Get the E-Book 📚

Laws

• In Business

  • Network effects is the law of universal gravitation [#003]

    • Management is anti-network-effects

  • Matthew Effect is the Theory of Evolution [#004]

  • Speed is the theory of relativity [#016]

  • Path dependence is Newton’s First Law (i.e. the law of inertia) [#032]

• In Management

  • The Peter principle is the law of gravity [#027]

Lessons [#001]

• Search the Hidden Truths, Then Stick to Them (a.k.a Non-consensus and Right)

• Become a π-shaped talent

• There are only two real core competencies:

  • The ability to discover opportunities

  • The ability to learn, continuously

• Luck accounts for 90% of success.

Product Management (#002)

• In most product categories, it’s not the first product that succeeds. Product managers play an important role.

• Product managers need to be both specialists and generalists.

• To be a great product manager, what matters most is your value system. And it should align with what the industry requires.

• Product management requires cross-disciplinary thinking to solve new problems that pop up all the time.

• The best combo is Professional PM's professional skills + Founder PM's mentality.

• Heart vs. Data [#023]

Mental Models

• Network Effects [#003]

  • Shape (Exponential, Linear, Asymptotic)

  • Scope (Honeycomb, City, National, Global)

  • Coefficient (the level of activity in the network)

• Matthew Effect [#004]

  • Most people don’t make choices on perfect information, they just go with the first thing that “seems right”.

• Market Concentration (1,2,3,7, ∞) [#005]

  • Big markets aren't enough; you need markets that allow big companies to exist.

  • Measured by concentration ratio (lagging indicator)

  • To determine the eventual market concentration of the industry, you need judgment. (leading indicator)

• Network Effects + Matthew Effect + Market Concentration [#005]

  • Pay attention to the interplay of different effects that benefit or suffer from scale, understand the trade-offs, and find the optimal balance through experimentation.

    => Rule of Thumb: Be top of mind (i.e. top three), while optimizing for capital efficiency.

Market [#007]

• Incremental Market vs. Reserve Market

  • Measured by penetration rate

  • The difference in customer acquisition costs between Incremental and Reserve markets is at least 10X

  • The key to win in a reserve market is through differentiation.

• High Frequency vs. Low Frequency

  • High-frequency Apps have unfair advantages (e.g. user acquisition costs, retention rates, consumer perception) over low-frequency Apps

  • The key to win in if your product is low-frequency is to go deep into the industry value chain.

Industry [#008]

• Chain Owner has the most initiative during an industrial change while other players can only be passive.

• Smiling Curve

  • The best place to be in an industry chain is either upstream - close to core inputs or downstream - close to the end consumers.

• The Ecosystem Model is prevalent for Internet businesses.

  • If establishing an industrial ecosystem can improve industry efficiency, then you must either become an ecosystem builder or join as a productive member, otherwise, you will be eliminated from the ecosystem.

• The Game between Upstream and Downstream

  • Market concentration affects the relationship between upstream and downstream.

  • Your upstream or downstream can interfere with the market concentration in your segment.

  • For a company, the best situation is that the concentration of both your upstream and downstream are very scattered, and only your concentration is high. This kind of luck is hard to come by.

  • If a) your upstream or downstream concentration is higher than yours, or b) because of your high concentration, your upstream and downstream have to game with you to change your concentration, the game will persist for a long time.

  • The most fierce games occur when the numbers of upstream and downstream companies are almost the same, and the numbers are small.

Entrepreneurship

• Timing [#009]

  • Timing > Environment > People

  • The window of opportunity for a category usually lasts for three months.

  • Due to FOMO, even for the wrong opportunity, you may still get funded.

  • How does a window of opportunity open?

    • Changes in PEST (Political, Economic, Social, Technological) factors

    • Change in Industry Practitioners’ Cognizance

      • All the great needs most definitely have been tried many times in the wrong way or at the wrong time.

      • Many teams working at the same thing will accumulate to the overall industrial cognizance of what works and what doesn’t.

  • Two rules

    • “If you believe something will eventually happen, try every 3 years." - Marc Andreessen

    • “As long as you can stay alive, it's always better to enter early than to enter late.” - Wang Huiwen

• Strategy [#010]

  • Strategy = Insight + Action Plans » Highest ROI (in different times & spaces)

  • Standard Strategy vs. Effective Strategy

    • Standard strategies (from consultants) are often ineffective when they are applied to a specific company.

    • An effective strategy for a particular company is like a doctor treating a malaise, there is a diagnosis, hypothesis, and testing process. A strategy working for a company is the result of multiple factors (e.g. product, operation, marketing) combined together.

  • Meituan used a Platform Strategy

• Total Addressable Market [#011]

  • When starting up, TAM is the most important thing to look at. It’s hard to guess right, especially when it’s early. Overestimation and underestimation can both be fatal.

  • The appropriateness of the investment depends on the judgment on future market size.

  • Three Ways to Judge TAM

    • Modeling based on data

    • Deduce based on human basic needs

    • Estimate based on analogous markets

• First Mover vs. Last Mover Advantage [#012]

  • Last Movers benefits from

    • The market needs less convincing

    • Knowing it can be done

    • Being able to zoom out and consider things from a market perspective and avoid blind spots that innovators are prone to.

Strategy For Product

• Product-Market Fit (PMF) + Diffusion of Innovations [#014]

  • Find a sharp enough entry point: rough-and-ready product targetted at Innovators & Early Adopters

    • Why now - why, at this point in time, this need has surfaced

    • What changed that changed the feasibility of a product

    • Ease of use increases as we move beyond early adopters

  • How to find the Early Adopters

    • The best situation is that you are the Innovator yourself or you often hang out with them.

    • Short of that, you can do it through the correct segmentation.

      • Age is often an important demarcation for early adopters
        

• Segmenting, Targeting, Positioning (STP) [#015][#016]

  • Segmenting - divide the market into many segments

    • Coordinate system, dimensions, altitude

    • Segmenting is not to choose the final market, but to choose an entry point. 

  • Targeting - select one of the segments as the target market

    • Now (entry point) & Long-term (eventual target market, network effects)

    • Competitive Environment

      • too broad a target market leads to unbundling

    • When choosing T, you must have product design, marketing efficiency, organizational ability, and capital efficiency in mind.

  • Positioning - form the cognitive connection between the market (demand) and the product (supply)

    • T is more focused on demand and customers while P is more focused on supply and products

    • Effective marketing is one that can quickly associate T and P.

• Price, Product, Place, Promotion (4Ps) [#017]

  • Price > Product > Place > Promotion

  • STP should connect with 4Ps

    • e.g. your product decisions strengthen your positioning in your target segment.

  • 4Ps are a combination

    • e.g. the price that people pay should correspond to a certain expected product experience

• Internet 4Ps [#018]

  • Bits vs. Atoms

    • bits travel a lot faster

    • bits have a close to zero cost of replication

    • bits are programmable

  • How It Affects the 4Ps

    • In traditional industries, all the 4Ps have their job to do and they're well separated.

    • For software, since the marginal costs are extremely low, the price can be compressed even to 0. When the price is 0, it’s Promotion by itself.

    • Because bits are programmable, a widely distributed free software Product can become Place or Promotion and earn from those activities.

  • Many powerful business models can be simplified as making one of the 4Ps 0 and instead charging from another P. The Internet greatly increases the possibilities to do so.

Strategy For Operation

• Tiering [#022]

  • No single product can satisfy all the needs in its category, and no single tiering (i.e. product mix) can satisfy all the needs. 

  • Two Types of Tiering

    • One Product, One Tier

      • The danger is a disruptor eating into your market segment

    • Multiple Products In One Tier

      • The danger is chaotic product mix and cannibalization

  • Tiering & Consumer Perception

    • When a brand doesn't form a sharp impression in consumers' minds, the consumer recognition of that brand would go down. That declined recognition would slowly make room in the STP segment and let a more sharply defined company occupy that space.

  • More Subtle Tiering

    • Coupons (price differentiation)

    • Membership

  • Clients Tiering for Software Companies

    • For B2B software companies, the most profitable segment is the “waist“ (mid-sized) customers

• Categorizing (The A/B Sides of the Internet) [#023]

  • A - Supply And Fulfill Online

    • A1 - Visible to direct correspondence

      • WeChat, WhatsApp, Line

    • A2 - Visible to friends, friend’s friends

      • Facebook

    • A3 - Visible to everyone, followers see first

      • Twitter, Weibo, Tiktok, Kuaishou, Instagram

    • A4 - Visible to everyone

      • Toutiao, Google, Baidu, Tencent Videos

  • B - Supply And Fulfill Offline

    • B1 - SKU-based Supplies

      • E-Commerce, Marketplaces, Hardware

    • B2 - Location-based Services

      • C stays, B moves

        • Delivery, On-Demand Services

      • C moves, B stays

        • Group-buy, In-store Services

      • C moves, B moves

        • Ride-hailing, Bike Sharing

      • Cross-Cities

        • Hotel, Travel, Plane/Train Tickets

• Categorizing & SuperApps

  • Categorizing determines which products can be made into one app and which can't.

Needs

• Needs > Requirement > Demand

  • Needs is the basic motivation of the users.

  • Requirement is the product design considerations based on those needs.

  • Demand is the quantity demanded by the market once the product is made, and it's usually used in tandem with the quantity supplied.

• Characteristics of Needs [#025]

  • The Unsatisfiability of Needs

    • Technologically Feasible > ROI Positive > Returns Over the Cost Of Capital > Can Support A Competitive Firm

  • The Time-Variability of Needs

    • Needs of individuals are perhaps always changing; it’s scenario-dependent.

    • For the whole human species, needs are everlasting. But the quantity of a need changes according to the overall societal conditions.

    • On a human desire level, needs are already written - the need either exists or it doesn't. But the intensity, prevalence, and feasibility are affected by many factors.

    • The window of opportunity for the great needs is like lightning in the sky. It is the moment when the positively and negatively charged clusters penetrate the air. At least one of the charged clusters needs to be in motion - either the need itself or the feasibility to satisfy it.

  •  Two types of needs.

    • Bottlenecked - the needs have always been strong and prevalent but it is difficult to realize them.

    • Catalyzed - although the needs have always existed, their intensity and prevalence have become greater in recent times.

• How to Identify Needs [#026]

  • Study the past successful failures - those that gained traction or almost made it but ultimately failed

    • All the great needs have been attempted by people in the past

  • The prevailing method of realizing needs at any point in time bottlenecks the needs being satisfied to the easiest-to-realize needs of an appropriate group of the population. In the next iteration, the realizable needs achieve a breakthrough, and the serviceable population expands.

• The Microscopic Needs [#027]

  • Rarely can users tell you what their needs are. Even though they may say they want certain things, they may be thinking in terms of “supply” - they already have the solutions in mind. It’s no different from the ideas that a product team or higher management think will work.

  • On user research:

    • Beware of confounding variables. e.g. Hawthrone Effect

  • Most of the time, you can only guess what people’s needs are. Therefore, the accuracy of your guesses is very important.

    • One way of guessing is to offer options for the product features and based on people’s selection of the options to deduce the underlying need.

  • How to listen to customers

    • In the early days of building a product, it’s probably futile to ask people who don’t use the product why they don’t use it. The more effective method is to ask the people who use the product.
      

Demand & Supply [#028]

• As products scale, or for B2B products, getting the demand & supply relationships right is critical. Sometimes, teams even have a strong incentive to get it wrong.

• The supply and demand of things are affected by price (i.e. price elasticity)

• Factors affect the demand-supply condition in an industry

  • Space [#029]

  • Time [#029]

    • Whether to develop a certain organizational competency within the firm or not depends on the characteristics of your industry.

      • e.g. If your outsourced your business analytics and R&D capabilities, when industry changes actually come, you don’t have good people who can respond.

    • Macro or micro level

      • The demand-supply condition changes slowly in the retail industry on the macro level (i.e. decades), but fast on the micro level (i.e. seasonal).

  • Segmenting [#030]

    • Usually, the high-end segment has an abundance of supply while the low-end segment has a shortage.

    • Regardless of whether it’s an oversupply or undersupply, in the high-end market, both supply and demand have a greater elasticity (more responsive to price change). In contrast, on the low-end, both demand and supply are more inelastic (i.e. things tend to be necessities).

    • The Hierarchy of Human Vanity

      • Necessity → Differentiation → Showing-Off → Class, Prestige → Non-monetary resources → Status, Tradition, Culture

  • Non-market Factors [#031]

    • They may cause some markets to be artificially imbalanced in terms of demand and supply, but these imbalances may have their reasons to exist. e.g. we don’t have the resources to satisfy a need unlimitedly. We’ll need to adjust accordingly to the context.

  • Online vs. Offline [#032]

    • The advent of the Internet industry and its development has broken the demand and supply equilibrium of the past. As it’s being broken, a new equilibrium is built with new business forms.

On Business Classics

• Industry Chain [#008]

  • Smiling Curve

  • Porter’s Five Forces

  • The Game between Upstream and Downstream

• Porter’s Three Generic Strategies [#010]

  • Cost Leadership

    • In the Internet age, products without the same scale just can’t deliver the same user experience. Cost leadership in today's context should be Experience Leadership powered by Network Effects.

  • Differentiation

    • With Internet products’ ability to offer personalized experience (i.e. each person gets his own product), the room for differentiation is greatly compressed.

  • Focus

    • Internet companies have a greater ability to operate multiple differentiated products as compared to traditional companies. It almost doesn’t matter if a company is focused on not. (i.e. there are too many strong scale/network effects)

    • What’s most important is the company’s core competence (i.e. A/B Sides of the Internet)

• 4Ps [#017]

  • 4Ps in the Internet Age [#018]

• Disruption Theory [#030]

  • Christensen probably didn’t consider segmenting. High-end disruption is real.

Books Mentioned

• Marketing (STP, 4Ps)

  • Marketing Management

  • Positioning

  • The 22 Immutable Laws of Marketing

  • The Innovator’s Dilemma (Segmenting)

  • The Pyramid Principle (MECE mental framework)

• Product Strategy

  • Diffusion of Innovations

• Needs

  • The Salmon of Doubt (Three rules of technology)

  • The Selfish Gene (altruism)

Find the article insightful? Please subscribe to receive more learnings from Chinese Internet companies and I’d appreciate it if you can leave a comment or help spread the word!

Subscribe now

I Want the E-Book 📚

The History

In 2003, Dianping was founded in Shanghai to solve the problem of finding good restaurants by crowdsourcing reviews from fellow consumers.

In 2010, Meituan was founded in Beijing to work on group buying.

Dianping was benchmarked against Yelp in the United States, although it was founded earlier than Yelp. And Meituan copied from Groupon. However, both Yelp and Groupon are considered failures by conventional standards, with a market value of about 1-2 billion US dollars. (Note: at the time he gave the lectures in Sept 2020).

The Groupon idea led to what was called the “Thousand Groupons War” in China in 2011.

At the height of the movement, there were four to five thousand group buying websites in the country. Dianping also entered the group buying market, and so became a competitor to Meituan.

The Thousand Groupons War happened very fast.

Meituan was launched on March 4th, 2010 - the first of its kind in China. By the end of that year, there were already close to a thousand group buying websites.

In October 2011, Meituan became the number one group buying website in China.

In 2012, most of the group buying websites went bankrupt. By then, the competitive landscape was clear - Meituan, Dianping, (Baidu) Nuomi, Lashou, and 55tuan were the sole survivors of this gladiatorial fest.

In mid-2012, Meituan launched a new product called Maoyan, to serve the movie ticketing market.

In 2013, we launched hotel booking services.

In late 2013 and early 2014, we started food delivery.

In October 2015, Meituan and Dianping merged. At this point, Meituan had 4 major business lines - group buying, movie ticketing, food delivery, and hotel reservations (OTA).

Our main competitors in group buying were Dianping, which received backing from Tencent, and Nuomi, which was acquired by Baidu.

In movie ticketing, the competitors were Alibaba’s Tao Piao Piao, Tencent’s Weiying, and Baidu’s Nuomi.

In food delivery, they were Tencent and Dianping-backed Ele.me, Baidu Waimai, and Alibaba-incubated Koubei.

In OTA, we were competing against Ctrip, Baidu-backed Qunar, and Alibaba’s Fliggy.

In every business line, we were burning money. The industry at that time likened us to Germany during World War II, fighting wars on multiple fronts. But Germany only fought on two fronts, and we fought on four.

In 2016, we launched merchant-facing restaurant management products.

In 2017, we launched Kuailv, dedicated to solving supply chain issues for merchants, including supplying fresh goods and financial services. In the same year, we launched Hazelnut B&B (i.e. homestay) and ride-hailing services.

In 2018, we acquired Mobike and filed for IPO.

In 2019, we launched Meituan Maicai (i.e. grocery e-commerce).

In 2020, we launched electric scooters/bikes, power bank sharing, and Meituan Select (i.e. community group buying).

Basically, every year, we have new businesses being launched, and none of them doesn’t entail a fierce fight.

In the foreseeable future, we will have many more new business lines, such as drone delivery and autonomous vehicle delivery which are being tested right now.

The Technology DNA

Meituan is a company focused on food. Our strategy focus can be summarised as “Food+Platform(s)”. It also covers travel, shopping, entertainment, accommodation, and transportation, as well as various services on the merchant side. Meituan's mission is "We help people eat better, live better".

(Note: The strategic focus has been adjusted to “Retail + Technology” since September 2021.)

There are many reasons why Meituan can do so many businesses well, one of which is our technology DNA.

In the relatively short [modern] commercial development in China, we had the good fortune to benefit from many dividends (i.e. population, urbanization, globalization, etc). For a lot of businesses, if they could seize just one of them, they would do well.

In contrast, if we look at the United States, where the commercial development history is longer and such dividends have been tapped out, we’ll see that the long-term driver for business growth is technology. Technology also includes scientific management practices and operational know-how. (Note: in other words, productivity).

For example, Walmart is actually a technology company. It launched its own satellite network in the late 1980s, which demonstrates the company’s commitment to investing in technology.

If a company wants to be competitive over the long term, at the core it must be a technology company, no matter what the company looks like from the outside.

One of the reasons why Meituan can win in the Thousand Groupons War is that Meituan has better technological capabilities than its peers.

For merchants to agree to be on a group buying platform, they usually would require the ability to settle their receivables from the platform any time they choose. If the group buying platform skips town or goes bankrupt after they’ve taken the consumer’s money, it’d be disastrous to the merchants’ cash flows and even cause the merchants to go bankrupt.

Meituan provided a system that does the accounting properly and pays the merchants any time they choose. As such, it gave merchants a sense of security when partnering with Meituan. In turn, it gives Meituan an advantage on the supply side in acquiring merchants. This played no small part in Meituan eventually winning the Thousand Groupons War.

Food delivery is a business that has prominent peaks and troughs in terms of demand. As a result, the servers need to be able to support both the peak and trough loads in a cost-controllable manner, which requires rather strong technical capabilities.

If we only aim to be able to handle the peak loads by stacking up servers, it’d result in overall low utilization rates. On the other hand, if we don’t do this, we won’t be able to handle the peak loads and suffer crushes. This makes load balancing a critical challenge. At Meituan, we let the servers handle user requests during peak loads and then run big data operations during off-peaks.

Technology Helps A Company’s Operational Granularity

An important metric to measure a company's operational expertise is its operational granularity.

For example, supermarket chains are run on a store-by-store basis. The boss is in charge of choosing the location, as choosing the right location is a core competence in retail. (Note: covered in Pt 22: Demand and Supply)

Once the location is selected, a general manager is appointed to run the store. If the location doesn’t become profitable after six months, then change the general manager. If it still doesn’t turn profitable after another six months, then it means you’ve got the location wrong.

The operational granularity in this case is pretty coarse. Still, some of the most famous supermarket brands in China run their operations like this. The market was forgiving during their development. So long as their locations are roughly right where they need to be, they won’t encounter major issues.

A level of granularity lower than this is by categories - knowing which categories of goods are the attention-grabbers, which ones are the money-makers, which ones are necessities, etc.

Further still, is to operate like 7-11, on the SKU level. At this level, it requires more systematic management, better data, etc.

The finest level of granularity is on the order level. In the era when goods were mass-produced, a money-making SKU is your holy grail. However, when goods were not mass produced (i.e. either handmade or personalized), not every order can be profitable.

The finer your operational granularity, the more logical your operation will run. (i.e. relying less on guesses or intuition). In the pre-Internet age, having a fine operation is almost impossible. Technology improved businesses’ ability to granularise their operations.

Operational granularity is also involved in setting KPIs for teams.

For example, if we want a business line to grow by 10%, then how to reasonably allocate this 10% into the business line’s constituent regions requires a clear understanding of the business operations in every region. And this is just a one-dimensional goal.

I’ll give you another example in food delivery. We need to know which cities should operate at a loss and which cities need to be profitable, and how much the profits or losses should be. Then we need to think about in a single city how we break down the overall KPI for the city to the city’s various business centers.

As such, Meituan's performance evaluation system is highly personalized. It needs to align with the business functions specific to that person. You can’t achieve such a sophisticated appraisal system without robust management systems in place.

Technology Helps the Digitalization of the Offline Economy

Before Meituan, I worked on a second-hand property resale website (TaoFang). It wasn’t a success.

At that time, when I visited real estate agencies for partnership meetings, I noticed that most of them didn’t have computers. In 2004, there wasn’t enough digital infrastructure to support an online real estate business. Hence, I sold the company.

By 2009, I started to notice that all the real estate agencies have been using computers. But it was too late, the market was already getting saturated. If I were to enter then, the only reasonable play I could think of would require the company to self-operate a lot of real estate agencies on the ground. I felt that was too “heavy”, so I didn’t pursue the idea.

Now, the foremost Internet-enabled real estate company in China is Lianjia (HomeLink). They started offline as a real estate brokerage chain, so they didn’t know much about the Internet in their early days. I saw their website domain was homelink.com.cn, so I bought the domain lianjia.com. (Well, I didn’t make much from them. Any money I made at that time was thrown into Meituan.)

Technology Helps With Targeted Subsidies

There are a few prerequisites for doing subsidies well. The first is segmentation - STP (Pt 13 - Segmenting, Pt 14 - Targeting & Positioning); the second is data analytics.

Without good data analytics, it is impossible to do your accounting correctly. Data analytics provides the foundation to have detailed customer profiles on which you test various subsidy schemes. Every subsidy scheme may require complex calculations. However, if you do your subsidies well, it’s quite possible to multiply the top line of the business.

Technology Helps Combat Frauds and the Black Market

One reason why Uber failed to compete in China was that during the subsidy war, their subsidies didn’t reach authentic users. It was siphoned away by the black market.

The average price for a ride was ￥16. And Uber may give the passenger a ￥15 subsidy and ￥10 to the driver. The subsidies were even higher than the cost of a ride. Because of this, some drivers would cooperate with their friends or interested parties to book fake rides to earn subsidies from the platform.

When there is an economic incentive, the black market that’s formed can be quite “industrialized”. They would also have good engineers working for them looking to exploit the system. Therefore, it’s necessary to have robust systems to fight the black market.

Fraudulent actors usually can be identified by certain traits. For example, a normal person wouldn't be at ten discrete locations in the city in a single day. To separate such usage patterns from normal users require strong data capabilities.

Artificial Intelligence

Technology also affects your ability to understand the needs of merchants and consumers, provide personalized recommendations, improve the efficiency of the supply chain, and more by leveraging artificial intelligence.

I’ll give you some examples of the application of AI at Meituan.

One dimension of the segmentation strategy for food delivery is the delivery radius. And the delivery radius is affected by the weather. During bad weather, we need to reduce the delivery radius to make lives easier for the delivery riders. How much the radius should be reduced is handled by AI. Under different circumstances, the delivery radius varies. So the delivery map is highly irregular, as it’s dynamically adjusted according to weather conditions.

Customer service also requires the use of AI.

It’s technically challenging to have customer service chatbots that can handle sequential conversations. For Meituan, most of our business lines have low average order values and low gross margins, so we can’t afford large-scale customer service teams.

Additionally, our business is local services, which are highly unstandardized, and we’re a platform, which won’t have enough control over everyone in the value chain. This means our user may require customer support often. Therefore, we have to use customer service bots to provide help at scale cheaply.

Of course, AI will be used extensively in emerging fields such as drone delivery, unmanned vehicle delivery, and consumer intent detection.

In conclusion, Meituan is a high-tech company.

[End of Course]

Find the article insightful? Please subscribe to receive more learnings from Chinese Internet companies and I’d appreciate it if you can leave a comment or help spread the word!

Subscribe now

P.S. Long piece, best viewed in browser.

Subscribe now

When you don't create things, you become defined by your tastes rather than ability. Your tastes only narrow & exclude people. So create.

- why the lucky stiff

A Designer’s Taste

1. Taste and Ability

The thing that a designer needs the most is good taste. When we say someone has good taste, we usually mean he has a keen eye for aesthetics or he is good at seeking out beauty. Surprisingly, when someone does have good taste, it's often not because he's particularly good at discovering beauty; it's because he's particularly good at recognizing ugliness. If he can see the ugliness, and he can't stand it, he’d naturally tend toward beautiful things.

Hence, having good taste is not just about beauty, it may have even more to do with ugliness. By “beauty” and “ugliness”, we don’t just mean how things look to our eyes, they also pertain to the beauty and ugliness of products, technology, systems, human nature, etc.

In the first couple of years when you start your career, especially for those of us who work in design, there may be a gap between your taste and your competence. Your work may not be as good as your taste would dictate. Don’t despair, for your skills would eventually catch up. Ultimately, it’s your taste that would determine your style.

Ira Glass, the creator of the hit radio program This American Life, shared with us his experience in an interview:

All of us who do creative work, we get into it because we have good taste. But it’s like there is this gap. For the first couple years that you’re making stuff, what you’re making isn’t so good. It’s trying to be good, it has the ambition to be good, but it’s not quite that good.

But your taste, the thing that got you into the game, is still killer. And your taste is good enough that you can tell that what you’re making is kind of a disappointment to you. A lot of people never get past that phase. They quit.

Everybody I know who does interesting, creative work went through a phase of years where they had really good taste and they could tell that what they were making wasn’t as good as they wanted it to be. They knew it fell short.

Everybody goes through that. If you’re going through that phase right now, you gotta know it's totally normal and the most important thing you can do is do a lot of work. Do a huge volume of work. Put yourself on a deadline so that every week or every month you know you’re going to finish one story. It is only by going through a volume of work that you’re going to catch up and close that gap. And the work you’re making will be as good as your ambitions.

I took longer to figure out how to do this than anyone I’ve ever met. It takes a while. It’s gonna take you a while. It’s normal to take a while. You just have to fight your way through that.

- Ira Glass

2. What Determines Your Taste

One’s taste is shaped by various factors collectively. The most direct, perhaps, is the culture in which one grows up. What is Chinese culture today? Some field photographs from the (former) creative director of frog design, Jan Chipchase, may give you an idea.

One can’t help but notice the gap between wealth and poverty, the contrast between Western culture and traditional culture, and the distance between idealism and reality.

For people who grow up in such a diverse and divergent culture, it stands to reason that they’d be more tolerant of differences and their tastes more flexible. You can say that Chinese people are extremely adaptable. They take whatever comes their way, the good and the bad. They could drink poisonous milk and eat gutter oil.

You can say that Chinese people are influenced by the philosophy of the Doctrine of the Mean (中庸) — doing things the way most people do and don’t deviate from the natural order. The bird that pokes its head out gets shot.

You can also say that the Chinese are dialectical. (Note: Dialectical materialism from Marxism). If asked to compare two things, the answer you usually get is, “This one is fine, but that one is also not bad. Each has its own virtues.”

Because of the contradictory environment that they grow up in, Chinese people have to have a great tolerance for ugly things. Otherwise, you won’t be able to make it. In China, people who express discontent toward ugliness are often shunned, chastised, and may even risk their lives. As such, most of us involuntarily accept our surrounding world. Critical voices are few and far between.

Interestingly, the public doesn’t think of those things as ugly. Only when someone points it out, do they finally realize, yeah, it indeed is ugly.

From melamine milk, and gutter oil to the poisonous drug capsules now, they were essentially the results of people doing nothing in the face of ugliness. China needs more people who can't tolerate ugliness, and a re-education to recognize what’s “ugly”. We will then discover beauty by exposing ugliness.

3. Tolerance of Ugliness

iD Commune (Note: an industrial design-focused blog in China) published an article called “The Perception of Beauty”, and the section about “taste” strongly resonated with me:

Taste in our colloquial context is usually centered around style, like how we would describe someone who is particular about the way s/he dresses as “having a good taste”. If we were to evaluate Steve Jobs’ taste based on his wardrobe, it’d be a very narrow and misleading view of what taste really is. Taste is an ability, not a style.

(Note: Actually, Steve Jobs was particular about how he dressed. R.I.P. Issey Miyake.)

We need not be punctilious about the language. Conceptually, having good taste is definitely not as simple as dressing meticulously or following fashion trends. Taste is an ability to perceive beauty. It’s a power. And every one of us possesses this power. Otherwise, we wouldn’t have masterpieces that captivate entire masses.

If everyone has this ability to perceive beauty, then what sets Steve Jobs’ taste apart? It is because some of us have more of this ability and others have less? Personally, I think it’s the exact opposite. Steve Jobs’ taste stands out from the rest because he’s intolerant of ugliness, almost to the point of extreme. This critical standard had affected him his entire life.

When Steve Jobs moved to a new house. His new home almost had no furniture for a decade, because he couldn’t find anything up to his satisfaction. He’s someone that would prefer nothing than something less than perfect.

Therefore, the intolerance of ugliness is the best explanation for some people whose tastes stand out from others. It’s also an explanation for why China can’t produce people like Steve Jobs. Because we have been inundated with ugly stuff our whole life, we’ve become very tolerant of ugliness. Don’t believe me? Let’s take a look at these incredible buildings.

4. Is Beauty Subjective?

Growing up, we were taught that beauty is subjective. It’s in the eye of the beholder; it’s a matter of personal preference. There’s no wrong way to paint a picture. You can’t make fun of your brother for coloring people green. “You like to do it your way and he likes to do it his way”. Your mother at this point is not trying to teach you about aesthetics. She’s trying to get the two of you to stop bickering.

Saying that taste is just personal preference is a good way to prevent disputes. The trouble is, it's not true. You feel this when you start to design things.

If you’re a designer, and you buy into that beauty is subjective, then you can’t go about doing your job. If beauty is just personal preference, then why do we need to design? Everyone's is already perfect: you like whatever you like, and that's it.

Experienced designers would realize that as their gain experience, their understanding of beauty is changing. There are some beauties that transcend individuals. Seeking out the commonalities between these recognized beauties becomes the primary approach to how they work.

In high school, Karl Marx taught us to look at problems dialectically. Everything is relative, including truths. This notion sticks with us even as adults. Once you start to think about design comparatively, you will find that there are striking similarities in the ideas of beauty across different fields. The same principles of good design crop up again and again.

Principles of Good Design By Dieter Rams

If there is such as thing as universal beauty, naturally there would be ways to create beautiful things that appeal to the public. German industrial designer Dieter Rams spent decades answering his own question - is my design good design? - and put forth his ten principles for good design.

Dieter Rams talked about design in the context of industrial design, which is the art of shaping manipulable products. In his context, design objects include all kinds of physical things that people want, from cutlery to chairs, telephones to cars. I will not discuss the differences between industrial design and other design fields here, nor will I go through Dieter Rams' ten principles one by one. I’ll just pick a few that I find interesting to share.

5. As Little Design As Possible

Another way of saying this is that “good design is simple”. You may have heard different versions of this many times in other fields as well. In mathematics, the simpler the equation, the shorter the proof, the better. For architects, it means focusing on carefully chosen structural elements, not excessive ornaments. Similarly, in writing, it means “omit needless words”, “brevity is the soul of wit” and “write a shorter letter”.

Ludwig Mies van der Rohe, who succeeded Walter Gropius as the director of The Bauhaus, a ground-breaking school of modernist art, design, and architecture, popularized the design philosophy of "less is more" in architecture. When modernist architects try to create buildings that abandon the ornamented traditional styles, they unconsciously designed buildings that still had ornamentation. Only when space and articulation replaced symbolism and ornament as keywords in architectural conversions, did good buildings begin to emerge.

John Ruskin believed that the soul of architecture was contained in the veneer of decoration that concealed the exterior walls. I think it’s fair to say today that you can have ornamented buildings, but don’t build buildings that are ornaments.

In fact, most things are inherently simple. We incessantly stress “less is more” because when doing design work, most people forget. For example, in web design, some people fear white spaces and always want to fill the page with bells and whistles. It’s actually quite stupid.

The design embellishments are just a diversion to cover up the lack of substance. When you force yourself to make things as simple as possible, you are forced to face real problems. When you can't cover up with ornaments, you have to do essential parts really well.

6. Good Design Is Timeless

In math, every proof is timeless unless it contains a mistake. This is why mathematician G.H. Hardy claimed, “there is no permanent place in the world for ugly mathematics.”

Aiming at timelessness (or at least long-lasting) is the best way to make yourself find the best solution to the problem. If you don't want your product to be replaced, then you have to make the best possible one. This is why the works of the great masters are compelling in any era and leave little room for those who came after.

In a similar vein, aiming for timelessness is also a way to avoid the undue influence of trends. Trends change - every year, every season, or even every day. If you want to make great products, you should avoid being controlled by what’s hot right now. Take Samsung and Apple as an example. Samsung is clearly a trendy company, it follows its customers’ needs closely. Whatever consumers want, they make. Apple is more concerned with making great products. They do not necessarily listen to everything consumers have to say. They don’t care about what’s popular. They care about finding the right answers.

If you look at the entrepreneurial boom in China, many startups are pursuing the current hottest trends (Note: 风口, fengkou, “Wind Gap”), often copying the latest hits in the United States.

New Internet concepts spring up almost every day. LBS (location-based services), SoLoMo (social-local-mobile), Cloud Computing, Pinterest, etc. The media hype has not only failed to inform, but has misled many entrepreneurs to think that if they had found a hit idea, they’d definitely be able to make it big, and users would flock to their app.

Reality suggests otherwise. Whether it was the previous center-of-attention group buying, or the explosively popular social game "Draw Something", they were all fads that slowly faded away from our lives over time. On the other hand, if something is long-lasting, its appeal must come from its own quality, not its popularity.

If good design is timeless, then how do we judge whether or not our work is affected by time?

One way is to make your work appealing to multiple generations of people. It’s difficult to predict the future, but one thing is for certain - the people of the future will not care about our fashion today. Teenagers today don’t care about their parents’ or grandparents’ fashion. It’s also true in the reserve. Your grandma doesn’t care about what kids are doing nowadays. If your work can appeal to people today as well as people 200 years ago, chances are it will still appeal to people 200 years from now.

7. Good Design Makes A Product Understandable

Good design makes products easy to understand. Many a designer worked very hard only to discover their efforts had been in vain because they focused on the wrong problem.

Use typography as an example. With the influence of the Bauhaus Movement in the mid-20th century, after the birth of Modernism, there was a boom in the usage of sans-serif fonts. This movement continued to even now. Designers everywhere in the world just slap the Helvetica font on everything they touch, because the sans-serif fonts look clean.

(Note: Sans-serifs fonts reign supreme today, as shown in below Twitter thread.)

However, whether having the serif or not (Note: in other words, aesthetics) is not the primary concern of typography. The primary concern should be legibility (i.e. effective communication). They should make content readable. Times New Roman is a serif font that’s easy to read. And there are many great serif fonts like it, such as Garamond, Baskerville, Carlson, Egyptienne, etc.

Great products are “thorough down to the last detail”. Even the most unnoticed details such as choosing the right typography require lots of research and comparison. To find the right fonts, the designers of the news reading app, Prismatic, studied the history of the various types of fonts and their developments, conducted a thorough analysis of the pros and cons of the specific candidates, and tested the fonts against their data:

Garamond is a French old-style type design from 1530 that was forward-thinking in geometric proportion and balance in stoke. Still, it had some imbalances, like the lowercase “a” and “e”.

Egyptienne, which is based on the Clarendon model, pairs really nicely with modern sans-serifs.

The varying thickness of the strokes and their balance in serif fonts work similarly to Chinese calligraphy. In good calligraphy work, the strokes are definitely not uniform. They are dynamic, in an overall balance. The human visual perception naturally creates an aesthetic response toward the interweaving of lines of varying thicknesses. Below, is a work by Piet Mondrian in the Neoplasticism Movement. The intersecting lines have varying thicknesses.

If we were to change the lines to the same thickness, the piece loses a lot of character immediately.

Sans-serifs are not all bad. They look great as article headings. What’s sad is that most designers probably don’t know any other sans-serif fonts other than Helvetica. There are many good sans-serif fonts, such as Akzidenz-Grotesk and Univers.

Akzidenz-Grotesk, created in 1898, is influenced by many fonts that come before it - Baskerville, Garamond, Carlson, Didot, etc. Akzidenz takes their geometric proportion and balanced strokes to a whole new level, and drops the serifs. Contemporary versions of Akzidenz-Grotesk descend from a late-1950s project to enlarge the typeface family. These efforts to achieve a more machined yet human feel led to Helvetica from Max Miedinger, and the works of Adrian Frutiger, like Univers.

Typeface design started with the invention of printing. Europeans claimed that printing was invented by a German named Johannes Gutenberg in 1450. Little did they know at that time that Bi Sheng in China had already invented movable type printing centuries earlier in 1040, as recorded in Dream Pool Essays by Shen Kuo.

The writing of Chinese characters has always been an art form known as calligraphy. Its influence spread throughout East Asia. However, Chinese today who write Simplified-Chinese barely retain the ability to appreciate calligraphy. We failed to inherit the historical developments in typography from our ancestors. Today, the Chinese fonts on computers are awfully scarce as compared to other languages. It’s a damn shame to lose our traditional culture like that.

8. Good Design Is Honest

An honest product does not make false claims or exaggerations. To a lot of folks, honesty and design seem to be at odds with each other conceptually. But the importance of honesty in design cannot be overstated.

Take for example the data charts that we see everywhere being used for justification and something we automatically ascribe credibility to.

If you believe that a piece of data or a chart is the most powerful, indisputable evidence - after all, “data doesn’t lie” - then you’re mistaken. Charts do lie. The father of information design and data visualization, Edward Tufte, said in his book, The Visual Display of Quantitative Information:

When a chart on television lies, it lies tens of millions of times over; when a New York Times chart lies, it lies 900,000 times over to a great many important and influential readers.

An example in his book is a chart that shows the change in traffic deaths due to stricter enforcement by the police against cars exceeding the speed limit:

Nearly all the important questions are left unanswered by this display:

A few more data points add immensely to the account:

Imagine the very different interpretations other possible time-paths surrounding the 1955-1956 change would have:

Comparisons with adjacent states give a still better context, revealing it was not only Connecticut that enjoyed a decline in traffic fatalities in the year of the crackdown on speeding:

In short, the same chart can be used to convey a lot of different meanings. Finding the best way of communicating information with integrity is the work of the designer.

Paul Graham’s Principles for Good Design

The co-founder of Y Combinator, Paul Graham, outlined his principles of good design in his essay, Taste for Makers.

(Note: Since we’re translating back to English, I thought I’d just quote Graham’s words directly when appropriate since he expressed these points with better precision, efficiency, and style. PG’s words are in italics.)

9. Good Design Is Suggestive

Literature works can be described as either descriptive or suggestive, as do paintings. Suggestive works usually engage people more than descriptive works.

Everyone makes up their own story about the Mona Lisa.

In architecture and design, this principle means that a building or object should let you use it how you want: a good building, for example, will serve as a backdrop for whatever life people want to lead in it, instead of making them live as if they were executing a program written by the architect.

In software, it means you should give users a few basic elements that they can combine as they wish, like Lego. In math it means a proof that becomes the basis for a lot of new work is preferable to a proof that was difficult, but doesn't lead to future discoveries; in the sciences generally, citation is considered a rough indicator of merit.

10. Good Design Is Hard

If you look at the people who've done great work, one thing they all seem to have in common is that they worked very hard. Hard problems call for great efforts. When you have to climb a mountain you toss everything unnecessary out of your pack.

A story about Wang Shi (Note: A Chinese real estate tycoon and an avid adventurer who completed the Explorers Grand Slam. He submitted the Everest at age 53 and 60.) stuck with me.

A friend of his described Wang Shi’s approach, “When Wang Shi was climbing Mount Everest, whenever he reaches a new camp after every phase of climb, he goes straight into the tent, lies down and doesn’t move and doesn’t speak. His teammates thought he was dying. Wang Shi just wanted to conserve energy and spirit. The only thing on his mind was to reach the summit. Nothing else.”

The same friend described another climber, “After every phase of climb, he’d talk to the media, announcing his progress and impart his new-found wisdom. At about 8,000 meters of altitude. He’s too exhausted to continue and had to scale down regretfully.”

The moral of the story is that when you set a goal, you need to give it everything you’ve got; you have to be able to endure the loneliness to scale to new heights.

In design, when the problem is hard, it forces the designers to go with a simple, elegant design. Fashions and flourishes get knocked aside by the difficult business of solving the problem at all.

Not every kind of hard is good. There is good pain and bad pain. You want the kind of pain you get from going running, not the kind you get from stepping on a nail. A difficult problem could be good for a designer, but a fickle client or unreliable materials would not be.

When Bauhaus designers adopted Sullivan's "form follows function," what they meant was, form should follow function. And if function is hard enough, form is forced to follow it, because there is no effort to spare for error. Wild animals are beautiful because they have hard lives.

Time always softens the pain and makes things look like more fun than they really were. But who said everything has to be fun? Pain builds character. (Sometimes it builds products, too.)

- Jamie Zawinski, the netscape dorm

11. Good Design Looks Easy

When we look at the work of great designers, we often think to ourselves, “That’s it? I could’ve done that.” Mostly this is an illusion.

A good, easy-to-use product may seem simplistic, but it’s the result of continuous improvements. The easy, conversational tone of good writing comes only on the eighth rewrite.

In science and engineering, some of the greatest discoveries seem like the simplest ideas. Listen to how Adam Savage tells the stories of how Eratosthenes calculated the Earth's circumference around 200 BC and how Hippolyte Fizeau measured the speed of light in 1849. Supposedly, anyone could have followed these simple, creative methods. But not many of us can make great scientific discoveries.

Some Leonardo heads are just a few lines. You look at them and you think, all you have to do is get eight or ten lines in the right place and you've made this beautiful portrait. Well, yes, but you have to get them in exactly the right place. The slightest error will make the whole thing collapse.

Line drawings are in fact the most difficult visual medium, because they demand near perfection. In math terms, they are a closed-form solution; lesser artists literally solve the same problems by successive approximation. One of the reasons kids give up drawing at ten or so is that they decide to start drawing like grownups, and one of the first things they try is a line drawing of a face. Smack!

Traditional Chinese ink wash paintings are composed of nothing else but ink of different concentrations and brushstrokes of varying techniques. You can say traditional Chinese painters have the best ability in the world to capture the essence of a subject. Alas, in the modern age, we forget these abilities or no longer talk about them.

12. Good Design Is Often Strange

Some of the very best work has an uncanny quality: Euler's Formula, Bruegel's Hunters in the Snow, the SR-71, Lisp. They're not just beautiful, but strangely beautiful.

This strangeness is the style that we talk about. Every artist wants to develop a personal style. The work that you do try to develop a style often isn’t good. But if you just try to make good things, you'll inevitably do it in a distinctive way, just as each person walks in a distinctive way.

Michelangelo was not trying to paint like Michelangelo. He was just trying to paint well; he couldn't help painting like Michelangelo. Mies van der Rohe and Piet Mondrian didn’t set out to be Modernists. They just wanted to do their best work reflective of their times.

The only style worth having is the one you can't help. And this is especially true for strangeness. There is no shortcut to it. The only way to get there is to go through good and come out the other side.

13. Good Design Happens In Chunks

This is a strange one. Great works often come in chunks. The inhabitants of fifteenth century Florence included Brunelleschi, Ghiberti, Donatello, Masaccio, Filippo Lippi, Fra Angelico, Verrocchio, Botticelli, Leonardo, and Michelangelo. Milan at the time was as big as Florence. How many fifteenth century Milanese artists can you name?

Something was happening in Florence in the fifteenth century. And it can't have been heredity, because it isn't happening now. You have to assume that whatever inborn ability Leonardo and Michelangelo had, there were people born in Milan with just as much. What happened to the Milanese Leonardo?

There are roughly a thousand times as many people alive in the US right now as lived in Florence during the fifteenth century. (Even more people in China.) A thousand Leonardos and a thousand Michelangelos walk among us. If DNA ruled, we should be greeted daily by artistic marvels. We aren't, and the reason is that to make Leonardo you need more than his innate ability. You also need Florence in 1450.

Nothing is more powerful than a community of talented people working on related problems. Genes count for little by comparison: being a genetic Leonardo was not enough to compensate for having been born near Milan instead of Florence. Today we move around more, but great work still comes disproportionately from a few hotspots: the Bauhaus, the Manhattan Project, the New Yorker, Lockheed's Skunk Works, Xerox Parc. (Silicon Valley)

At any given time there are a few hot topics and a few groups doing great work on them, and it's nearly impossible to do good work yourself if you're too far removed from one of these centers. You can push or pull these trends to some extent, but you can't break away from them.

14. Good Design Is Often Daring

At every period of history, people have believed things that were just ridiculous, and believed them so strongly that you risked ostracism or even violence by saying otherwise.

This problem afflicts not just every era, but in some degree every field. Much Renaissance art was in its time considered shockingly secular. Einstein's theory of relativity offended many contemporary physicists, and was not fully accepted for decades.

In George Orwell’s Nineteen Eighty-Four, the government tightly controls information to keep the population in line, especially by banning books. Ring any bells? Censorship in China dates back to the Qin Dynasty, known as “the burning the books and burying-alive the [Confucian] scholars” in 213 BCE. Ovid was banished from Rome by the emperor Augustus for writing Ars Amatoria. Even in ancient Athens that celebrated intellect and free speech, reading certain books might still get you into trouble.

American activist Wendell Phillips said in a speech:

What gunpowder did for war, the printing press has done for the mind, and the statesman is no longer clad in the steel of special education, but every reading man is his judge.

Great ideas are often bold and stand against the ruling class. This helps to explain why most Chinese Nobel laureates either won their prizes overseas or in prison.

The ridiculousness of what we’ve discussed above may seem painfully obvious to us now. However, people then might not feel there’s anything wrong as they live through it.

In a post-screening audience Q&A for the movie In the Heat of the Sun (1994) at the Shanghai American School, a young girl asked the film’s director, Jiang Wen, “The teens in the movies were living through the Cultural Revolution, why were they so happy?”. Jiang Wen responded after a short pause, “I can see that you seem pretty happy right now at this stage of your life. When you grow older, you’ll realize what a terrible time we’re living in.”

In the same way, we seem to feel that everything around us is acceptable - C’est la vie. Only decades later, we realize that the previous era was awful.

What we need to worry about right now is not the Orwellian “Big Brother”, but our turning a blind eye to ugliness. If you want to discover great new things, you should pay particular attention to them.

It's easier to see ugliness than to imagine beauty. Most of the people who've made beautiful things seem to have done it by fixing something that they thought ugly. Great work usually seems to happen because someone sees something and thinks, I could do better than that.

Giotto saw traditional Byzantine madonnas painted according to a formula that had satisfied everyone for centuries, and to him they looked wooden and unnatural. Copernicus was so troubled by a hack that all his contemporaries could tolerate that he felt there must be a better solution. Steve Jobs thought all the “smartphones” in the market looked terrible, so he sought to create the phone he wanted.

Intolerance for ugliness is not in itself enough. You have to understand a field well before you develop a good nose for what needs fixing. You have to do your homework. But as you become expert in a field, you'll start to hear little voices saying, What a hack! There must be a better way. Don't ignore those voices. Cultivate them. The recipe for great work is: very exacting taste, plus the ability to gratify it.

Lessons

• Good taste is not about discovering beauty; it’s about detecting ugliness.

• Good design is not concerned with fashion; it focuses on making the best possible product.

• Ornament is often a cover-up for the lack of substance. Good design needs to solve the real problem.

• Good design is hard but looks easy.

• Good design is daring, and constructive.

References

• Original Article in Chinese

• “The Taste Gap” by Ira Glass

• Ten principles for good design by Dieter Rams

• Taste For Makers (2002) by Paul Graham

• Is There Such A Thing As Good Taste? (2021) by Paul Graham

• Content-centric design by Prismatic

Find the article insightful? Please subscribe to receive more learnings from Chinese Internet companies and I’d appreciate it if you can leave a comment or help spread the word!

Subscribe now

Subscribe now

Factors Affecting Demand-Supply: Online vs. Offline

We’ve covered many factors that affect the demand and supply conditions of a need or an industry - space, time, market segments, and non-market factors. There remains one most important factor since we’re talking about Internet products - online vs. offline. In a given market, demand may outstrip supply in the offline world while there’s an abundance of supply in the online world. The Internet has indeed a profound impact on demand-supply conditions.

Food

Take food delivery as an example. In the past, you go to restaurants to eat. No matter where you live, the demand and supply of food are affected by space. A McDonald’s right next to where you live and a McDonald’s two kilometers away are valued quite differently by you. Depending on the location, every restaurant has an effective radius. If we were to believe in the efficient market hypothesis, then the supply of restaurants is in a dynamic balance with people’s demand for non-home-cooked food.

(Note: I’m gonna reuse the graphs from the last issue. Without food delivery, the limiting factor is your willingness to walk; you have ~1.5 burger joints available to you.)

With Internet and food delivery, the McDonald’s right down your apartment is basically the same as the McDonald’s two kilometers away. The two McDonald’s that were in balance with the demand in the offline world now may be oversupplying with Internet-enabled food delivery.

(Note: With food delivery, the limiting factor is for that much delivery fee, how far the restaurant and/or the delivery staff is willing to travel for you. Each restaurant has a bigger serviceable radius. You have 3 burger joints available to you.)

This will impact McDonald’s business in many ways. They have to rethink how big their stores need to be, the density of the stores, etc. If food deliveries were a major part of the business, the restaurant can have a bigger back of the house and a smaller front of the house. The location of the restaurant can be in a more inconvenient part of the street since they don’t need to draw in foot traffic.

This also has implications for Meituan’s business. If we already have a McDonald’s on the platform within this 1-km radius, the offline sales team doesn’t need to try and get another one. Of course, in actual operation, we may not have the spare energy to think about this. But in theory, if a company can optimize to such a level that they don’t waste energy trying to get another McDonald’s in an area when they already have one, they would be more efficient.

Food delivery will make McDonald’s business more efficient as well. They can open delivery-only stores that don’t need front of the house and don’t need prime real estate to draw in foot traffic. The stores can also be designed for express checkout for the delivery personnel.

"You're not in the burger business; you're in the real estate business." - Harry J. Sonneborn (McDonald’s first president & CEO)

Shopping

Take e-commerce as another example. When we shop things offline, whatever is it that we buy, we’re limited in how far we’re willing to go to get it. We’re also limited in our ability to query items physically. These two limitations result in three outcomes, two of which are failures:

1. There’s no supply of such a good in your actionable radius. (failure)

2. You can’t find it even if there’s a supply. (failure)

3. You find what you want. (success)

After Taobao moved all these goods online, the actionable radius doesn’t matter anymore. As long as it’s not something you need immediately, even if there’s just one supplier in the entire country, you’ll be fine. As such, it results in a severe oversupply in the online marketplace while the offline world is in dynamic balance. The advent of the Internet industry and its development has broken the demand and supply equilibrium of the past. As it’s being broken, a new equilibrium is built with new business forms.

Building New Business Forms: Meituan Delivery Case Study

Path dependence is Newton’s First Law in business

When Meituan just started our food delivery business, most of our offline sales team came from our group buy business. In our group buy business, our initial guiding principle was “mad visits, mad deals” (Note: Figurative meaning - visit as many merchants as possible, list group buy deals as early and as many as possible). This in essence is to relentlessly increase the supply.

Our first target customers for food deliveries were college students. When we hit the campuses, I felt the “mad visits, mad deals” strategy wasn’t working. I told the team, we have to break our path dependency - “mad visits, mad deals” doesn’t work on campuses. But it was to no avail, people didn’t listen. The force of path dependence was too strong.

This is a pattern in cognitive psychology - [confirmation bias]. People only want to listen to things that are congruent with what they already believe in, and they tend to ignore contrary data. When an article is lauded on the Internet, it’s usually not because it taught people something new, it’s because it confirmed what they already believe in. Path dependence is Newton’s First Law (i.e. the law of inertia) in the business world.

As such, I tasked one city manager to open in a new city, and they can only open in one school. He recruited 6 salespeople to operate in the school. The salespeople continued on with the “mad visits, mad deals” mantra. But after a while, the orders remained flat.

Remember that Meituan and Ele.me were very different back then. Ele.me had already established a foothold offline. They were helping the restaurants to manage their delivery orders, so the restaurants they were bringing online already have delivery demand. They were operating in just 12 cities. Meituan however immediately started off with 20 plus cities, some of which Ele.me didn’t think were suitable for food delivery.

This particular school that we experimented in had hadn’t much exposure to food deliveries, so there was no demand. Then we onboarded so many restaurants from the get-go. Each restaurant had very few orders.

At that time, the delivery service was provided by the restaurants themselves, not the platforms. For restaurants that serve foot traffic, chances are they don’t have spare manpower to provide deliveries. As such, the restaurants would prioritize the in-store customers first - they were greater in number and they have more interactions with the restaurants.

Here we have a negative feedback loop. Few food delivery customers mean few orders for the restaurants. → The insignificant business results in poor service from the restaurants. → Poor service from the restaurants results in a negative product experience for food delivery platforms.

Chase the Right Rabbit

At the time, the Chengdu city manager did something different. They opened in one school and only listed eight restaurants, and then spent all their energy doing promotions in the school - handing out flyers, giving out discounts, etc. He told the restaurants, “Tomorrow you’re gonna tons of orders flooding in, you’d better hire some help.”

Due to the aggressive promotion, the order numbers start to show. Since only eight restaurants were listed, each one of them got a significant number of orders. These restaurateurs then realized the strong demand for food deliveries.

After each day, the team would evaluate which restaurants didn’t provide a good delivery service to the customers and would advise them to improve. Then Meituan would do another round of promotion. As the service improved, the customer retention rate also increased.

One iteration further, as the orders and customers’ user experience go up, more restaurants were added, but not too fast. As the number of restaurants, the number of orders, customer experience, and delivery efficiency grow in tandem, the operation began to run smoothly.

Thus, it’s critical to recognize at any phase of the business whether there’s a demand shortage or a supply shortage, as well as know how to deal with each situation. If you got it in reserve, the whole business is doomed.

Knowing this allowed Meituan to meet consumers’ demand with just a few restaurants and when the food delivery experience wasn’t that great.

Keep On Evolving

Later on in 2015, when Meituan came out from college campuses to work on the white-collar market, our sales team was already highly trained. Our core competency is “groud push”. (Note: offline sales & promotions, onboarding merchants.)

We’re at such a level that suppose we were to operate in a school, we’d know how many buildings are there, how many people are in each building, and we’d draw out the battle map, and do our sales and promotion. On the second day, we could calculate how many users we added in each building. Then based on the new users added to each building, calculate compensations for the ground salespeople. The management sophistication is very high.

Even though we were confident in our abilities, we flopped hard in the white-collar market at first.

First of all, ground sales don’t work because you can’t get into office buildings. So we stationed loudspeakers and booths near the subway entrances during the morning peaks. And we were chased away by Urban Management police.

We kept trying but the orders were not growing. Finally, we realized that the needs of white-collars are quite different from those of the students.

White-collars have a higher standard of restaurants and they have higher requirements for delivery speed. In addition, a high percentage of restaurants in the business districts belong to restaurant chains and each shop doesn’t have much autonomy. Restaurants in schools are usually self-owned. You asked them to recruit a delivery man, they can get it done in a day’s time. The branded restaurant chains, however, they’ll ask you what should the delivery staff wear and they'd need to request uniforms from the headquarters.

At this point, it’s important to switch to a supply-driven strategy (i.e. onboard those who can offer delivery options asap). Pretty soon we realized that a lot of good restaurants don’t have the ability to provide a food delivery option. Then a platform-operated delivery fleet becomes important. And that’s where we are today.

Find the article insightful? Please subscribe to receive more learnings from Chinese Internet companies and I’d appreciate it if you can leave a comment or help spread the word!

Subscribe now

Subscribe now

Factors Affecting Demand-Supply: Non-market Factors

Despite the attention and emphasis that we place on market factors, the non-market factors are also numerous and consequential, even in [a supposedly free-market economy like] the U.S.

For example, in September 2019, the state of California passed a law (Assembly Bill 5) that would require Uber and Lyft to classify their drivers as employees rather than as independent contractors and provide them with mandated employment benefits. (Note: Uber, Lyft, Doordash, Instacart, and Postmate then supported Proposition 22, passed in November 2020, which grants them an exemption from this law.) This is a classic non-market factor, from the legal side.

Non-market factors usually don’t matter in our operations, but when they do, they matter a great deal. Take ride-hailing for example. Ride-hailing is among the sectors in the Internet industry that are most affected by non-market factors. From the birth of Uber to the global popularity of ride-hailing, Uber has always had to deal with various authorities from governments of the world, [on top of industry competition, and so do Uber’s competitors]. Some regulators are more amenable to ride-hailing while others are less so. London revoked Uber’s license to operate in the city twice in the past and Uber ran into a lot of problems in Japan too.

(Note: Legality of ridesharing companies by jurisdiction)

The topic of interest here is how the non-market factors affect the operation of businesses and the demand and supply conditions. They are harder to understand than market factors because they work in much more complicated ways and it requires you to look at things from outside the industry. Non-market factors usually have their reasons to exist. To the free-market liberals, these non-market factors may seem completely unreasonable, but a lot of liberals are not fully open-minded.

Roads

Let’s start with a simple and mundane example - roads.

Roads are generally considered to be public goods. They are public infrastructure but they carry various market costs, like construction costs and maintenance costs. They also carry other indirect costs. If the roads are too wide, it may not be conducive for retail businesses in the city. People can’t see what’s on the other side of the street or are unwilling to walk over; a convenience store can only serve half the street or even less.

In additional, not only do you have a road in front of the block, but you also have roads at the sides of the block. For people who come from the side of the block, it’d be quite a distance to get to another block. Unfriendly traffic design to convenience stores is unfriendliness to consumers. Furthermore, in countries like Japan, convenience stores are a major employment source. They are also called mom-and-pop stores for a reason.

(Note: This part can be hard to understand without visualization. Hence I drew up the graphs below to help.)

(Note: To solve this, you need to apply the concept of Manhattan Distance with some modifications. Suppose our block is a square with sides of 60 meters, similar to the 200 ft x 200ft city blocks in Portland - What Is the Ideal Size for a City Block? And our road is a two-direction single lane road - each lane 3.5 meters, plus 3 meters each side for pedestrain walkways. Thus, the road in total is 13 meters wide.

Bob at the bottom-right corner of a M2 block wants to get to the store. Without diagonal crossing - hence applying Manhattan distance, he has to cross the road at least twice. The steps in Manhattan distance translate to how many times you have to cross the road. Depending on where you are at your block relative to the convenience store, you have to count the sides of the block into your walking distance as well. In Bob’s case, his total walking distance from the store is 60 + 13 + 60 +13 + 0.5 * 60 = 176m.)

(Note: With our previous foundation, we can now answer how the width of the roads impact retail businesses in cities. Suppose the walkable distance is 400 meters. Beyond that, people just don’t bother or they go to another shop. Our block size and road width are same as before - 60m and 13m. Within the walkable distance, all corners of M5 blocks can be effective serve by the convenience store while some corners of M6 blocks can be covered as well.)

(Note: Now with wider roads. A Class I city road in China is 80 meters - two-direction eight lanes plus verges and non-motor vehicle lanes as well as 15m-wide pedastrian walkways on each side. Our block size is still the same. Within the 400 meters walkable distance, now we can barely touch the M3 blocks.)

As such, there are limits as to how wide the roads should be. In that limited space, who gets to use it is the debate on the usage of public infrastructure. If you’ve been to the U.S., you may see a carpool lane in the middle of the highways. Only cars with multiple people in them can use this lane. If you use this lane, it means the road is better utilized - for the same period of time, same distance, and the same width of space, the road carried more people. Public infrastructure is used efficiently.

In another example, the bigger cities are all developing their public transportation. Subways are highly developed in Japan. (Note: Tokyo Has Built The World's Best Subway.) Subways are highly efficient in terms of the usage of roads. They don’t even take up surface space, and you can build multiple storeys underground.

In urban planning, you’d need to answer questions like how much time of the day a lane is reserved for buses, or how many carpool lanes you want to have on your highways.

Ride-hailing as compared to public transportation would be a much less efficient mode of transportation. Therefore, we cannot have too much of it. However, if you limit it too much, it’d be hard for people to get a ride.

Now we get to a philosophical question of should the need of getting a ride whenever you want be unlimitedly satisfied? Obviously, roads as a public good can’t support it.

An Ecosystem Perspective

You could also see these non-market factors as market factors as well. It’s just that the “market” is bigger than your industry. We need to look at it from the whole ecosystem perspective.

(Note: There’s an argument that some startups succeed by regulatory arbitrage - here and here. Uber is cheaper than taxis by avoiding licensing, labor protection, insurance, safety inspections, etc. and Airbnb is cheaper than hotels by skirting zoning laws.)

If we had priced ride-hailing higher, would that solve the problem? After all, before Uber, taxi rides were expensive.

Here comes another question. Using pricing to determine who gets to use the road, is that fair?

Maybe I’m not in the income bracket to use private transport every day, but I do today because I have to send my family member to the hospital quickly. Are you not gonna provide me with this transport option?

As such, we shouldn’t use a pure market approach to solve everything. The non-market factors may cause some markets to be artificially imbalanced in terms of demand and supply, but these imbalances may have a rationality to them.

In these kinds of markets, the tools that we have been using in managing demand and supply may not work anymore. You have to adjust your business model, product design, and operations to the context.

Find the article insightful? Please subscribe to receive more learnings from Chinese Internet companies and I’d appreciate it if you can leave a comment or help spread the word!

Subscribe now

Subscribe now

Factors Affecting Demand-Supply: Segmenting

Segmenting is not easy. Different industries require different approaches and each segment within the industry would have its own demand-supply condition.

A common situation in segmenting is that the high-end segment has an abundance of supply while the low-end segment has a shortage.

Let me give you an example. In The Innovator’s Dilemma, Clayton Christensen was confident about his theory. However, a good explanation of the past isn’t enough to count as a good theory, it has to be able to predict the future. As such, Christensen predicted the rise of the electric vehicle (EV) industry, and that EVs will replace the fossil-fuel-burning cars that we have now. The path for EVs’ ascend will follow that of low-end disruption - starting with a niche market before eventually moving upmarket to take over the whole market.

The EV industry has all the right ingredients for a low-end disruption. EVs have fewer components than fuel cars, so theoretically it should cost half of what it takes to make a gasoline-power car to make an EV. And electricity is cheaper than fuel. In all, EVs should be a cheaper option, and it’s prime to enter a low-end niche market that incumbents don’t deem profitable, like school buses or mobility aids for the elderly (Note: where range and speed are not as much a concern).

Unfortunately, the late Christensen got it wrong. Elon Musk started on the high end. First with sports cars (Roadster), then Model S, Model X, Model 3, and the upcoming Model Y. We shouldn’t discount Christensen’s theory just because of this. The disruption theory is one of the most important works on innovations. What he probably didn’t consider is segmenting.

(Note: For the record, Chistensen didn’t intend to predict the future of EV. He wrote the chapter as how he would approach the problem. And he did get some things right.)

I’ll also amend something I said earlier - that there are only two situations: demand outstrips supply or supply outstrips demand. That’s still true, but it may mislead you to focus on the surface. What’s worth paying more attention to is the underlying rule that brings about either situation.

Why did Telsa start with the high-end? One important attribute of the demand-supply condition with regard to high-end and low-end is [price] elasticity. Regardless of whether it’s an oversupply or undersupply, in the high-end market, both supply and demand have a greater elasticity (more responsive to price change). In contrast, on the low-end, both demand and supply are more inelastic (i.e. things tend to be necessities). The high-end customers have more disposable income, the relative cost of trial and error is low, and they can afford to buy a lot of products until they find something they like.

The Hierarchy of Human Vanity

Here’s the hierarchy of human vanity.

First, it’s the necessities. In the earlier years of consumerism, people compare whether your household has a TV, a fridge, an air conditioner, etc.

Then the degree of necessity declines, you will be okay without it but it’s still a case of haves vs. have-nots. iPhones are great, but Xiaomi works fine too. Phones have a particularly strong showing-off property since it’s in your hands all the time.

One level further, the item has even less practical value and is primarily used for showing-off, like expensive watches.

What comes after that? Well, we’re both rich, but you’re new money, I’m old money. I buy things your parents or grandparents wouldn’t dream of buying.

Further, if we’re both old money, money and prestige don’t matter anymore. I’m gonna compare time. I have a lot of idle time on my hands; I play golf all the time.

Further still, even if I have all the time in the world, I’m just not gonna do anything with it. I’m gonna keep my nails so long that I can’t even eat by myself. I’ll have servants to take care of my daily routine.

Give Them A Reason

For segmenting, people in different classes would have different things that they care about. Elon Musk was dead-on about the high-end customers - not only do they have the riches, but they also care about the environment. So Teslas were sold for their environmentally-friendly image. You’re so rich, try this scarce product that I have. It might not have enough mileage so it’s impractical for normal usage. But, first, you have to be rich to be willing to buy a car that you only occasionally drive, and second, you can show the world you think long-term, you care about the environment and the future of humanity.

From this entry point, let customers feel that they’re successful, and compassionate that they’ll buy a car that they won’t and can’t drive every day.

In the high-end market, the demand is so elastic that you just need to give them one reason they’ll be willing to try. In fact, all new technologies and new products have two paths of entry. One is from high-end, and the other from low-end. Most of the products that enter from the high-end are demand elastic. Even if the products are not “must-have”, and they’re not cheap, some people will be willing to buy if you give them a reason.

(Related: What Happened To Tesla’s First Car — The Roadster?)

Even if an industry would eventually go mainstream, rarely is the product suitable for the masses from the get-go. It has to make its way to the mainstream by premiumization if you’re coming from the low-end or commoditization if you’re coming from the high-end.

Tesla started with the most high-end market of sports cars. As its volume increased and technology matured, the costs came down and performance went up, which enabled Tesla to make products to serve a new segment that’s less high-end but greater in volume, and a positive feedback loop formed.

So, in short, the master plan is:

Build sports car

Use that money to build an affordable car

Use that money to build an even more affordable car

While doing above, also provide zero emission electric power generation options

Don’t tell anyone.

- Elon Musk

Different segments of the same demand would have drastic differences. As the demand goes from high-end to low-end, the necessity of the product increases. This is an important property of demand-supply from the segmenting perspective.

Epilogue

This issue I had a particularly hard time understanding. Wang Huiwen’s core message is simple enough - bringing price elasticity into demand-supply consideration. That’s Economics 101. However, he raised a counterpoint to what I’ve been thinking about startups all along - start small, start niche. And Elon Musk would probably have agreed with him. So I had to reconcile the differences. Below is my process. You can follow along if you want, and I’d be interested to hear what have you learned.

For me, I think “start small, start niche” is still true, but there are a lot of nuances to it. Ultimately, it’s about figuring out what people want.

Disruption Theory

• The original disruption theory: Disruptive Technologies: Catching the Wave (1995) and the book The Innovator's Dilemma (1997)

• A 2015 revision/clarification, considering Uber and Tesla not as disruptive: What Is Disruptive Innovation?

• Two Theories of Disruption

  • New-Market Disruption: Established companies invest in technologies that their mainstream customers value (“sustaining technologies“) and ignore new technologies that don’t serve the needs of their mainstream customers. New entrants enter the market serving niche segments (nonconsumers) who find a use of the new technologies and the performance of new technologies improves at a rate faster than consumer demand (“disruptive technologies“) and they capture more market share as the technologies gradually satisfy mainstream use and eventually dominate.

  • Low-end Disruption: Incumbents provide ever-improving products to their most profitable and demanding customers. These products overserve the needs of less-demanding customers. New entrants enter the market by offering these low-end customers “good enough” products that are cheaper.

• Discussion on “High-End Disruption”

  • Adherents to the HBS school of thought don’t seem to favor such a theory. The Persistent Myth Of High End Disruption

  • By seeing innovation as beyond just increasing performance, high-end disruption takes over the market by increasing affordability. High-End Disruption: Using Affordability to Measure Innovation

• Flaws with Disruption Theory

  • Ben Thompson - 1. What Clayton Christensen Got Wrong 2. Apple and the Innovator’s Dilemma

    • The theory was developed at a time when most buyers are businesses. Unlike businesses, consumers are not perfectly rational. (Bounded rationality)

    • Not every product attribute is measurable.

    • “Good enough“ - some things, like user experience, can never be good enough.

  • I personally find it troublesome that a lot of the debates tend to focus on definitions and putting things nicely into two buckets - is it “disruptive”? If not, then everything else is “sustaining“. I don’t think that’s the whole picture.

    • Ben Thompson rightly points out another kind of technology - “Obsoletive”

High-end vs. Low-end for Internet Companies

One should note the difference between online businesses and offline businesses - the marginal costs of serving an additional customer (zero vs. non-zero, #18 - 4Ps in the Internet Age). The implications are:

• It’s no longer that some companies serve the high-end segments and some serve the low-end segments. Instead, one company can serve everyone. Beyond Disruption. In that light, new theories/strategies need to be applied, like Ben Thompson’s Aggregation Theory or Meituan’s Platform Strategy.

• Innovations have to start small. The high-end segments with a greater willingness and ability to pay could be better as the initial customers. (As is the case for Under)

High-end vs. Low-end across Geographies + Technology Adoption Cycle

Naturally, one of the most important factors affecting disposable income is geography (i.e. developed vs. developing countries). With globalization and the ability of Internet companies to serve just everyone on the planet, it’s interesting to consider segmenting across markets.

• Again, I’m citing Ben Thompson - The End of Trickle-Down Technology

  • A lot of people in developing countries are technology enthusiasts and visionaries who simply don’t have much money. They don’t want "trickle-down” technologies that are passed down from the high-end segments. They want the newest things. New business models and companies need to be built for them.

  • Conservatives in developed countries have been underestimated in their willingness to spend for products with better user experience.

Find the article insightful? Please subscribe to receive more learnings from Chinese Internet companies and I’d appreciate it if you can leave a comment or help spread the word!

Subscribe now

Subscribe now

Factors Affecting Demand-Supply: Space

The demand-supply condition of an industry is affected by space.

Take the online dating market in China as an example. Overall, there is more supply of males than there is demand. (Note: The overall male-to-female ratio in China is about 105:100, but for younger age groups, it’s higher at 112 - 118. Population by Age and Sex 2018, Population by Sex, Marital Status and Region 2018, National Bureau of Statistics).

If we look at the segment of college students, at Tsinghua University, there is also a surplus of males, but at the country level, there are relatively more females in higher education. (Note: 51% of undergraduates, 50.9% of postgraduates, and 58% of vocational school students are female. Source: China National Program for Women’s Development 2011 - 2020, National Bureau of Statistics.)

In rural areas, the male-to-female ratio is even worse. Females usually have an easier time migrating to cities, therefore, in urban areas, and especially in big cities, there are more females in the reproductive age groups than there are males.

The gender ratio will affect a lot of things. If you’re making an online dating site targeting urbanites or one that targets the rural market, which side should you promote to? Which side is more likely to be paying members? The two options may have completely opposite answers.

Factors Affecting Demand-Supply: Time

How frequently the demand-supply condition changes varies from industry to industry.

For example, in the ride-hailing industry, the demand-supply condition changes at a rapid pace. In a typical one-week cycle, it’s undersupplied during the [weekday] morning peaks, balanced at noons, oversupplied in the afternoons, and then undersupplied again in the evenings and nights.

(Related Reading: Understanding Supply & Demand in Ride-hailing Through the Lens of Data by Grab)

Some industries change very slowly according to time. At the macro level, the demand-supply condition of the retail industry is slow to change. It probably takes a whole lot of changes in human habitation, energy costs, transportation, etc. to bring about a significant change in the retail industry. In the past, that was about once every 30 years. Some changes are due to the changes in economic, political, and societal conditions (PEST). Some changes are due to the changes in industry practitioners’ cognition and operational competency. These changes speak to a general trend - as the retail industry continues to innovate, the gross margin declines.

Every 50 years or so, retailing undergoes this kind of disruption. A century and a half ago, the growth of big cities and the rise of railroad networks made possible the modern department store. Mass-produced automobiles came along 50 years later, and soon shopping malls lined with specialty retailers were dotting the newly forming suburbs and challenging the city-based department stores. The 1960s and 1970s saw the spread of discount chains—Walmart, Kmart, and the like—and, soon after, big-box “category killers” such as Circuit City and Home Depot, all of them undermining or transforming the old-style mall. Each wave of change doesn’t eliminate what came before it, but it reshapes the landscape and redefines consumer expectations, often beyond recognition. Retailers relying on earlier formats either adapt or die out as the new ones pull volume from their stores and make the remaining volume less profitable. - The Future of Shopping, Darrell Rigby for Harvard Business Review

Changing slowly could be a scary thing too. A 30-year-old young man may have enough business and life experience suited to be an entrepreneur in the retail industry. And when he’s 60 years old, he may have made something for himself. Just when he’s about to retire, the industry undergoes a transformation.

Walmart is probably the best offline retailer in the world in terms of revenue and market cap. But when they started, they were the underdog. They faced a powerful, much larger competition, Kmart.

People usually think the greatest advantage in the retail industry is cost leadership from Porter’s Three Strategies. An advantage in the cost structure leads to cheaper prices, and cheaper prices win over customers.

At that time, people thought cost leadership mostly come from bulk purchases to get better prices from suppliers. The greater your sales and order quantity, the stronger your bargaining power with suppliers, and the greater your competitiveness. As such, in the retail industry, the brand with more stores and more sales would be more competitive.

When Walmart just started competing with Kmart, Kmart already had thousands of stores while Walmart only had a few dozens. So for quite a long time (1962 - the 1970s), Walmart didn’t compete with Kmart directly - direct competition means opening a store within one of Kmart’s stores’ (say 5, 10 miles) radius.

Kmart’s founder Harry B. Cunningham retired in 1972. As he was stepping down, he told his team to watch out for Walmart, although they’re still small now, we need to keep an eye on them. After Cunningham retired, Walmart opened a store in Kmart’s territory and competed successfully, and the rest is history.

This is how it goes in the retail industry. A disruption happens only once a few decades. The disruption usually stems from the change in demand-supply conditions. The change might be slow, but it’s scary.

(Note: It doesn’t help that many retailers became the subjects of financial engineering by private equity and hedge funds and not actively managed by operators anymore.)

Organizational Implications

The pace at which the demand-supply condition changes has a meaningful impact on companies.

The food delivery business is similar to retail. Walmart has to operate store by store. Food delivery companies have to operate cluster by cluster (”Honeycomb”). The success of one store has little bearing on another store, other than the collectively larger scale which may bring some benefits.

Walmart’s slogan is “Every Day Low Price”. How are they able to do that?

Most people’s first response would be because Walmart has so many stores, they have economies of scale in the supply chain, so they can have lower costs. But that’s not true. The very first store that Walmart opened already achieved everyday low prices. The lower prices didn’t come from economies of scale. (Note: “Every Day Low Prices” as a strategy was adopted in 1974 when Walmart had 78 stores and Kmart probably had 1,000+ stores.)

Additionally, when your store carries a great variety of goods, your operation would become exponentially more complex. It’s the same in food delivery. When your area has more merchants, more consumers, and more delivery staff, your operation becomes more complex.

How to improve the operation? You’d need people dedicated to solving this problem.

At that time, Meituan didn’t have a team that does this kind of job, so I wondered if we can poach a team from another company. We first looked at Internet companies, but they didn’t have this kind of talent. Then I asked HR to look for people who work at the nexus in offline retail, but no luck either. Finally, I asked the offline retail bosses why they don’t have these kinds of jobs which are so important. And they told me when they have such a need, they just go to consulting companies like MBB. So I poached some consultants to form the business analytics team of Meituan.

I wondered, “why the retail companies which face complex operational and organizational problems don’t have dedicated business analysts on their teams?“ Following that thought, I looked at which other jobs they should have but don’t. And I discovered that most retailers don’t have a product R&D team either, but Walmart does.

After much deliberation, I think this is because their industry is too slow-changing. Business analytics and product R&D jobs won’t have enough to do in a slow-changing industry. These teams tend to be costly. In a slow-changing industry, the ROI of having such a team is very low, so low that it’s cheaper to hire consulting firms when the need arises.

Similarly, for their technological needs, most retailers won’t develop their own software or have a software engineering team. Their software probably comes from SAP.

As such their rate of iteration is very slow. When the industry undergoes a transformation, and the company doesn’t foster talents that respond to changes. The external people won’t be able to deeply understand the company’s operation and is slow to iterate. When the industry undergoes a change, and the external solution is not able to respond as fast as the change is taking place, the company will be disrupted. This is why it’s scary when the industry changes very slowly.

When you run a company, you will face the question of whether to develop a certain organizational competency within the firm or not. Be it a tech team, product team, business analytics, or else. These are important strategic decisions, and it depends on the characteristics of your industry.

This is one of the most important organizational implications of the change in demand-supply conditions in response to time. Some industries change rapidly while others are most slow-moving. One is not necessarily better than the other. The key is to understand what fast and slow mean.

We also need to consider whether the demand-supply conditions changes are on the macro or micro level. For example, the demand-supply condition in the retail industry is fast on the micro-level. The SKUs usually have a seasonality, so it’s measured in seasons, which is pretty fast and can be hard to manage. Retailers usually outsource the hard-to-manage SKUs and adopt a partnership model. For example, in a supermarket, for a category that changes every season, they may bring in a partner to have a store-within-a-store. (e.g. cosmetics)

Find the article insightful? Please subscribe to receive more learnings from Chinese Internet companies and I’d appreciate it if you can leave a comment or help spread the word!

Subscribe now

The Importance of Getting Demand & Supply Relationship Right

Most of what we’ve covered so far is on the consumer front, and it’s meant to help you understand more easily. When it comes to the business side, the relationship between demand and supply becomes very important. You can even say it’s everything - almost all things related to demand and supply will affect your strategy.

Understanding demand and supply is hard. Although you can only be in one of two situations: 1. demand outstrips supply or 2. supply outstrips demand, it’s hard to know at any point in time, which situation you’re in.

In our day-to-day work, these are the common problems that we’d encounter that have to do with demand and supply:

1. The team doesn’t proactively determine the situation of demand and supply. As a result, the operation lacks focus, and the approach is basically “throw it against the wall and see what sticks“.

2. The team knows that demand and supply each affects the other, but can’t make a clear judgment about which one is more important.

3. The judgment about demand and supply is correct, but the operation is not guided by the prevailing demand and supply conditions.

If any of the above three situations occurs, then the team would usually work in the opposite direction of the demand-supply condition. In fact, they may even have a strong incentive to get it wrong!

For example, if a company is in a situation where supply outstrips demand, it should be doing more work on the demand side. In reality though, more often than not, it’s still pushing hard on the supply side.

Why?

If supply outstrips demand, then the demand is very important. The demand side buyers know they’re important, and they’d be very hard to entertain. In comparison, the supply side sellers would be a breeze to spend time with because they’re the ones who are in a hurry and have something to ask for. As such, the team would have a strong incentive to continue working on the supply side and pretend the harder-to-deal-with side is less important. This was a frequent occurrence in Meituan.

Location, Location, Location

Here’s another example from the retail industry to demonstrate how frequently people get the demand and supply relationship wrong.

I asked the bosses of convenience store and supermarket chains, “In the retail industry, which one is more important, demand or supply?“

Without exception, all of them answered that supply is more important.

It doesn’t gel with my business common sense - with the industrialized manufacturing that we have today, most goods should have more supply than there is demand.

Therefore, the next question to figure out is - is this how they think, or is this how they act?

So I asked them another question, “At your company, what work do you absolutely have to do yourself?“

They all answered, “Choosing the location“. One of them even said he personally chose the location for close to a thousand stores they have all over the country.

For retail, location is the aggregation of demand. They say supply is more important, but their actions are pretty revealing.

(Note: “location, location, location” in the real estate business means supply is paramount.)

So why would they say supply is more important than demand in retail if they don’t really believe so?

Firstly, if the bosses take care of choosing the location (demand), then the other people at the company need to work on the supply. They can’t be telling their employees that supply is not important, what really matters is the demand. They need to let the employees know their work is important and valuable, so they stress the importance of supply day in and day out, and it becomes something of a habit.

Secondly, overall, the company spends much more time on supply than choosing the location of the stores. Choosing the location is insight-heavy while supply is operation-heavy. There are so many details and tons of work dedicated to supply. People naturally think that the things that we spend the most time on are the most important things, but that’s not always the case. We need to be conscientious in recognizing what the important things are, then allocate our time and resources to those important things, so that we can have a better ROI. It’s always the deficient side of demand-supply.

Lastly, the bosses say that supply is more important and tell you the hard work they have to do on the supply side to make the industry less attractive to new entrants. If they tell you that getting the location right is the most important thing in the retail industry, it makes the industry look as if there’s no barrier to entry at all, and many people would enter the playing field.

Consequences of Getting Demand-Supply Wrong

The consequence of getting the demand and supply relationship wrong is dire. A recent example is Renrenche. (Note: A used car marketplace; raised US$760M from Tencent, Didi, and Goldman Sachs; once one of three top players in the category; imploded with founder departure and fire sale in late 2020). The Renrenche team knew consumer psychology well and they improved a lot on the demand side by optimizing for a better consumer experience. However, the used car industry is one where demand outstrips supply. If you don’t have good quality second-hand cars, no amount of good consumer experience makes a difference. This is where Guazi (Note: raised US$3.9B.) and Renrenche differ. Most of Guazi’s operations focused on the supply side.

Solving for the lacking side

In the jobs market, companies are recruiting and people are looking for jobs. It’s hard to know which one is more lacking, the jobs or the job-seekers.

On the Meituan platform, is there a lack of service staff job opportunities from merchants or is there a lack of service staff candidates? Overall, it’s the people that are more lacking than posts.

Internally, we’ve worked on a recruitment product for merchants before. When the product team reported, they said that all the merchants have a need to recruit staff. Therefore, it seemed that there was a demand for this product.

However, they’ve got the demand-supply relationship backward. Labor doesn’t want the service staff job because it’s hard work and low pay. To conduct research into whether merchants have a demand for a recruitment service is useless. Of course, they do. What’s more deficient in the industry is the supply of service staff. Only when you have solved the problem of labor doesn’t want to take up the service staff job, did you make a contribution to merchants and society.

Following that thought, if the supply of labor is what’s lacking, remember that the supply and demand of things are affected by price. In labor-deficient countries, they usually have two solutions. One is immigration. The other is career progression with skills training and salary increase. For example, different levels of service staff get different certifications, and the best service staff work in five-star hotels. Therefore, offering skills training programs and certification systems are probably more effective than a recruitment site.

(Explanation:

At the prevailing market equilibrium, the market-clearing wage - what restaurants are willing to pay and what labor is willing to accept, for food service staff is P0. The total labor force in the food service industry is Q0.

With the convenience, variety, taste, etc. that food delivery offers, more people would want to order food from restaurants as opposed to cooking at home → the demand for restaurant food increases → the demand for restaurant staff increases, D0 → D1.

If restaurants think they can still recruit staff with what they’ve been paying, P0, and based on the demand they’re experiencing, they’d want to have QD amount of staff. The perceived labor shortage is therefore QD - Q0, and this is what they’d say to Meituan people if being asked, “do you need a recruitment service?”

It is need/demand for a product, but you can’t fulfill it without changing the fundamental market dynamics. You have to increase the supply of labor by paying more, whether in wages or other forms of remuneration.

As wages increase, some demand for labor gets competed out, before reaching the new market equilibrium at P1 and Q1.)

Find the article insightful? Please subscribe to receive more learnings from Chinese Internet companies and I’d appreciate it if you can leave a comment or help spread the word!

Subscribe now

The Microscopic Needs

What we have covered in the previous lectures are the Big Needs that create a new industry category. Recognizing Small Needs is equally difficult because it’s very hard for us to see what those needs really are.

Generally speaking, almost no user will directly tell you what his/her needs really are. What they’re saying is usually demands or requirements, and rarely do they speak to their needs. We’re not talking about the economic need here (i.e. quantity demand at a certain price), but the underlying reason that triggered the idea in their mind to want something / seek a solution. For example, the company’s sales team may say that they need a CRM (customer relationship management). But CRM is quite a complicated product, you can’t be making a Salesforce for them right?

(Note: We may be a little too punctilious about the wording of needs, demands, requirements, wants, etc. here and this might still sound confusing to you. How I understood this point here better is actually through the language of Supply and Demand. Demand Thinking - Episode 1: Feature requests aren’t demand. Watch from 0:00 to 9:47.

When your users say they need a CRM, or permission controls in a project management software, these kinds of feature requests may sound like needs or demand from the users, but they’re actually supply-side ideas - they already have the solutions in mind. It’s no different from the ideas that a product team or higher management think will work. Or in design language, form vs. function.)

On User Research

In our daily lives, a girl in a relationship probably won’t say directly what her needs are. She may just be mad sometimes. If you can figure out why, that’s when a happy life starts. At least, that’s the case for me, haha.

At work, clients probably can’t tell you what their real needs are either. Like what Henry Ford said, “If I had asked people what they wanted, they would have said faster horses.”

Well, you say, I’m a product manager, I’ll conduct user research to find out for sure. But user research can be tricky. In user research, there’s a thing called the Hawthorne Effect.

During Frederick W. Taylor’s Scientific Management (or Taylorism) era, they conducted numerous studies to find out how to improve the efficiency of assembly line workers. One such study varies the light intensity at the manufacturing plant to see if it affects workers’ productivity. And they found at both higher or lower levels of light, workers’ productivity had increased.

It wasn’t because the current light intensity was the worst for productivity, but as workers knew they were being observed, they worked a little harder for that period of time. These kinds of confounds and artifacts are common in social science experiments.

Human psychology is a highly intricate system. What you can observe is the final outcome. You have no idea how many twists and turns it had gone through inside someone’s mind before being manifested in the physical world. You can’t tell whether it’s the truth or a lie. Sometimes, the person you’re asking doesn't know either.

Suppose you’re sitting face-to-face with another person. If you stay silent, the other party might just start talking. Many people would be willing to say anything to break the awkward silence.

If your product experience is not great, conducting user research may not help you much, because rarely can users truthfully feedback on their product experiences. For one, if others can use the product but I can’t, it makes me look stupid. So in the feedback, I probably won’t give too many details in how I operated, avoiding things that reflect badly on me.

When we encounter setbacks in life or work, telling the story may make us feel super vulnerable. We’d be too embarrassed to recount the story truthfully and what we let out is just a mood.

There’s an app for that?

Sometimes, a company facing some internal problems may ask the product managers to make a product to fix everything. However, if the underlying problem is really a management problem and not a tools problem, then you won’t be able to make a good product. Because you don’t understand the real need for the product.

Of course, it’s not very realistic to expect your boss to be a great leader. The law of gravity in management is the Peter principle: people in a hierarchy tend to rise to “a level of respective incompetence” - they are promoted until they reach a level at which they are no longer competent.

Can you even product?

Meituan initially had 3 versions of CRM and it was due for an update. I was in charge of product and marketing, so this was under my purview. However, I was having too much fun with marketing and PR, so I had no time for product. As such, I delegated the task of developing a new CRM to willing subordinates.

Once they were done, the product was not usable at all. Even after another round of modification, it’s still not up to expectation. In the end, even Wang Xing (CEO of Meituan) questioned me, “Dude, can you even product?!”

“Can you make a good product?” is a question that a product manager will face throughout his career. Even if you’ve reached the god-level like Allen Zhang (product leader for WeChat), every day, a billion people want to teach him how to make a good product.

(Note: Here is a popular rant on WeChat’s UI - “In 10 years, Allen Zhang made WeChat an ugly monster”. It’s in Chinese but I think through the browser translation you’d get the picture. Below, I translated one picture for illustration.)

In the entire Chinese internet industry, there are no more than 20 people who have earned the industry recognition that they can make good products. So I had no other option but to work on the CRM myself. Luckily, at that time (late 2011), we were able to convince “Ah Gan” from Alibaba to join Meituan.

(Note: Gan Jiawei, employee No.67, ex-VP of sales at Alibaba, ex-COO of Meituan-Dianping, now a partner at Hillhouse Capital. Regarded as one of the best in B2B sales in the Chinese tech industry.)

I told Ah Gan that the previous two versions have failed and it’s critical for us to have a great CRM software. We only have a dozen product managers at our disposal, so I made a list of requirements to let Ah Gan choose 3 from. After we were done, the CRM product we have was the best in the group buy industry.

(Note: Some context for the importance of the CRM. Ah Gan has a formula for sales: Revenue = Customer Unit Price * Prospects Density * Area * Efficient of Customer Visits * Close Rate. At Alibaba, their customers were the suppliers of goods, and they were located in industrial parks and the like, so they’re highly dispersed. Their customer unit price was ¥50K/year, which was considered high, but the prospects density was low and it’s hard to close. As such, the sales reps need to cover a lot of ground, sometimes even across provinces to find customers. Towards the end, each sales rep served about 50 customers, and with a unit price of ¥20K/year - the product became cheaper as is the case for tech, the avg. annual sales per rep was a respectable ¥1M/year.

For Meituan, it’s a different case. During the “Thousand Groupons War”, Meituan’s customer unit price was almost zero, because merchants pay for performance, not upfront. However, their prospects density was super high - restaurants and service providers are clustered in urban areas, so they were relatively easier to visit and easier to convert. For a B2B business, the salespeople are its production capital and you have control over two variables - scale and efficiency.

When Ah Gan joined Meituan in Nov 2011, the sales team was about 1,000 and they had about 20-30K merchants on the platform, and on average, each rep served 27 merchants. That’s much lower than Alibaba, which has a higher customer unit price. Based on the expected customer unit price and prospects density, if Meituan’s CRM can support, each sales rep should easily serve over 200 merchants.

As for how big Meituan needed to be, if they’re talking about serving local services providers all over the country, which was estimated to be over 6 million in the country at that time, with a 200-to-1 merchant to rep ratio, their headcount would be 20-30K people.

At IPO, Meituan had 28,458 employees in sales, marketing, and business development and 4.4 million active merchants.)

You can only guess

This is the “rough and ready” that we talked about before. Why we couldn’t make a good CRM is because people from product, sales, business intelligence, HQ, and the local teams all want different functionalities. They also have different ideas on how to manage the sales team. Even if different people like Ah Gan and Wang Xing have made the same requirement, they may have used the same words to convey different expectations. And the end product didn’t meet either one’s expectation.

A lot of the time, you can only guess what people’s needs are. Therefore, the accuracy of your guesses is very important. One way of guessing is to offer options for the product features and based on people’s selection of the options to deduce the underlying need.

Like when girlfriends are mad, the boyfriends keep asking what they want to do, where they want to go, etc. One possibility is that through the elimination of wrong options, you arrive at what she really wants. The other possibility is after exhausting all the options, you find that her needs can’t be satisfied. Thus, we have to continuously offer options until we find what users really want.

For the longest time, Meituan has always wanted to make an internal tool to score merchants. At first glance, it should be simple enough to build and many departments do feel the need for such a tool. However, when the first version was out, no one was satisfied.

So I asked Ah Gan, do we what to separate merchants with high and low customer satisfaction rates? He said nope. Then I asked, is it to separate merchants with high sales and low sales? No again.

Finally, we figured out that the quality control team doesn’t want to separate merchants by sales or customer satisfaction, they want to isolate the merchants who have taken advance payments but can’t pay Meituan back. They are the most likely to leave the platform and hurt the company. For the same “merchant scoring” need, what the sales team wants and what the quality control team wants are totally different.

How to listen to customers

It’s like this not just internally within the company, where different teams want different things, external customers are the same. It’s common to hear from your customers that your product lacks a certain feature. But when you actually made that feature, they may not necessarily use it. They only said so because you asked. They just may not be your target customer.

Customers may potentially have real answers, but it’s difficult to dig them out from them. If the product is hard to use, they may not directly say it, as it may make them look stupid or it may embarrass you.

In the early days of building a product, it’s probably futile to ask people who don’t use the product why they don’t use it. The more effective method is to ask the people who use the product. These people must have a certain degree of approval for your product, but they might just be facing some problems with it.

(Note: In the Segmenting section, we’ve mentioned Superhuman’s engine to find PMF. They first segmented out the users who aren’t of the target profiles, then they focused on the users who would be very disappointed or somewhat disappointed if they could no longer use the product. Same wisdom on whose feedback you should listen to.)

When we talk about needs, there are some misunderstandings that need to be clarified. Many people would be quick to propose solutions. (Note: supply-thinking.) But solutions aren’t the need itself. Like in the CRM example, there can be countless requirements for the CRM. But at that time, the ones that really mattered to Meituan were only a few.

Find the article insightful? Please subscribe to receive more learnings from Chinese Internet companies and I’d appreciate it if you can leave a comment or help spread the word!

Subscribe now

Subscribe now

How to Identify Needs?

Two questions here. The first is how to identify needs. The second is how to think about it. I'll provide a method that I think works based on my personal experience. If all the great needs have been attempted by people in the past, then we need to figure out why they didn't succeed. There are many reasons why they failed, maybe it's because of the team. That's why we need to study the successful failures - those that gained traction or even almost made it but ultimately failed.

From Personal Web Pages to Social Networks

For example, in social, Facebook is the one that reached an unprecedented scale. If we trace back the development of social networks to when the Internet was just beginning, there were many new ideas and phenomena then. One of them is personal pages (made by or for an individual), and many personal pages were pretty successful, reaching tens of thousands of visitors. (Note: There were just several millions of people online.)

Building personal pages is an industry phenomenon. What Need it reflected is hard to say. But as Fu Sheng (CEO of Cheetah Mobile) has said, phenomena are rules manifested. Even if we can't really figure out exactly what need it is, it is manifested in reality. 

Personal pages developed for a while then phased out. Afterward, something called ChinaRen (founded in 1999) came out, and in the US, there's something similar called GeoCities (founded in 1994). These two web hosting services are the next iteration of personal web pages and the user base was also much larger. 

(Note: Cameron’s World - A project that restores some archived GeoCities sites to their full glory. Sound on.)

Why GeoCities and ChinaRen took off? Because it was hard building personal web pages. You need to have content, technology, equipment, and money to pay for web hosting, and you can design. It's rare for an individual to satisfy all these conditions, that's why personal pages never took off. For ChinaRen, the web hosting is free (Note: under a certain web space size, like 2MB as it was for GeoCities.) and you don't need to know how to code, you just work with the modules. This led to several orders of magnitudes increase in the potential user base. Regardless of the underlying need, the barriers to entry decreased, and the ease of use increased. 

However, ChinaRen failed technically. They had a hardware failure and lost the files for three hundred thousand sites. Gone was ChinaRen.

(Further Reading: An Ode to GeoCities)

Around 2000, blogs start to gain popularity. Blogs are a further iteration of web hosting services. For a personal page or GeoCities page, you visit it once to find out who or what this page is about, then that's it, you probably won't visit the page again. As a result, the site builders lost the motivation to maintain the site. 

Perhaps that intoxicating feeling explains why a lot of the pages on GeoCities seemed frozen in the gestational state, their most prominent feature some kind of wacky “under construction” graphic. After the thrill of setting up the site wore off, the creators seemed to get bored of the daily work of maintaining a home page. And what was the point, anyway? After all, it quickly became obvious that setting up a Web page wasn’t a surefire way to find fame, wealth, or dates. That was especially true as more and more people came online, creating a glut of home pages. Now that everyone had one, no amount of flashing text could make your page stand out. — Slate, How GeoCities Invented the Internet

Blogs fixed the format, and encourage you to write new posts every once in a while. Posts are displayed in reverse chronological order, so that the most recent post appears first, at the top of the web page. This tells bloggers and readers that content is updated regularly. As a result, usage by writers and readers increased significantly.

The problem with blogs is that readers don't know when bloggers posted new content. Different writers post at different frequencies. At this time (2005 - 2006), RSS gained widespread use as blogs reached mainstream popularity. 

Even with RSS, there is still a big problem and that is the creators and readers community was still a small segment of the population - people who have a need for cutting-edge ideas. 

At this point, the industry split into two paths.

The first one is Weibo (Twitter). As data costs become cheaper and cheaper, mobiles phone come with more and more powerful cameras, and personalized recommendations become possible with machine learning, the microblogging branch gave rise to Instagram, Toutiao, and Douyin/Tiktok. (Note: Interest-based, content-centric, unidirectional follower model.)

The other branch is Facebook. (Note: social-centric, bidirectional follower model).

When we were working on Xiaonei, it was the heyday of blogging. As such, we were facing a decision on whether to build a blogging feature for Xiaonei. The decision was to build a blog-like product, but should the blog be open to everyone or just friends? If it's the former, more people will be able to see the content, and writers will have a stronger motivation to produce content. But this is an assumption to be tested.

An American site gave us inspiration. Even if you write rubbish every day, your friends would still check out your page. If you make the product to be default open, then the content creators know they're writing for the public. If it's default closed, then it's writing for friends. The content would be vastly different. Even if each piece of content may not be viewed by as many people, the friction to create content is much less, the result is more people writing and reading content on the whole. There are far more people caring about friends' gossips than the latest research findings. 

At this time, there's still a big problem - you don't know when which of your friends wrote a post or uploaded new photos, so you have to click on their profile pages one by one, which is a bad user experience. As such, Facebook borrowed the RSS idea to create the News Feed. The News Feed was a milestone product for Facebook. At this step, the final form of Facebook was basically established.

From the personal pages all the way in the beginning, to web page hosting, to blogs, to Facebook and Twitter, to Instagram and Douyin/Tiktok, each wave is bigger than the last, richer and richer the content formats and more and more mainstream the users. If we see this development from the lens of Diffusion of Innovations, this diffusion is achieved through different product forms iterating across different eras, and through the iterations, covering different segments of the population. 

The users for personal web pages are the Innovators. The users for web hosting services are between Innovators and Early Adopters. For blogs, even more Early Adopters. And for the subsequent Facebook, Douyin/Tiktok, Weibo, the user penetration rate got higher and higher. The nature of the information may not have changed much, but the form has changed a lot.

As the form changes and usage become more widespread, content has also become more diverse. The diversification of content is closely related to the changes in infrastructure - network speed, data costs, the prevalence of smartphones, pixels on phone cameras, phone CPU, etc.

With better infrastructure and lower barriers to creation, users can access more content, in richer formats, so the content has become more diverse. When we were using Photoshop on PCs, it's hard to image editing photos on mobile phones. When we were editing photos on mobile phones, it's hard to imagine editing videos on mobile. When we were editing videos, it's still hard to imagine beautifying live video feeds. All this is because of the change in infrastructure which allows the broader coverage of user bases.

So what need is it exactly? Today, we all can name a few needs. It's hard to pinpoint exactly what it is. The most important thing to realize is that the prevailing method of realizing needs at any point in time bottlenecks the needs being satisfied to the easiest-to-realize needs of an appropriate group of the population. On the next iteration, the realizable needs achieve a breakthrough, and the serviceable population expands.

The waves of development are accompanied by changes in understanding. For example, blogging was realized around 2000, but technologically, it was feasible since 1995, but the industry practitioners' understanding hasn't iterated to this point. Therefore, we need to understand the key elements in realizing the needs at every point in time and seize the corresponding opportunities. From this perspective, the live streaming vertical is the biggest benefactor of the 3G → 4G → 5G development.

The Iterations of Photo Albums

The desktop era of Facebook already has an album feature. The first generation of albums is paper-based. Then it's the compact discs (CD). Then the albums on PCs. And finally web-hosted photo albums. In fact, the step from PC albums to online albums was not a natural next step, because, on PC, you can already view your photos anytime you like. Putting them online runs the additional risk of the account being hacked. However, one advantage of online albums is that it's much easier to reference a picture when writing an article or sharing it with friends. 

Facebook albums have a tagging people feature. When I first saw it, I was shocked. What a cool feature! The prerequisite for this iteration is the social graph in Facebook.

The next step for photo albums is mobile. Previously, you have to upload the photos somewhere after you've taken them. With mobile phones, photos can be immediately synced to an album or shared with friends. For any site with a social graph, realizing this feature is very easy. This led to a significant reduction in costs to take, upload and share photos, and subsequently, the conversation rate increased significantly and the production of visual content in social networks increased drastically. For the consumers, there is more content to see, so it's also a better experience. 

Physical Constraints, Content, Product Forms

YouTube of the US is great. The YouTube of China is not Youku. Youku is corresponding to Hulu or Netflix. Bilibili is somewhat like YouTube. However, the user base of Bilibili in China today is an order of magnitude smaller than the user base of YouTube in the US. Bilibili has a DAU of 30M, YouTube's DAU is hundreds of millions to a billion. (Note: Bilibili - 61M DAUs, 223M MAUs - 1Q2021; YouTube - 122M DAUs in the US, 2B+ MAU worldwide - 2020.)

Why YouTube is so much bigger than Bilibili?

YouTube was founded in Feb 2005. At that time, much fewer people have video cameras in China than in the US, so Chinese entrepreneurs like Youku (founded in Dec 2003) or Tudou (founded in Feb 2005), tried for a while and discovered that the supply of content was too little, so they had to pivot into something like Netflix or Hulu to provide professional-generated videos.

In fact, YouTube users spend as much time on the service as Douyin or Kuaishou Users, and there is a large variety of content.

Why Douyin and Kuaishou have a different form than YouTube is because when they entered the market, the key elements have changed. Douyin and Kuaishou users use their phones to shoot. Even today, most of the videos on YouTube are shot by digital cameras. The physical constraints lead to the difference in content. Douyin and Kuaishou users shoot with their phones and due to internet speed, bandwidth, etc., they can only do short videos.

YouTube is now a Web and mobile mixed product. And the videos on Douyin and Kuaishou are now getting longer. This is to say, even with different entry points, which match with the state of infrastructure at that point in time, the two product forms may evolve in the same direction. 

This is related to the previous section. When we're building products, we need to consider why a need can be satisfied in the current time window. The reasons are usually because the needs have shifted due to economic and social reasons. Or because of the advancements in technology and infrastructure, a previous infeasible need has become realizable. These two combine to become a breakthrough point for the need. You need to match with the breakthrough point to create a product. In many product decisions, you need to consider why the breakthrough happened and match your decisions with the underlying reasons. 

Find the article insightful? Please subscribe to receive more learnings from Chinese Internet companies and I’d appreciate it if you can leave a comment or help spread the word!

Subscribe now

Subscribe now

Needs in Product vs. Demand in Economics

Needs is an important topic that's difficult to talk about because it's very nuanced. The product manager job was born in the consumer goods industry. Although product managers have to think about users' needs every day, in tech, we don't really have a standard definition for Needs. 

What's worse, the Chinese word for "needs" corresponds to several English vocabularies. As such, we frequently encounter the words "needs", "requirement", "demand" in our work. (Note: Conversely, it’s hard for us to distinguish the three and words like “want”, “desire”.)

In economics, there's a clear definition for demand:

during a given period of time, at various prices, the quantity of a good that consumers are willing and able to buy

Most of the time, when product managers talk about needs, they aren't referring to this definition. They just borrowed the word. It's just so that in Chinese, it's all the same word. Hence it leads to tons of confusion. 

If we were to compare these three English words, it roughly goes like Needs > Requirement > Demand. Needs is the basic motivation of the users. Requirement is the product design considerations based on those needs. And Demand is the quantity demanded by the market once the product is made, and it's usually used in tandem with the quantity supplied. Here, we're talking about Needs. (Note: capitalized for specificity).

Even so, we still find it difficult to explain in one sentence the difference between Needs and the economic definition of demand. Why is there such a big divergence between Needs in product management and demand in economics?

In the last lesson, we've mentioned that to go to a place we could take a plane or a train. In the economic definition of demand, these two products already have established forms. However, in the product discussion about Needs, product managers have to figure out what the user is really trying to get at by offering multiple alternative solutions, and then give him the most optimal one. 

(Note: The form vs. function distinction is important here. I recommend this talk, Position, Position, Position, by Ryan Singer, the former head of strategy of Basecamp, as well as his podcast series, Demand Thinking.

Clayton Christensen’s wording for users “hire” products to make progress in specific circumstances is also great. A lightweight JTBD Framework.)

Recently, air tickets have become much cheaper due to the COVID-19 pandemic. To go to Shanghai from Beijing, it's cheaper and faster to go by plane. If we don't emphasize the objective behind the user's action - to go to Shanghai from Beijing, then we won't be able to offer the appropriate solution. As such, the product that the user buys is a solution to his Needs. Using demand from economics to talk about Needs in product management makes things tricky. 

“The product you’re selling is a utilities combination under a set of boundary conditions“ - Yu Jun, former product leader at Didi, Baidu

If you want to be a product manager, you must be wanting to make a product that a lot of people would use. A product that a lot of people would use must satisfy a lot of people's Needs. In other words, it serves a Big Need. Needs can be big or small. Our understanding and judgment about Needs then become critical. And we will need to try to understand Needs from multiple dimensions. 

The Unsatisfiability of Needs

Rule 1

Out of all human needs, needs that can be satisfied are far less than needs that can't be satisfied. 

Many needs can't be satisfied due to the current state of science and technology. For example, to live forever is a perfectly valid need, but it can't be satisfied. The first time I realized that I was depressed for half a year. Early on, we also realized that humans can't fly. The Wright brothers eventually succeeded to some degree, but many people who tried to fly died. In short, most needs are unsatisfiable. 

If you become a product manager one day, and everyone tells you that you need to satisfy the user's needs, you may feel bad when you can't. However, if you had realized that most needs are unsatisfiable, you would be calm dealing with that guilt. What a powerful realization!

Rule 2

Within the set of needs that can be satisfied under our current state of technology, ROI-positive needs are far less than ROI-negative needs. 

Humans have all sorts of needs. In economics, we have defined many needs with the production value and manufacturing cost unequal. (i.e. public goods, externalities). If you work on B2B products, such needs won't be unfamiliar to you. Clients have odd needs ("requests", "requirements") all the time.

Rule 3

Even for ROI-positive needs, can these needs support the scalable operation of a commercial entity? Needs that can support a scalable commercial operation are far less than needs that can't. 

To satisfy a need, it's not enough that doing so is ROI-positive. The ROI has to be of a certain scale. Because you need to raise capital, and investors have requirements on the rate and the size of return on that capital. (Note: Debt or equity -> WACC, weighted average cost of capital). A scalable need may not have to be able to support a standalone company; it could just be enough to support a department within a company. 

Rule 4

For needs that can support a scalable commercial entity (e.g. a business line), we need to consider whether they can allow a company to survive competitively in the marketplace.

I get asked all the time why Meituan doesn't do X that's quite suitable for our business. What this means is that, for some needs, they are only suited for companies of a certain scale, but you can't create a company serving those needs.

The needs that can support you to create a company are indeed very few. A lot of product managers manage but a small function. This set of needs are not enough to support an independent company.

In conclusion, when we sit down and think about whether there are big, valid business opportunities in our society at any point in time, the default answer should be no. There are so many people, so many entrepreneurial talents in the world, if there are very good opportunities, why hasn't someone done it already?

For any Big Need, at any point in time, the default is there is no opportunity. Having this realization would make you a strong critical thinker, but it would also cause you to miss a lot of opportunities. However, big, valid needs are indeed extremely rare. 

The Time-Variability of Needs

The other question is - are Needs everlasting or changing?

From the standpoint of an individual, needs are perhaps always changing. For example, at this point in time, do you want to have kids? That’s part of the reason why I retired.

Are needs changing for the collective group?

To answer this, we have to find out how needs come about. The mechanism through which needs come about is complicated. Under a specific circumstance, a specific need arises for that specific person. When a myriad of prerequisites coalesces, the needs of people are produced as a result.

As described in Maslow's Hierarchy of Needs, for most people, the desire to satisfy the higher tier needs arises when the lower tier is satisfied first. Conversely, if the lower tier needs are suddenly taken away, the higher tier needs would also collapse. The film, Back to 1942, illustrates this. 

One of the human values is altruism, which is the opposite of egoism. A group of altruistic individuals appears when the wealth level is high in society. But even when everyone was poor, this altruistic gene is expressed in a small group of people. When societal wealth reaches a certain level, this group of individuals becomes larger.

(Related reading: The Selfish Gene)

Therefore, for the whole human species, needs are everlasting. But the quantity is changing, and it’s closely related to the overall societal conditions.

We need to be objective when analyzing needs. Most of the time, when people talk about needs, it's usually centered around "I", and not considering the societal infrastructure, wealth level, education level, cultural zeitgeist, etc. Changes in the macro conditions bring changes to Needs. These macro markers are more related to the broad needs of the population. 

Toshifumi Suzuki (former CEO of 7-Eleven) said the key is to read the world's changes. If we cannot understand the changes in fundamentals, we cannot seize the industry opportunities brought about by them.

Among the schools of thought on history, one is called materialism. The materialistic historical view believes that the emergence of heroes is caused by changes in environmental factors and that the environments shape heroes. For example, without Einstein, others would have come up with the theory of relativity, because people had already observed scientific phenomena which cannot be explained by existing theories. Similarly, we have to go back to the most basic theory to look at Needs.

The needs that we can observe of course are Needs. But the changes in these needs are driven by many underlying factors. The manifested needs may not have changed, but due to the changes in deterministic factors such as new technology, the feasibility of satisfying those needs has changed. There are also some needs that may be amplified by the changes in the social environment, wealth level, and human knowledge, etc.

On a human desire level, needs are already written - the need either exists or it doesn't. But the intensity, prevalence, and feasibility are affected by many factors. 

Only by understanding why a solution is appropriate for this exact need, can we satisfy needs in an effective manner. A product manager has to consider a lot of needs. A reasonably large company has a variety of specially developed tools to manage the needs (“requests”, “requirements”), so the product manager will be buried in the countless needs. Only by grasping the causes of major and critical needs can we know what the higher ROI needs are. Therefore, we must always evaluate the prevalence, the intensity of the needs, and the ROI to realize the needs.

The window of opportunity for the great needs is like lightning in the sky. It is the moment when the positively and negatively charged clusters penetrate the air. At least one of the charged clusters needs to be in motion - either the need itself or the feasibility to satisfy it.

A/B Sides of the Internet and Needs

In the last class, we talked about the A/B sides of the Internet. Now let’s talk about why the division, why Meituan emerged at that point in time, and why Wang Xing and I chose this direction. 

A very important reason was that most of these B2 companies appeared after 2010. Before, it was very difficult to obtain the user's location. After the rapid development of the mobile Internet at that time and everyone had a smartphone, the location data in the smartphones made it possible to realize many things that could not before.

In the early days of food delivery, Daojia could only serve customers with an AOV of ¥80. It was because there was no smartphone that could locate the delivery person in his delivery route, so when the delivery person was dispatched, he could deliver a single order. Without knowing the real-time location of the delivery person and the direction of his future movements, which are data needed for a dispatch system to allocate orders, the delivery efficiency and costs couldn't be improved much.

A critical function at Meituan is the dispatch team. The main job of this dispatch team is to make good use of the real-time location data of the delivery riders and put the infrastructure to work. A delivery rider empowered with real-time dispatch has a far greater delivery capability than one without.

When I went to Europe for business exchanges, I learned that because the population of Europe is sparser, the delivery riders can only deliver 8 to 10 orders a day. A delivery rider of Meituan can deliver 20 to 30 orders a day due to the central dispatch system. Real-time location data also play a part here.

Meituan has a bike-sharing business, and we also started e-bikes this year. At present, e-bikes seem to be doing pretty well. As such, Meituan internally has been discussing whether they should expand e-bikes overseas. 

After studying the market, they discovered the costs of e-bikes in China are very low. The supply chains of other countries do not have such a low-cost structure. Furthermore, the penetration rates of mobile payments in other countries are relatively low.

E-bikes depend heavily on mobile payments, so e-bikes can only be realized after the mobile payments infrastructure become highly developed. Listing down the mobile payment penetration rates of the major countries in the world, you’d find that China is at over 30%, while the others are all at 2% - 3%. Therefore, it’d be tough to do e-bikes in other countries.

Now, they are discussing whether they can use e-bikes as a way to build mobile payments. So the idea is totally flipped, and the strategic decision has changed - it has become whether to expand mobile payments overseas.

This demonstrates that infrastructure is very important. The reason why many things become feasible is that there have been breakthroughs in infrastructure, and infrastructure has reduced the cost of implementing this feature [to satisfy this need].

The Heart of A Child

We humans will learn a lot in the process of growing up, but we will also realize that many things are unachievable, and gradually we start to manage our expectations. This leads to even if something is possible to achieve, we may not be able to realize it and still thinks it's impossible. 

Therefore, there are two types of needs. One is that the needs have always been strong and prevalent but it is difficult to realize them. The other is that although the needs have always existed, their intensity and prevalence have become greater in recent times.

To have ”the heart of a child“ (赤子之心) is the first type of Needs. Examples include immortality, flying, and every wish coming true. 

The mission of Meituan Delivery is to "put the world in your hands". The ideas that really move people's hearts are some of the most prevalent needs. It's just that they are managed as we grow up. So the first type of needs are the greatest needs, and they gradually become achievable.

However, breakthroughs are hard, so it is very important to maintain a heart of a child. Otherwise, you will let yourself give up thinking about these ideas and pushing them forward.

(Note: Recent Virgin Galactic and Blue Origin’s successful launches are very good examples of this type of Needs.)

Innovator’s Dilemma - Why now?

There is another question here - why this need is achievable now? 

If you make a product one day, you will face a dilemma. When something has been realized, everyone will think that's how it should be. But before it's realized, there would always be people questioning why would anybody use this thing. 

For example, when I was working on Xiaonei, when the site was already up and there were hundreds of users, a friend came over and questioned why would anybody use this thing. He also spewed out plausible-sounding theories. Social networks are popular in the United States but they won't work in China because Americans and Chinese are very different. Chinese women are very conservative. American women are very open. It won’t do for Chinese women to upload photos on social networking sites and be commented on by others.

At that time, Wang Xing and I didn't have a strong grasp of human nature. We just thought the Six Degrees of Separation theory was cool, and that's why we made the site. Anyway, when you talk about your idea with anyone, you can always find all kinds of theories that oppose it.

All great needs have two very important characteristics: 1. "great idea, but not possible" and 2. "no one wants that".

We've mentioned earlier in this course that Thomas J. Watson allegedly famously denied the public's need for computers - "I think there is a world market for maybe five computers". I'm 42 years old this year. But I still dream several times a year that I fly over the campus of Tsinghua University, flapping my wings, and seeing the young students running in the field. And I'm acrophobic in real life! So great needs will always face these two questions at the beginning. 

If you meet someone who is more sensible, he will ask you the third question - why now? why no one has tried it or succeeded before? 

As mentioned before, all great needs must have been tried in the wrong way at the wrong time by the predecessors. If not, then it’s not a great need.

A recent great need that's being attempted is self-driving cars. Of course, I don’t know if it'll succeed this time around. According to reports, one of Larry Page’s research ideas, when he arrived on Standford's computer science doctorate program in 1995, was self-driving cars. From this perspective, this great need has been in existence for at least more than 20 years.

Find the article insightful? Please subscribe to receive more learnings from Chinese Internet companies and I’d appreciate it if you can leave a comment or help spread the word!

Subscribe now

I’m recruiting some users for a private beta test for my product. In short, I’m building a product that curates the highest signal-to-noise ratio tech content from the web and offers a distraction-free reading experience. It’s for founders, investors, product managers, and operators. If it sounds interesting to you, you can sign up here. It’s free. I’ll let you know once it’s ready.

Sign Up for Private Beta

On with the content. Wang Huiwen is gonna talk about Needs next. For Chinese product managers, they do it somewhat differently than in the West when we talk about solving people’s problems or making something people want. They think very deeply about it, almost on a philosophical level. That’s why I decided to publish this Product Management Roadmap first so you can see where Needs come in in the big picture.

Product Management Roadmap

This roadmap is made by two Chinese product managers - Liu Fei (刘飞, WeChat Official Account: liufeinotes 刘言飞语), who wrote a best-selling book on product management and used to work at Didi; and Shao Nan (少楠, WeChat Official Account: southennanmu 产品沉思录), who writes great content and works at DXY.cn (a content & online community for healthcare).

Below, I’ve translated some of their discussion on this roadmap on their podcast.

I’d say it’s heavily influenced by the school of thought of Yu Jun (product leader at Didi, ex-VP of Product at Baidu; top 2 product guys in China, the other one being Allen Zhang of WeChat). I’ll cover more of him in the future.

(If you want a high-def PDF version of this roadmap, share this newsletter on Twitter/Linkedin/Facebook or with your friends/colleagues via Slack, etc, and send a screenshot of the share to chinaplaybook@substack.com)

Understanding the Users

Needs Analysis

Value Analysis & Product Design

Product Iteration

Interdisciplinary Knowledge

Legends

Podcast Discussion Transcript

Overview

Liu Fei:

First of all, let me briefly summarize the product thinking roadmap. It has a few modules. The first big module is Understanding the Users, it's strongly connected to Needs Analysis - your analysis has to be based on real users' needs.

The Product Design module follows because you design your product based on those needs. In between Product Design and Needs Analysis, there's a module called Value Analysis.

Similarly, Value Analysis is strongly tied to Needs Analysis. In short, needs are the problems that you unearthed from the users and values are the solutions from a product point of view to see if you can solve those problems for your users.

And Product Design is, okay, now I have some clear ideas on how to solve those problems from them, how do I put them on paper or screen? How would the ideas look like in a product, like an app for example?

After we've designed a solution, the next part is Product Iteration.

What I understand as the workflow of a product manager is roughly these few parts.

At the bottom, we have a few complementary appendices.

The first one is Interdisciplinary Knowledge. When I first got into the industry, and I guess you probably experienced the same, everyone didn't have much concept on interdisciplinary knowledge. But over the years, you started to hear more and more people talking about that regardless of which discipline you came from, you should be learning from other disciplines, especially from economics, sociology, psychology, etc. Right?

As for the other two appendices, one is the PM Career Development, which is quite simple and should be self-explanatory; the other is the Core Abilities of a PM, which is closely related to career development.

So, this is the overview of this roadmap. At first glance, are there any differences from your product system?

Shao Nan:

Well, first of all, here's what I find interesting about this map. I see this map as two halves. I used to buy a paper map whenever I go to a new city. For those maps, the actual map is at the center of the page, then you have sites of attraction at the periphery - which bus to take to get there, what are the local delicacies, etc. 

So, the first half is the essential stuff, the core workflow. The second half is the attractions that make your journey more enjoyable, and info to make sure that you don't get lost. That's my first reaction. 

Back to the product roadmap, for the Understanding the Users part, you and I are highly consistent, no difference there. However, I missed a whole module, which is the Value Analysis. I have not much concept. I probably have done some work in this area before, but I didn't have a formula or have abstracted it as a knowledge module. I'll have to think about it seriously. So that's a knowledge gap on my part.

On the other hand, for Product Design, I'll probably add something more to it, which is Service Design. 

Yeah, because at that time, I found that, hey, a lot of experiences are not done on the phone. They’re separated from the pure app experience. For example, you have to negotiate with offline merchants, and you'd have things like third-party transactions. So I'd have to put a lot of thought into Service Design. 

Liu Fei:

Right, ok, so the presentation is not what we traditionally think of as the product. 

Shao Nan:

Exactly. It evolves with time, right? In the early years, the pure app experience might be enough. Looking forward, our knowledge now may become anachronistic in the future. That's why I'm like, how fun, same map, but what people see are different. 

How the map came about

Liu Fei:

I'll talk about how my product thinking framework came about.

First of all, this map is not terminal and it's not a standard. Not everyone in the industry needs to agree with this or reference this. It's just my personal understanding. From a space perspective, this map is just one man's viewpoint. From a time perspective, this is just a summary at this juncture. I may add, delete, modify in the future. With that out of the way, I can talk about how I gradually came to this model.

When I first got into the industry, my first job is at Smartisan (a smartphone manufacturer). At Smartisan, I think I had some level of knowledge and understanding of the user. But not as complete compared to now. 

The knowledge and understanding were more restricted to the user scenario or the very detailed minor points about a user, not the complete profile or mentality of a user.

It was like, I have a feature, I think about whether the users will like the feature or under what scenario the users would use the feature. 

So this is a rather microscopic understanding. Yeah, so on the micro understanding, I was on the right track. But I severely lacked the macro understanding. 

Of course, that was also due to the circumstances we were under. We didn't have the resources to conduct a lot of user research. So we didn't have the macro kind of understanding like who are the users on our platform, how do they live, what are their mental states, what's the distribution like across different demographic segments, would this group of users like this feature, etc. 

We didn't do any sort of analysis, but maybe the deeper understanding of the users can come from actual practice in the Product Iteration module. 

So my understanding at that time lacked this chunk. My core understanding was just starting from micro and design bottoms-up. Indeed, it was also starting with the users, designing the feature based on the user scenario. But the path in the middle is very simplistic - you have a scenario, the users need it, make the feature. Yeah, that was my product framework, it's so basic, a very short path. 

Shao Nan:

We all started like that. At least, you had some concept of the user. When I started, we were building iOS apps, around 2010. Only at that time, I started to have this notion that oh, I'm making a product. 

The methodology was just "the light bulb moment". Let's make a thermometer. Let's make a date countdown feature. Let's make a calendar... It's all starting from the self, giving zero care to who the users would be.

Just now you mentioned micro understanding. It reminded me of a user research case that left a deep impression on me. When I went to Baixing (a classified ads site), I think we all had some concept of the user. But all the methodologies that we used at that time were borrowed from advertising companies. 

Liu Fei:

How so?

Shao Nan:

For example, like methods used by Unilever, the FMCG industry, such as focus group.Of course, they have many methodologies, but they're not suited for the tempo of Internet companies or the customer groups they serve. 

For example, the users of Baixing would be the relatively lower-income groups. When you invite them to the building for customer interviews, you couldn't get really get anything out of them. They were dumbstruck with the fancy offices. Ask me anything, whatever you want, I'll tell ya. 

Liu Fei:

Got it. 

Shao Nan:

Yeah, so I remember clearly this uncle, we needed him to pay for something. To pay, you need to install a plug-in on IE (Internet Explorer). The uncle says, “okay, I can pay.”

We were standing behind him like, why didn't he click the button, why didn't he click the button. And the uncle started to sweat because of all the stress. He wiped his forehead and said, “don't worry, I can pay.”

After struggling for 10 minutes, he finally gave up - “sorry, I can't pay.”

Liu Fei:

He treated it as a task. I have to complete it. It's completely different from his normal living state.

Shao Nan:

Right. We've talked about Interdisciplinary Knowledge. When the Internet industry was just starting, we vaguely know that we should be doing user research, but we could only borrow experience from other industries since we've never done it before. We borrowed focus group. We borrowed eye-tracking. Wow, eye-tracking right, so advanced. But it's useless, haha. 

So I feel quite deeply on this understanding the users part. As compared to before, our methodologies really evolved a lot. 

……

Shao Nan:

Then what happened for you?

Liu Fei:

As mentioned, my product thinking was a very short path at my first job - I discovered a scenario or need, then boom, we have this feature. 

Since that time, and actually part of the reason why I left, I started to have some sense about Internet products. Whether it's the experience at Smartisan, or conversations with friends who work in product, or even observations from what people shared online, I started to feel, hmmm, people build products differently, and I wanted to experience other ways of building products.

That's why I went to a more operations-driven industry - on-demand nail services - and started up. Afterward, I worked in food delivery and ride-hailing, and the understanding deepened more and more. 

However, at the time when we were starting up in nail services, I reflected retrospectively, I was only more aware of the early parts of the map - users and needs. But even for users and needs, I didn't analyze very well and I had some misconceptions. 

For example, the understanding of the users at the time. My understanding of the users are just the customers of nail services. I didn't know much about the nail technicians.

I thought they were not the users. They were not under the purview of a product manager. You know, we pay them and let them go to the place to serve. Really, no kidding, my understanding was really that shallow. 

Of course, if you think this way, you wouldn't think too much about the needs of the two parties on a marketplace, the match between supply and demand... and consequently, your understanding of user experience would also be very shallow. 

My understanding of user experience is just like those early UX-type product managers - whether the app looks nice, has cool visuals, how are the interactions, etc. 

Of course, on-demand nail service is the experience of the full process. Your product in fact is the entire service from start to finish, not the app. If it's the entire service, then some key issues with regard to services, such as punctuality, the quality of the nail services, the service attitude of the technicians, etc. become core product points. 

If you only care about the app, then you only care about the pre-order experience. But in fact, the pre-order experience is relatively unimportant. What matters more is the after-order experience - how you dispatch orders, how to improve nail technician's service quality, how to manage them etc. 

Of course, I only learned this later on. When I was starting up, my understanding was really simplistic and shallow. 

Shao Nan:

That realization came about pretty much around the same period for both of us. 

Here I want to interject an interesting story. In the early days of Airbnb, they have something called Project Snow White, I've specifically studied it.

At that time, the two founders felt it was very difficult to make any progress because booking an Airbnb is such a complicated user flow. Say I'm going to rent my house to you, where should I put the keys, right? What should you say when you enter the house? What shouldn't you say? It's so complicated. Users don't know what to do. 

Of course, we all started at the same line. The founders are also designers, and they had never done a service business before. 

Later, the founders went to Disneyland and they found that, wow, the experience is so nice here. That's when they realized that a lot of experiences are not done within the "product". They're done outside the "product". 

Once they realized that, they started to think well, how does Disney develop their cartoons? 

If you've studied animation or film, you know they have something called the storyboard. That's how they outline a story. For example, today this person goes out the door, turns right to unlock his bicycle, rides the bicycle forward, etc. You have a camera language. 

So after they came back from Disneyland, the Airbnb founders hired a Pixar illustrator to draw out the personas of a few typical users. For example, a poor student, an old couple, a rich person with 3-4 houses, etc. Draw these people out on paper, these are the User Personas. That's when I first learned of this thing called user personas. 

The second thing you need to do is, you have to tell a story. For example, what's the poor student's story? Right, I'm a poor student, but I want to travel to Europe, and I don't have any money. Maybe I can rent out my house and let it earn money for me, while I travel in Europe?

So this is the student's story. The illustrator then started to draw, maybe 15 - 20 frames. With the frames, the scenarios become very clear. For example, when I meet the host for the first time, should I bring a gift for courtesy and rapport, right?

When the host writes the introduction for the house, since the guest is here to experience local life, he should probably tell the guest what's good to eat around here, what fun places, which area in Brooklyn that you shouldn't go to, etc. 

This kind of thing you can never think of looking at a screen. Is this how you felt when you came to that realization?

Liu Fei:

Yes, it's very similar to my journey. Actually, the first phase was, I found that the product or the app didn't seem to matter a lot. I was feeling rather depressed. Like, wow, product managers don't matter in this business. 

The second phase was when I started to realize, oh, this bigger thing is the product. Product managers don't just need to work on the app. If you have thought through how to solve the various problems through the product, then you've graduated to become a mature product manager. So the second phase was the enlightenment.

……

Liu Fei:

Yeah, you know, we all know these words at the time. Users, needs, scenarios, values, we know them. But in fact, our understanding of these concepts, and the methodologies that are based on them, we can only say we had a very superficial understanding then, or our understanding was problematic. It wasn't until later as we did more work, then we started to have a clearer understanding of these concepts. It's a cognitive process. So I'm thinking, for some junior PMs or newbies, they look at this map, and they'd think, it's not that difficult, I know this, I've heard of these things. 

To really understand it, you have to be making products and participating in the operations to have a visceral appreciation. 

The Answer is in the Field

Shao Nan:

Right. Actually, I have a training method. I was trained this way as well. When I first went to Baixing, one of the things that we needed to do was to conduct user research via phone interviews. That's no methodology, and no one to teach you. You just need to talk to some people today. What do they not like about our website? 

You know, the first time I picked the phone, my hands were shaking. Like, what do I even say to this person? You don't know whether it's a man or a woman, old or young, you don't know anything. It's just a number.

But anyway, it's an excellent way to train how to understand users. 

Here's my view, some people may not agree. Some people say, okay, I know how to conduct user research. But they've only ever sent out surveys or looked at usage data, or even observed people from afar. I'd say it's not on the same level as talking to the user 1-on-1, trying to understand his motivations, his concerns, etc. It's like kung fu, you have to put in the work. 

Liu Fei:

I always tell the younger PMs around me, the best way is you have to, have to be in the field, or you interact with the users directly. For example, if you're building a content product, the users just consume content. Of course, there's no point going to their houses. But you have to talk to them, observe how they consume content. This is the field. It doesn't mean you have to stand beside them. 

This I have experienced deeply, especially after I worked on nail services, the times that I go on-site significantly increased.

When I was doing food delivery, my company started to have rules that you have to deliver food yourself. X orders a month. It wasn't that stringent. At Didi, it was even more stringent. We have to register as drivers and we have to drive X orders or have to earn $X amount. Of course, this is just for the driver-side PMs. For the passengers-side, you don't have this requirement.

For the driver-side, it's a strict requirement. If you don't know how to drive, you sit in the car. You have to be in the car. Especially in the later phase of the company, Mr. Yu Jun asked our driver-side PMs to have an entire day to drive on the road every month or every quarter. Why? Because previously it still wasn't authentic enough. 

Some of us just drive for one order, two orders, during lunch break, whatever. You'd never experience the mental and physical states of the drivers under the real work intensity. 

Under the driver's real work intensity, for example, after you drove for 10 orders a day, how's your energy level, your mental sharpness when you receive orders, and what's the chance of getting into accidents when you drove too many rides. Or if you're dispatched to somewhere really far, when this situation happens, you'd start to worry about how to take care of your next few orders of the day, etc.

When you actually care about your income and your number of orders for the day, that's the most authentic experience. That's when you can really discover a lot of problems that the drivers face.

Echoing what you said earlier, only after you spend enough time in the field, to actually experience the lives of the users, you can train your understanding of the users, and the understanding of the scenarios. 

Previously I thought going to the field is not that different from having a phone interview. You get some users in the office, you ask them questions. It's not that different right? Actually no. I discovered that in the field, the amount of information you can gather is much larger. 

For example, food delivery. The delivery rider gets the meal from the restaurant, goes on with the ride, arrives at the consumer's door, delivers the meal, then leaves. Isn't that a simple process? You go to the field, you also see him performing the same process. Isn't it?

No. When you're in the field, you’d find that, oh, for example, to get the meal, the restaurant may be crowded, and your delivery rider is squeezed there along with everybody else. There are other delivery riders trying to fight with him to get their meals first. 

Or queuing, which happens very often. Or the restaurant prepares the meal too slowly. Or you have some disputes - his order is that, my order is this, who gets this dish? Or you have a poor internet connection, the order doesn't come out in the dashboard, how does the delivery rider communicate with the restaurant owner? Or the delivery rider found that he can't get this meal from this restaurant in time, but he's rushing to another place, how is he supposed to handle that?

When you're in the field, you'll encounter all sorts of problems and the amount of information is overwhelmingly big.

So I think, for me, only through experiences like this, I started to have a good sense of how to solve the users' problems. It started when I was doing nail services, but at that time, I only discovered the problems, I didn't know how to solve them. Only at Didi, this product sense was groomed well. 

When I first went to Didi, I set a goal for myself that I'd spend an entire quarter to understand the drivers. I had become intentional in understanding the users, using the effective methodologies that I know of. 

Shao Nan:

Yeah, I also felt that deeply. You mentioned food delivery, I also worked on food delivery. Some very simple things, like for the restaurant, when you have five riders coming in, you have to make sure they don't take the wrong order. Or you have to make sure you have a pair of chopsticks inside the package. If you forgot the chopsticks, for sure, it's a bad review. 

Another point is that some people ask me why I don't believe in surveys. Because you'd have deviations. Then they say, if I have enough quantity, would that be okay? I think not. You can discover a lot more information being in the field. 

A lot of people use surveys in the wrong way. I want to expand on this point. 

They say surveys are good. You can have a large number of responses, and they can inform on some user behaviors. You go to the field, you can only see 3 or 5 people. 

Well, on-site research is qualitative. I've seen people do on-site research for quantitative reasons, like doing 500 in-person interviews. You've completely missed the point. 

If you go to the site, as we talked about, you can find out a lot of unexpected things, then you can think through those unexpected things and come up with solutions. Like poor internet connection. What's the probability of that? You can't be always offline right? Only in certain spots, you have a poor connection. So how do you deal with that? Or queuing at peak hours. Or your rider delivered the wrong order. Or your rider's vehicle broke down. Etc. These things are when you see the signs, you can see the patterns. 

Of course, you don't just need to go to the field, you also need to synthesize diligently.

It's hard work going to the field. However, if you do this for 5 years, regardless of what industry you're in, or whether you changed job, you're just different from people who spent one year on-site. 

I used to lead a junior PM. When he just came, I asked him to just do user research. He's been doing that for 3 years now. Now when I ask to him to conduct a user search, he said he'd just interview three users, and he'd tell me the insights that he gathered. When he first started three years ago, he'd call 20-30 people a day.

Three years back, his summary was superficial, like it's a female, 34 years old, why she uses DXY.cn (a content & online community site for healthcare).

Now he'd say, this is a mother living in a first-tier city. She's conflicted inside because her parenting ideology is very different from that of her parents. She craves recognition from people who can understand her position. But at the same time, she doesn't have that much money to hire a private physician or go to high-end clinics. What she wants is identity. 

You see, it's the same person, just three years apart, the kind of user insight that he can get out are completely different. 

Liu Fei: 

It's about knowing the purpose behind these methodologies and how and when to use them. Like on-site that we've talked about, is to discover the problems and gather qualitative information. You know what questions to ask. Only when you know what questions to ask, and your questions are specific enough, can you qualitatively define the problem and the information that you gathered is valuable. 

From User to Users

Shao Nan:

Right. Let's move on to the next question. You see our recollections were pretty similar, we both started not paying attention to the people. We initially just focused on the small behaviors, like what color of the button, and then shifted to start focusing on the person - the whole person or a few types of person. 

When did you start to realize that this is still the individual(s) and move on to the group or the industry? Like for ride-hailing, you have the drivers, the car rental companies, the municipal governments, you know, interested parties. When did you realize that?

Liu Fei:

Right. As I mentioned, I realized that the product is this bigger thing when I was working on nail services. When I moved on to food delivery, I started to deliberately train myself in this area. 

As for from individual user to the collective group, it should be when I was working on food delivery. That's when I realized I need to have a macro understanding. You need to have some judgments about what the industry is and where it is going. Who are the users within this industry? What's the logic that this industry operates on?

I think Mr. Yu Jun said it the best. It's just the matching between supply and demand. All the products that we use, are in essence transactions. Of course, that's a higher level of abstraction which I learned later. 

When I was doing food delivery, I know that let's say you're doing logistics fulfillment, your upstream is the delivery platform, and your downstream is your riders. You deal with the merchants and the riders. Between these three parties, what role do you play?

What are the needs of the delivery riders? Why would they be willing to come onto your platform? What are the needs of the merchants? Why would they come onto the platform? Etc. So at that time, I started to have these concepts.

At Didi, of course, my understanding got even deeper. We'd think from the macro perspective. Like you mentioned, the drivers are one part of the supply. Supply, first of all, they face the consumers. There's also the participation of the platform in that process. 

How do you get so much supply as a platform? You need to have a lot of people. Where do people come from? People, I'd give you a few examples. It could be the rental companies, or the small collectives, or even just a driver that knows a lot of other drivers.

The other question is the cars. Where do you get the cars? You need to partner up with a lot of car companies. What's the ecosystem around that?

Later on, you'd face government regulation. You have to deal with compliance. Or even where your car plates come from. Etc. 

Slowly, you learn the underlying logic of the industry. From this underlying logic, you can further analyze a few layers. The logic for driver-passenger matching. The logic for orders dispatch. The logic for your service. Etc. From each logic, you can further discover a lot of problems. 

So gradually, you start to have a map of the entire industry in your head. And you can locate where your part fits in on the map. 

This is drastically different from when we started - I observed some minor user behavior, boom, I made this feature to fix that. We have a much longer path, and the solution would also be more complicated. 

Interdisciplinary Knowledge

Liu Fei:

The central problem is that you won't be able to learn everything. Good product managers have a strong problem-solving mindset. What problem am I solving right now? What subjects do I need to know to solve this problem?

Previously at Didi, I was a platform-side PM, so I needed to learn a lot about economics, algorithms, and demand and supply matching. The demand and supply matching impacts the business, the ecosystem, and the platform. To balance it, we need to use some algorithms and implementation methods.

If I was working on service strategy, like my peers dealing with driver and passenger disputes every day, I had to study law. Every day they study law. They need to know how to make rules, how to enforce them, and even consider what would be the potential public opinion, what's justice, etc.

That's an important task, designing a feature when you need to consider public opinion. You can't just say it feels fair and reasonable to me, or to us as a platform. Because that's would be a black box. You have to consider how drivers and passengers would take it. You can't just be I'm doing right by my own values anymore. No. 

(Only these points are translated due to the length limit. For me, studying this map alone is like a 10X product learning experience. I highly recommend keeping it somewhere close by and revisit regularly.)

👋 Hi! I’m Tao. As I learn about building products & startups, I collected some of the best content on these topics shared by successful Chinese entrepreneurs. I translate and share them in this newsletter. If you like more of this, please subscribe and help spread the word!

Subscribe now

This article is based on this lesson and a talk he did previously.

Design is categorizing

The product leader of WeChat, Allen Zhang, said that design is categorizing. To put things into their rightful place is exactly designing. In our daily life and work, we tend to look at things specifically - the task at hand, the industry, industry characteristics, product features. Rarely do we consider the big picture to look at the categorization of an industry, even though it's a very meaningful exercise. 

If we were to categorize the entire Chinese Internet industry with just one cut, how should we do it?

There are many angles from which we look at the Internet industry. One of which is to divide by technological sophistication. We dichotomize technologies into A) "traditional" - those that have been entrenched in our lives and we take for granted, and B) new technologies such as AI and big data. Another angle is by what industry Internet integrates into, such as education, F&B, retail industries. (i.e. "EduTech", "FoodTech", "Retail Tech"). The angle that we choose result in meaningful differences.

The angle that I'm going to talk about is leaning towards categorizing by the “traditional” Internet characteristics.

A/B Sides

If we were to categorize with just one cut, we can divide these companies into two categories - one supplies and fulfills online and the other supplies and fulfills offline. Let's call them A-side and B-side. Notice the difference in their areas.

A-side Internet companies are video sites, live streaming, online games, etc. B-side companies are Alibaba, JD, etc. Most of Meituan-Dianping's businesses also belong to the B-side. 

The area on top is relatively smaller because the area corresponds to the entire industry's GDP. For example, Alibaba and JD are in the retail industry. The retail sector is a very big contributor to China's GDP. (Note: The wholesale & retailer sector accounts for ~10% of China's GDP, while Information & Communication, Software, and IT Services accounts for 3%, and Culture, Sports & Entertainment accounts for 0.7%. 2016 data.)

There's another angle we can talk about here. Although the industry of video, live streaming, etc. is not as big as the retail industry, it receives a greater degree of influence from the Internet. Things like online games are “pure Internet”. But for retail or e-commerce like Taobao, since the fulfillment and delivery are conducted offline, the Internet has a less degree of influence on this industry. 

This degree of influence will affect their gross margins. The area below is much bigger, but the companies at the B-side don't necessarily enjoy a high margin. Generally speaking, if you look at companies' financial reports, you can see the A-side companies usually have high gross margins, and B-side companies usually have low margins. These are some of the ideas that we can get from just one cut. 

B1 and B2

If we were to cut one more time and only once more, we can build upon the previous basis.

This may be what it looks like: A-side remains unchanged, but we further divide the B-side into B1 and B2. B1 is SKU-based supplies. B2 is location-based services (LBS).

Why can these two be divided? 

Their information architectures, product interaction processes, and business operations will be very different. But where do these differences mainly manifest?

For A-side, where supply and fulfillment are conducted online, the core competencies of the company lie in product design, understanding the users, and managing communication, socialisation and content. These capabilities are common in A-side companies. 

For B1, the company's core competencies lie in understanding SKUs, understanding the supply chain, and understanding pricing. (Note: Genki Forest is actually a good example.)

For B2, if you were to inventory B2 companies, you'd notice a remarkable commonality among them - a large offline team. Having a large offline team is one of the biggest differences between B1 and B2 companies. 

Differences Between B1 and B2

Scale Effects

In SKUs retailing, suppliers enjoy scale effects. Most of the suppliers for Taobao and JD have scaled production. However, in services, the suppliers such as delivery riders and Didi drivers don't enjoy scale effects. SKUs can be manufactured, but people have anti-scale effects. (Note: Wang Huiwen sees Network Effects as a subset of Scale Effects.)

When facing large order volumes, the merchants on Taobao benefit from diminishing marginal costs of production and fast inventory turnover.

Large order volumes for services will drive up the marginal costs. Additionally, there is a high requirement for timeliness for services, consumers can't save up their consumption to make it convenient for producers.

As such, this led to a fundamental change in business logic. The business logic for service providers is to peak shaving and valley filling.

Merchant Recruitment

The merchant recruitment is also different. Taobao recruits merchants nationally. As such, there are plenty of alternatives for a category (or SKU) and they don't have to spend too much effort on merchant recruitment. For Meituan, however, the merchant that we recruited in a city can only serve the consumers in that city. Therefore, we have to do "ground push" in every single place to recruit merchants, which requires the commensurate organizational capabilities.

As a result, Taobao recruits merchants in Hangzhou (purely online), whereas Meituan's merchant recruitment teams are all over the country. 

Infrastructures

B1 and B2 also require different infrastructures. The infrastructure for Taobao is logistics and payments, whereas for location-based services (LBS) - food delivery, bike-sharing, ride-hailing, restaurant reviews, etc. - the core infrastructure is maps and location-tracking devices. 

Consequently, the importance of the location-tracking feature in the app - how fast, how accurate, how precise the app measures user's location - is an defining attribute of B2. Taobao is a great product, but for Taobao, the user's location is not that important, so the app may not necessarily report location with speed or high accuracy and precision. This [feature prioritization] has to do with core competencies. 

Looking at A, B1, B2 as a whole

Developmental Perspective

Chronologically, A-side is where the first Internet products - email clients, browsers, and web portals - appeared. They lean to the A-side judging by their attributes and came out first to make the Internet itself more useful and usable. B1 came after, like e-commerce websites. B2 companies came about around 2010. Based on the timeline, we can see the three waves.

Why B2 began to emerge since 2010?

One of the reasons is that location is very important. Location is comprised of two parts - the first is the user's location, and the second is the supplier's location. Whether it's the user's location or the supplier's location, we need to get the location fast and conveniently.

The prerequisite for that is the prevalence of smartphones. Without smartphones, getting the location of users or merchants would be extremely difficult. Therefore, location-based services came about together with the proliferation of smartphones. Smartphones opened up this space. Before smartphones, there were companies operating location-based services, but they had a hard time operating. 

Typical Companies and Products

Let's take a look at the typical companies and products in the three categories. It'll look something like the below graph.

When we talk about a company, we often talk about the company's DNA or core competency. A company's DNA or core competency is determined by the category that it operates in.

We've talked about the different characteristics of A, B1, and B2. The different characteristics shape the corresponding companies' core competencies. They also to a certain degree shape the relationship between different companies as well as determining which industries a company can operate in or not.

From this graph, we can see a few things. Overall, every company here has a main coordinate, but we can also see some companies may be involved in another category. Generally speaking, the success rate of overstepping is not high.

For example, Tencent. As one of our investors, Tencent is our good friend. As an A-side company, they tried to enter into B1 with Yixun (B2C e-commerce). However, they realized that them operating in B1 was not going to be able to compete with Liu Qiangdong (founder and CEO of JD), so they sold Yixun to JD.

After that attempt, Tencent changed its attitude to solely focus on operating A-side businesses and participate in non-A-side businesses in the form of investments. Tencent invested in us Meituan, Didi, Ele.me, Lianjia and they couldn't get in Gaode Map (Amap, or AutoNavi). Of course, they also invested a lot in B1.

For Tencent, B1 and B2 require different competencies that it doesn't have, so they changed their strategy to investment.

Similarly for Alibaba, they want to be in the A-side. For example, Alibaba's Digital Media and Entertainment Group acquired Youku Tudou (video), Xiaomi (music) and also invested in films and television. Overall, the results are not too good, because A-side requires different core competencies than the B1 core competencies that Alibaba is familiar with.

In terms of the differences in core competencies, the distance between B1 and B2 is relatively closer, and the distance between B and A is relatively further.

Xiaomi

The more observant people may think that my chart is not accurate. For example, I put Xiaomi in B1. I did spend some time mulling over Xiaomi and I think it's accurate to put Xiaomi in B1.

Between B1 and B2, Xiaomi's core competencies and business characteristics are more evident in B1. B1 tests a company's understanding of the supply chain. Xiaomi really has a very deep understanding of the supply chain now. This categorization is by the different companies' main business.

Meituan’s Competition

On a related note, I want to mention something that's frequently brought up. People usually have doubts and opinions about Meituan Dianping. They think we have relatively more enemies and we often get into fights.

In fact, if you look at this graph, do you find some patterns?

From when we started in 2010 till now, we have many competitors, especially in 2015. In 2015, we have four businesses:

1. Group Buy - the competitors are Nuomi (acq. by Baidu) and Dianping (pre-merger)

2. Movie Tickets - Nuomi Movie (Baidu), Weying (Tencent), Tao Piao Piao (Alibaba)

3. Food Delivery - Ele.me (Alibaba), Kuobei Waimai (Alibaba)

4. OTA - Ctrip, Qunar

You may notice all our opponents are concentrated in B2. In fact, the B2 category is the most fought over battleground in China's Internet industry since 2012. The bike-sharing war, the ride-hailing war, and the food delivery war were all in B2. 

Why is it so fiercely fought over?

Because the B2 category only just started in 2012, and new opportunities just emerged. Everyone has a similar level of cognition of the opportunities, a similar level of industry understanding, and similar core competencies. The scenarios that we see are also similar. The similarity in core competencies results in us clashing in scenarios, it's unavoidable. (Note: “Scenario” is a term I’ll cover more on.)

ByteDance

While we’re on the topic of competition, today, many think that ByteDance is a belligerent company. If you look at it closely, is any of ByteDance's oppponent in B1 or B2?

ByteDance works on information stream and short videos.

Who's its competitor for short videos? Kuaishou. ByteDance's Weitoutiao's (Note: "Mini Headlines", allows content creators on Toutiao to engage with fans.) competitor is Weibo. ByteDance's Q&A product, Wukong Wenda, is competing with Zhihu.

ByteDance mainly operates in A-side, although it still has many competitors, they concentrate the fights in A where they have core competencies.

The reason why ByteDance faces so much competition is that the categories that they operate in have also been undergoing rapid changes in the last few years. New opportunities emerged and hence the intense competition.

4 Scenarios of LBS (B2)

If we take each of the three categories as a whole, we may think A-side is what we call "pure Internet", B1 is what we used to refer to as e-commerce. If we were to cut B2 one more time, how would it look like?

Since it's location-based, there are the consumer's location and the supplier's location. Whether the location moves then becomes an important element. We can categorize B2 into these few scenarios:

1. The consumer doesn't move, the merchant/supplier moves. Delivery, on-demand services. Typical examples are Meituan Delivery, Ele.me, 58 Daojia (home services)

2. The consumer moves, the supplier doesn't move. These are in-store services. Example companies are Meituan Dianping, Lianjia (property), and we may also include Airbnb.

3. The consumer moves, the supplier moves. Ride-hailing, bike-sharing. Typical companies are Didi, Mobike, Uber, EvCard (car rental).

4. Cross-cities. Hotel booking, travel, transportation tickets. Typical companies are Trip, Meituan Travel.

There are some intricacies with the different scenarios. For example, hotel booking is mostly used cross-cities, but there is also a high percentage of same-city hotel bookings. As such, the in-store business and cross-cities business consume the same supply.

Herein lies the problem. The roots of Meituan Dianping is the group buy business, and one of the group buy categories is the inventory from hotels for in-store consumption. (e.g. discount hotel stay for staycation). But you can't prevent consumers to group buy hotel bookings in a different city. (i.e. you get the group buy discount rather than the OTA discount.). Cross-cities hotel hookings is Ctrip's (Trip.com) scenario. Naturally, Meituan and Ctrip would come into conflict.

This has nothing to do with the company's values or whether the company is belligerent, it's a natural conflict due to user scenarios.

Ctrip used to be a major shareholder in Yongche (ride-sharing). If Yongche had done a better job, Yongche would have been a competitor to Didi. Since it didn't, Yongche is not competing with Didi. (Note: Yongche pivoted to more niche use cases.). Ctrip and Qunar today also offer private car service and so they may still be competing with Didi.

One of the important scenarios for private car service is getting to and from airports, so Ctrip and Didi's businesses have some overlap. When two companies' main scenarios are similar or they share some overlap, inevitably, there will be some friction. If the two companies are belligerent, a fight is unavoidable.

China and US Markets

US Benchmark

These are the four main scenarios of LBS.

Let's look at it from another perspective. For A, the leader today is Tencent. Who's the US equivalent for Tencent? Facebook. Google is corresponding to Baidu.

As we can see today, Tencent and Facebook's market caps are basically equivalent.

China’s leader in B1 is Alibaba. Alibaba is corresponding to Amazon. And their market caps are roughly equivalent. (Note: before recent market changes and Ant.)

But for the leader in B2, who's the US benchmark? In Meituan's development, one of the biggest concerns from all the investors, especially the US investors, is that it's very hard to find a suitable benchmark company in the US.

Why LBS differs In China and the US

We spent time trying to understand this. In LBS, why is there such a big difference between Chinese and US companies? We think it's mainly these four factors:

1. Labour cost.

   Large scale fulfillment team is needed for LBS. Our food delivery business on an average day roughly has about 500K people delivering. When we talked to our South Korean counterparts, they told us their biggest worry is that their monthly turnover of delivery riders is 30%. This is a societal infrastructure. China differs greatly from countries like the US.

2. Population density.

   We can see the US cities have a much lower population density comparatively. Corresponding to the low population density is the low density of merchants. The low density in both greatly affects location-based services. Regardless whether a lower population density is desired, for delivery, the impact is definitely negative. It affects delivery time and the time to get the meal.

3. Population size.

4. Generational competition.

   Looking at all the industries from a historical perspective, at different time periods or every once in a while, the industry would undergo an upgrading. For this upgrading, let's call it a generation. The US retail industry from when it started till now, it's probably at the 4th, 5th generation. If the mainstream retail format in the US market is 5th generation, and now there's Internet retail, which we‘d call the 6th generation, the generational competition is between the 6th gen and 5th gen companies.

Generational competition - Beating the Status Quo

What's the number 1 competitive element in an industry?

It's the status quo, not the competition between competitors. The US has a long commercial history, so many industries and businesses have developed very well over several generations. If we call the last generation companies as the conservatives, then the conservative forces in the US are very strong. In China, the conservative forces are very weak. 

When I just started working on the food delivery business, I conducted some market research and found something that surprised me.

The largest delivery site in the US is called Grubhub, but they only had hundreds of thousands of orders a day. Domino's Pizza had more orders than it did! Big chains like McDonald's and KFC didn't want to work with Grubhub either because they can offer delivery themselves. The delivery service has become a standard in established F&B chains. Additionally, the Americans eat pretty much the same things every day, unlike the Chinese, we have a lot of varieties.

This led to the last-gen solution being rather satisfactory, leaving very little room the the next-gen solution.

From the research that I did at the time, there were only three companies in China focusing on food delivery. In total, their orders were less than 100K a day. There's no wide consumer awareness [of this product experience]. Their finances were not great either. If we see them as our last-gen, when we started, they were very weak.

What does this mean? It means when we started, we were not just bringing the Internet to the food delivery industry, we were building the food delivery industry itself. In fact, since we're building the food delivery industry itself, our impact to the industry is much greater than just digitalizing the industry, our commercial value would also be greater.

This is a core difference between China and the US. Considering all these factors, we can understand why American investors can't understand Meituan.

Lastly, a simple judgment - for the LBS category, China is a much better market than the US due to our population density, labour cost, population size and generational competition.

Categorizing the A-side

Previously, we’ve covered the categorization of the B-side. Now, let's talk about categorizing the A-side. If we cut the A-side once more, we can divide it into 4 categories by the visibility of information:

1. A1 - WeChat, WhatsApp, Line

2. A2 - Facebook

3. A3 - Twitter, Weibo, Tiktok, Kuaishou, Instagram

4. A4 - Toutiao, Google, Baidu, Tencent Videos

When sending messages in WeChat, only the recipient can see. (Note: In moments, comments from a friend's friend on a friend's post won't be visible to you unless you're also a friend with the friend's friend. In short, you won’t see a user’s activities in WeChat unless you’re a direct contact.)

In Facebook, other people can see your friends. (Note: Unlike WeChat, you can see your friend’s friend’s activities on Facebook.)

Sending a tweet in Twitter will be visible to everyone, but your followers will see it first.

In Google and Toutiao, all the information are visible to everyone.

If a product incorporates features that show information with different visibility, then it becomes very hard to segregate the users.

For exmaple, when Weibo first came out, some users would have private conversations on Weibo's public domain. If we incorporate WeChat and Weibo's features into one product, then users would get mixed up and it's fatal for user experience.

On categorizing by information visibility, Facebook is the winner. Their products - Facebook, WhatsApp, Instagram - are all separated and they occupy A1-A3, only leaving A4 to Google. 

Similarly, despite the dominating success of WeChat, Weibo's survival is not affected, because the two has different information visibility. Weibo has 200M DAUs. 

Heart vs. Data

The core product methodology differs for products with different information visibility. Allen Zhang's product philosophy is fundamentally different from ByteDance's product philosophy. Allen Zhang says you have to use heart to make products. At Bytedance, it's all about data - whichever product has the better metrics wins. As such, to date, Tencent still hasn't succeeded in information stream and short videos. Of course, ByteDance's Feishu (Lark) is not making any progress either.

This is due to the fundamental differences in product methodology. To oversimplify, the difference is, for A1 and A2, you mainly rely on heart, and for A3 and A4, you mainly rely on algorithm.

Tencent has long relied on human brains rather than machines, which resulted in them not being strong in machine learning algorithms, espcially after they sold their search business to Sogou. However, it's not that they don’t understand algorithms, it’s just their ability to apply algorithms is not that strong.

Superapps

Categorizing also has implications for which products can be made into one app and which can't.

Taobao and Alipay have huge traffic, but putting Ele.me there didn't do much.

WeChat and its Moments have such great traffic, but Weibo is still thriving.

It's also why Meituan is able to consolidate so many things into one app, because most of them are in B2.

Douyin and Toutiao are separate apps. They didn't incorporate Douyin into Toutiao despite Toutiao's huge traffic.

All these have to do with categorizing. Categorizing correctly involves core competencies, resource allocation, consumer psychology, and organisational abilities.

(Lastly, a graph to summarize everything.)

👋 Hi! I’m Tao. As I learn about building products & startups, I collected some of the best content on these topics shared by successful Chinese entrepreneurs. I translate and share them in this newsletter. If you like more of this, please subscribe and help spread the word!

Subscribe now

Subscribe now

Apologies that this issue came a bit later in the week. Hope you enjoyed the Genki Forest series. Back to Wang Huiwen's course. This is one of the harder sections to translate. Some pondering and appreciation are required on your part. I chose the word "tiering" to denote it's a verb, an action. Some related language for the concepts described here includes market segmentation, high-end vs. low-end / brand hierarchy, product mix, etc. It's slightly different from the segmenting, positioning described in STP. Tiering might connote it's one-dimensional, but as you'll see, it can be more dimensions. In essence, it’s dividing any distribution of users, clients, product value, etc. into several chunks according to some attribute.

The important things to note are a) it's an act, practiced in actual operations (not just strategizing) and b) we only know the truths from actual practice.

Also important to note is that in China's tech parlance, there is a dichotomy between product (tech, product, design), and “operation”, which is a catch-all for everything else. No less attention is paid to operation.

Tiering

Why do we need to use tiering in our operation? It's related to the STP. No single product can satisfy all the needs in its category. 

If it's still not obvious to you, try changing how you define the category. 

Use chewing gum as an example. You may think chewing gum is a solution for bad breath. For most people, the cause of bad breath is tooth decay. If the cause for bad breath for most people is tooth decay, then dentists and chewing gum belong to the same category. 

It all depends on whether you're looking at it from the demand (problem) or the supply (solution) point of view. 

Let's look at another example. In most places, the railway is a state-run monopoly and it seems there are no competitors. However, trains solve the problem of transportation. So airplanes are one of the competitors to trains. Even within trains, we also have:

1. high-speed rail and conventional rail (speed)

2. business class, premium class, and economy class seats (comfort)

3. soft sleeper and hard sleeper (long-distance, comfort)

No single product can satisfy all the needs in its category, and no single tiering (i.e. product mix) can satisfy all the needs. 

Different tiering can have vast differences. If you didn't tier your operation but your competitor did, it would result in you being uncompetitive in their tiering. (My understanding: it's up to you to define but there may be some consensus in the industry.)

Tiering is hard. The challenge of operational tiering is great at both the cognitive and organizational levels. 

For example, in the Group Buying business, if the company feels that they have too few users and they want to have more, the simplest way is to engage the top merchants ("head") because they are the most attractive brands to consumers. 

On the other hand, if the team feels there are too few merchants, they would engage the bottom merchants ("tail") because they are the greatest in number. As a result, the middle merchants ("waist") get ignored. 

Why tiering is so hard is because it's not just testing the company's operating logic, it's also a test on the company's organizational capabilities. Using just one KPI to evaluate the performance would result in errors. 

Types of Tiering

There are two types of tiering. The first is tiering with different products. The second is tiering with one product. 

One Product, One Tier

For example, the x-axis is usage/capability (from personal on the left to business/enterprise on the right), and the y-axis is price. 

On the top-right quadrant is IBM in the 1980s and on the left-bottom quadrant is PC (personal computer). On the surface, the two don't belong in the same market, as such, IBM didn't pay much attention to PC. However, as quantity, cost, experience, and performance change in one market, the relationship between the different tiers may also change. The gap between IBM's business computer and PC was getting smaller and smaller. 

If this is a gif (you’re welcome), you can imagine at first the top-right quadrant is a very big area, so big that IBM ignored the PC segment, which is a very small area. As the bottom-right quadrant gets bigger, IBM also dedicated a small team to work on PC. Since it's a small team, they couldn't manage the whole value chain of PC so they came up with a standard to consolidate all the suppliers. Even for the PCs that we use today, aside from Apple's Macs, they still use IBM's standards. 

As time passes by, the PC market grew bigger, the cost got lower and the experience got better, and it started to eat into the business computer's market. 

It's a similar situation for Alibaba to de-IOE (Note: midrange computers from IBM, database from Oracle, storage systems from EMC Corp - a staple backend stack for some big tech companies. One of the reasons to reduce reliance on IOE is because the quality of open source projects got better.). 

Therefore, the seemingly unrelated two industries may come into conflict as time passes. If you started out in a market that will be squeezed by others, then your hard work in that market may just be a fool's errand. 

Multiple Products In One Tier

There is another terrible scenario with the tiering. If a company operates products belonging to two different tiers simultaneously, like P&G, which operates many shampoo products, there is a danger to mess up the positioning of different products during the process. 

Rejoice or Header & Shoulders, which one is more high-end? 

If people have different answers to it, it means the tiering has failed. They must have done something that destroys the tiering to cause such a chaotic product positioning.

Let's start with the answer first. Head & Shoulders (anti-dandruff) is the lowest, followed by Rejoice (smooth hair), and the highest tier is Vidal Sassoon (styling). But today, we're not sure about the relationship between Head & Shoulders and Rejoice.

That's because, no matter how great the business, one day the market will be saturated.

If the Head & Shoulders team found out that their market has saturated, what would they do? They'd quietly raise the price of the product, and to match with the increased price, they'd have to position the product slightly more premium in their advertisements. 

Alternatively, the Rejoice team might have to lower the price, lower the cost and lower the brand positioning in their advertisement accordingly, if their segment became saturated. Gradually, the consumer perception changed and the two brands converged in their minds.

Tiering & Consumer Perception

I'll give you another example on consumer perception. I'm a loyal customer for mineral water and soda water, and I always go for one brand. Because, for the other brands, their bottles would become flattened as you drink. As a business person, it'd be quite awkward that my bottle gets deflated when I'm in meetings. 

In the supermarket, the hardness of the water bottles is positively correlated with the price. A common practice among beverage makers, when faced with profitability pressures, is to make the plastic bottles thinner since most people won't notice it.

It's a subtle change but it's also very core. When a brand doesn't form a sharp impression in consumers' minds, the consumer recognition of that brand would also go down. That declined recognition would slowly make room in the STP segment and let a more sharply defined company occupy that space. 

Tiering & Brand

Every company has some product lines and one or two flagship best-selling products. Each product has its own marketing team. As the flagship product has a higher total profit, the team would be given higher flexibility on how to spend money and they are more likely to spend money in all sorts of places.

General Motors used to hire Tiger Woods to endorse its mass-market car brand, Buick. But if Buick is represented by Tiger Woods, who can represent Cadillac?

There might be positive feedback in the short term, but it's gonna be negative in the long term. This is one of the tricky things in a company's operation. 

More Subtle Tiering

Coupons

Yet another type is you don't realize consciously that they're tiering. A typical example is McDonald's discount coupons. This coupon is not easy to use. It requires various combinations to save a few dollars or get a gift. This is actually a tiering method (i.e. price differentiation) for McDonald's to expand its user base.

This is an example that illustrates the importance to understand the 4P theory and STP theory well. 

There is a mantra in the Internet industry which is to create the best user experience possible. However, the McDonald’s coupon experience is clearly very troublesome. This is because this experience is a filtering mechanism to create a tier for the price-sensitive but time-insensitive customers. If you don’t create the coupons, you miss out on a customer tier and it's bad operation.

Many Internet people pursue the ultimate user experience. It's because the zero product replication cost and the large enough user base make the user experience the biggest value creator. But in some industries, those bad user experiences are sometimes part of the product design.

Membership

Costco has been very successful with its membership model, and Amazon Prime is also doing very well. As a result, creating a membership model became a fascination in the Chinese Internet circle. 

However, creating a successful membership model is an advanced operational technique and not all companies can master. For example, McDonald's certainly took the coupons into consideration in Pricing. Same for Costco, it also took the membership into consideration when it set product prices.

Low-end Disruption or High-end Disruption

In the past, investors always said that Meituan Delivery which started with the campus market is too low-end and can’t make money. Dajia is more high-end targetting the white-collar market. 

To explain the outcomes that we see today, we need to think about how we do tiering in the Internet industry.

Intuitively, everyone thinks that targetting high-end is more profitable. Baidu Waimai seemed to be more high-end than Meituan - in the early days, they had more neat rider uniforms, more premium advertising, and sleeker App design. The question on everyone's mind was - so in the future, will it be…

1. high-end and low-end delivery platforms co-exist

2. high-end disrupts the low-end

3. low-end disrupts the high-end

(like many cases in the book "The Innovator's Dilemma").

Regardless of which view you hold, there are supporting examples from the past. For high-end disrupts low-end, there are Tesla, Porsche, Ctrip vs. Qunar. For low-end disrupts high-end, there is the hard-disk industry. There is also WeChat which disregards the high-end/low-end dichotomy and is used by everyone. 

In the delivery business, the most likely scenarios are ii) high-end disrupts low-end, followed by i) high-end and low-end co-exist. (Note: He didn't explain why, but you can think about it yourself.) The reason why Baidu Waimin failed is that they're fake high-end - they have neat uniforms and premium advertising and but their average order value is even lower than Meituan's. 

Clients Tiering for Software Companies

When Meituan was acquiring an F&B management software company, our strategic investment team did a tiering based on the relevant companies in the market - the company that is chosen by the top F&B companies must be the one who's most competitive.

This logic is correct at first glance. There were two biggest companies in this space, TzxStar and "Choice", and Meituan made an offer to both companies. We made the offer to "Choice“ first, but "Choice" went with Alibaba, so we acquired TzxStar. 

After the acquisition, we discovered that TzxStar had very bad financials. After some research in the industry, we found that companies with good financials are those with a low share of top F&B clients in their revenues. 

The top clients would have a strong negotiation power, and they would also a lot of customization requests. Doing customization is a black hole for software makers.

Additionally, there are still new entrants in the F&B management software space. What a software company would do is to work with a few top F&B brands on some friendly-priced, customization-laden projects to establish credibility, then serve the mid-tier ("waist") brands. The software companies that work only for top brands are very hard to make money. 

One of the reasons why B2B software has progressed so slowly in China is because right now, China has a big "head" and a fat “tail”, but not many “waist” businesses. This makes it extremely hard for business software makers to find a suitable customer group to productize. 

Find the article insightful? Please subscribe to receive more learnings from Chinese Internet companies and I’d appreciate it if you can leave a comment or help spread the word!

Subscribe now

Subscribe now

This is the last part of the Genki Forest Case Study series. See part 1 and part 2.

If you want the original slide deck, share this newsletter on Twitter/Linkedin/Facebook or with your friends/colleagues via Slack, etc, and send a screenshot of the share to chinaplaybook@substack.com

Share

05 Flat Channel Management

After analyzing product, let's look at channels. After all, there is no avoiding that Genki Forest would have to come face to face with the incumbents in offline sales.

Based on our understanding, Genki Forest has achieved a total sales volume of over ¥2.5B RMB (~US$390M) across all its channels in 2020, far exceeding its goal.

On breakdown by channel, the biggest chunk, ~40% is taken up by traditional channels such as F&B retailers and schools (varies by geography, here is an average estimate), followed by convenience stores (CVS) (35%) and e-commerce (25%). Key Accounts (KA) such as big supermarkets account for the remaining 1%.

We estimate, by 2020, Genki Forest has already set up over 700K sales endpoints over the country and has over 2,000 salespeople to interface with the distributors.

From our research, Genki Forest's offline sales model is not too different from those of other traditional companies.

Their key to achieving a breakthrough in channel is to leverage technology in order to implement flat management for higher organizational efficiency - quite characteristic of Internet companies.

As shared in his speech, the way for a small company landlocked by giants to break out and win is through efficiency.

He understands it as the Power Law: if a man has $3 billion dollars compounding at a rate of 1.1 per day, and another man has only $100K but compounding at a rate of 1.2. After some time, the second man will be richer.

The "Power" is more important than the base.

(Note: This sounds more like Compounding, but compounding is a power law.

In another example, he explained why it may be a good idea to double down on your winners, as he painfully learned.

Suppose you have a product scoring 100 is compounding at a rate of 1.2. Your ego or misjudgment tricks you into making another product since you think you’re so good at it and there’s another opportunity. Splitting your resources into two projects so you have 50 x 1.2 + 50 x 1.2, which is the same as before, 100 x 1.2. But you feel great because you made two successful products.

But you’re better off pouring all your resources into the first product (Happy Farm in his case), to focus on improving the power from 1.2 to 1.4.)

For big companies, they have almost unlimited resources. Habitually, they solve every problem through money.

The employees at these companies spend most of their time on "managing up" - how to make powerpoints for reporting, how to get budget, how to help the superior hit his KPI. Naturally, the "power" is greatly reduced.

On the other hand, the extreme lack of resources at small companies can keep the company running at top speed, thereby greatly improve the "power" and it becomes the key to success.

He has personally lived through it with Elex's YAC (Yet Another Cleaner) antivirus software beating a big company (Baidu) with a hundred times bigger budget.

We will cover some examples of how specifically Genki Forest increases the "power" in their management.

1. Digitalize Office Work

Starting from 2018, the entire Genki Forest company has gone paperless - anywhere from manufacturing to approval, reporting, etc.

They first used DingTalk and developed tools and processes based on Alibaba's infrastructure. Using DingTalk as the backend, the front-line salespeople could use the "Store Canvassing System" to keep track of sales and profit targets. Through the smartphone, the company could also see how operations people have been hitting the ground as well as receiving daily and weekly reports.

Later, Genki Forest switched over to ByteDance's Feishu (Lark) as the office automation system.

In addition, Genki Forest also developed their own store canvassing app, with functionalities such as listing out the outlets, route planning, issuing receipts, data dashboard, etc., to allow front-line personnel to plan and optimize their canvassing.

All these sound great but how do they actually impact efficiency?

Let's look at an area of operation in detail. From a sales rep to a regional director, there are five layers between them. If the person in charge of any of the layers takes more than an hour to approve, there will be a monetary penalty. As such, the approval process and issue handling can be done extremely fast. To apply for a new campaign, you can get the necessary approval within 3 hours, whereas in a traditional company, it'd take 3-5 days.

Aside from within the company, Genki Forest also uses digitalization to improve the management of distributors.

The following image shows the one-button order app for retailers, "Genki Select" 元气良品. The product prices vary for different retailers.

2. OKR (Objectives, Key Results)

Genki Forest has introduced OKR to the company relatively early, in place of the traditional KPIs. The benefit of OKR is to let employees have clearer goals and to boost their self-motivation.

To illustrate, a regional manager could set the Objective to be XXX million sales revenue, and the Key Results would be the number of endpoints, number of cities, number of shelve listings, number of fridge placements, etc. Two other common Objectives are a) team building and training and b) improving operational quality.

The setting of objectives trickles down from a national level to individuals. Eventually, these objectives are enforced through performance reviews.

More could be said but we'll stop here.

3. Amoeba Management

Those familiar with management studies may have heard of it. Created by Kazuo Inamori in the 1960s, the system divides the company into small units called "amoebas", each responsible for its own management and profitability.

In this system, every employee plays a major role and voluntarily participates in managing the unit, achieving what is known as "Management by All."

For the sales team at Genki Forest, that's exactly how they operate. Every region is divided into smaller teams. Each team is in charge of its own operating profits. 10-15% will be distributed among team members while the remaining belongs to the company.

This allows maximum flexibility for the salespeople. To a certain degree, it also helps Genki Forest with budgeting and output management.

For a traditional company, the whole year's marketing budget has been decided last year. Regional sales directors would have to apply for budgets from the headquarter and go through multiple layers of approval. That leaves very little room for adjustment.

The Amoeba system is still a rather innovative approach considering how beverage companies have been running traditionally. For example, Nongfu Spring implemented its first Amoeba project in 2019.

4. Offline Data Feedback

For traditional FMCG companies, one of the main reasons for their lack of "flatness" is that the wholesales-retailers are not willing to provide offline data [if they have it in the first place]. As a result, the company's headquarter would have a long delay to get the latest data from the front lines. That's a problem that Genki Forest is trying to tackle as well.

We found that Genki Forest employed a data provider called "WinWin". (Note: 码上赢 - "Win on code" - I guess QR code or bar code.)

"WinWin" integrates with the POS machines at retail stores to directly collect transaction receipts data. In turn, WinWin provides accurate sales data to Genki Forest.

In 2020, Tang's Challenjers Capital even invested in WinWin in their Series A+ round.

A frequently used metric by Genki Forest is the "Coke Index", to quantifiably evaluate the sales levels.

Why an index?

If you solely look at the absolute sales figures of your products, you would be easily affected by confounding factors such as seasonality. As such, you need a stable reference point to judge whether a new SKU is doing well.

From industry figures, the classic 500ml Coca-Cola Coke has an 80% placement rate in all sales endpoints all over China, and its sales are relatively stable across the four seasons.

Defining one week as a sales cycle, through comparing the "Coke Index", Genki Forest can continuously monitor which SKUs are selling well and need to double down, as well as which are not and need to be taken off the shelf.

The speed at which this kind of insight is generated is several times faster than the traditional methods.

An example of how this is applied is the "Alienergy" energy drink. The product was first tested in selected convenience stores. The data quickly reported that it's not selling well and it's immediately being taken off the shelves. Now they're testing with "Alien" electrolyte water.

Of course, the retailers being unwilling to provide data is not so easily solved.

In its 2020 Distributor Conference, Genki Forest announced its plan to install 80,000 smart vending machines so they can monitor endpoint sales data in real-time. Comparing to Coca-Cola's 870K and Nongfu Spring's 360K, it's a much smaller number but it's a start.

In the following section, we'll explain the two characteristics that we discovered about Genki Forest's strategy on channels: heavy on convenience stores, heavy on e-commerce.

People who have worked in the beverage industry know that it's basically impossible to make money from placing in convenience stores because of the high commission. Logistics, fridge usage, marketing fees ... taken together, the commission could be as high as 30-40%.

Genki Forest still decided to place their products in Bianlifeng and 7-11. We think it's because they are eyeing the growth potential of convenience stores, and the overlap between the target audiences of convenience stores and Genki Forest.

As shown in the following chart, convenience stores are the second-fastest growth channel, only behind e-commerce. In 2020, the convenience stores market size in China has reached ¥250B RMB.

On the other hand, using 7-11's data to illustrate, the customer profile of convenience stores are mainly young people aged 20-29, and the largest two customer groups are white-collar workers and students.

However, due to high the barrier to entry to get on convenience stores' shelves, it has been a process.

Before 2020, Genki Forest outsourced the work to distributors. Genki Forest supply the goods, the distributors would negotiation with local convenience stores. It was quite efficient.

However, as there is a middleman involved, the channel fees were never transparent, and Genki Forest risks the drop in operation quality and the convenience stores going out of business.

Hence, starting from 2020, Genki Forest started to self-operate by working with large convenience store chains.

As for Genki Forest's e-commerce channel:

Data shows, the proportion of Genki Forest's online sales is much higher than that of traditional companies. For example, Nongfu Spring's e-commerce sales only account for 5% of the total sales. In 2018, Genki Forest's online sales account for more than 30% of the total sales. Tmall and JD are the main sales channels. As can be observed, 2020 is the break-out year for Genki Forest. Just its Tmall sales volume quadrupled and totaling ¥450M RMB.

Based on our analysis, we think it's due to significantly increased marketing spend.

From the e-commerce platforms’ data, we can also know that Genki Forest's sales are driven by hits. The Soda alone accounts for almost half of its total sales.

On the other hand, other than the Milk Tea released in 2020, the other products are lackluster in comparison.

We originally thought as since e-commerce is a self-operated channel, it should have a higher gross margin. Unlike offline channels, it could be a profitable channel. In the end, we found that although marketing campaigns are done by the company itself, the supply of goods is still through distributors.

That is to say, Genki Forest itself is still earning the small spread from distributors, no different from offline channels.

Plus, the retail price online should be lower than offline. With the advertising, live streaming discounts, the gross margins online could even be lower than offline.

We speculate that Genki Forest is prioritizing expanding the scale of the channels while keeping an eye on operating costs.

06 Building a DTC channel with Private Domain Traffic

Admittedly, Genki Forest is more than just the marketing hype. But undoubtedly, the omnichannel marketing capabilities are key to Genki Forest's success.

Based on our estimation, Genki Forest spent over ¥600M RMB on marketing online and offline. As a percentage of revenue, it's significantly higher than other beverage players.

Many other upstart FMCG brands adopted a strategy of "encircling cities from rural areas" (a Maoist war strategy) to break through the blockade by incumbents. For example, Feihe (an infant milk formula company) started from 3rd and 4th tier cities and slowly built up the premium brand before eventually conquering the 1st and 2nd tier cities.

Genki Forest did the opposite - encircling rural areas from cities, and concentrated its marketing on the Gen Zs.

This "attacking from high ground" strategy is a lesson that Tang learned from his gaming days.

In a 2015 interview, Tang attributed the success of Clash of Kings to this strategy:

"We dare to spend ¥1.8B RMB of our ¥2B revenue on ad placements, in global cities such New York, London, and Moscow to increase the brand awareness. Occupying the higher ground than go lower rather than the other way around is why Clash of Kings has become the most successful Chinese game in the international markets."

In 2020, Genki Forest sponsored 6 variety shows and a TV drama on satellite TVs and media channels such as Mango TV and Hunan TV, which focus on the female audience. At the end of 2020, Genki Forest paid ¥150M RMB for the sponsorship slot on Bilibili's 2021 Spring Festival Gala and gained 1.5 million followers on its new Bilibili channel.

Genki Forest's marketing strategy has already been widely covered so we won't repeat it here. That said, we’d like to expand on a recent focus of Genki Forest - Private Domain Traffic.

Genki Forest started laying the groundwork for its private domain traffic in 2020 - establishing an e-commerce store with the WeChat mini-program, setting up a WeChat service account, and using WeCom (WeChat for Work) to keep contact with fans - all on top of the WeChat groups it has already been running.

From the data that we gathered, Genki Forest already has over 200K followers on its WeChat private domains (WeCom + Service Account) and the mini-program e-commerce store has raked in close to ¥10M RMB in sales in the last 5 months.

These may not contribute much to the company's top line. However, this DTC channel is an excellent place to gather customer feedback and test new product ideas. For Genki Forest who would like to be as close to customers as possible, the DTC has great strategic value.

Next, we will break down how Genki Forest's private domain traffic operation works using the industry-standard "Traffic - Incubation - Conversion" framework.

First, on traffic, the primary method for generating traffic is through ad placements in WeChat's "Moments" (friends' feed). Intrigued users are led to Genki Forest's landing page and directed to follow its Service Account (where content, e-commerce, membership are conducted). The attraction is usually new product releases or heavy discount promotions. Users who want to get the discount vouchers must add Genki Forest's customer service to contact on WeCom, thereby becoming part of the "private domain", then go to the mini-program to place orders.

From our experience, the customer acquisition cost (CAC) with this kind of operation is ¥10+ RMB per follower. With that, we estimate that Genki Forest has already spent several million RMB on generating private domain traffic.

The second part is "incubation", where there is a much higher degree of freedom on how to go about it.

For Genki Forest, they:

• Group message 2-3 times/day a batch of product information in WeChat groups, mostly about discount deals.

• Direct Message 1 - 2 times/week new promotions through WeChat Business.

• On the mini-program, organize "Experience Officer" campaigns and invite users to participate in new product testing. The experience officers would publish their reviews and they form a content community.

Last but not least, the conversion.

We think the key metric here for Genki Forest is not the GMV from the private domain, but how much useful user data they gathered.

The private domain team probably doesn't have a hard sales KPI.

Unlike other e-commerce channels such as Tmall, Genki Forest lists a lot more SKUs on their mini-program marketplace - aside from Genki Forest's own products, they even listed oral care products and gift boxes from other brands.

In addition, Genki Forest is also selling annual membership priced at ¥98 RMB a year. The main feature of the membership is to save money. From the data, about 10K people paid for the membership, so there are quite a lot of loyal fans out there.

To conclude, in order to let you guys have an easier time synthesizing all these, we put up a process chart for Genki Forest's private domain operations - from getting new visitors to converting them to place orders and eventually become members.

07 From Finite to Infinite Games

Let's recap:

Genki Forest, through innovating on three fronts - product R&D, channel management, and marketing & promotions - brought the high-efficiency Internet operating mentality to the FMCG industry, thereby accelerating the rate of trial and error and laying the foundation for making "hits".

• Arbitrage Mindset: "Geographical Arbitrage" - bring mature concepts to China; "Neuro Connection Arbitrage" - Japanese-theme brand positioning to be associated with high quality in consumers' minds.

• Rapid Testing: Borrowing the A/B testing methodology from the Internet industry; Creatively collecting online and offline data. Data tell which products could become hits.

• Customer-centric Subsidization: Consumer demand as the highest priority. De-prioritize pursuit for gross margins. Focus on improving the product quality (taste), rather than maximizing scale and profits.

• Flat Management: Digitalization & Office Automation. Use the Internet companies' management philosophies to improve organizational efficiency.

• Build Up Private Domain Traffic: "High Ground" brand positioning and marketing; use DTC channels to keep close to customers, and to accelerate demand discovery and product testing.

While all these sound logical, the cost of innovation is to take risks.

No matter how much Genki Forest respects consumers' demand, it can't keep losing money forever. Eventually, they have to turn a profit, and to do that, they have to:

1. Control manufacturing costs

2. Raise prices paid by distributors.

In the past, Genki Forest has been relying on outsourcing factories and naturally, the production costs would be high.

On the other hand, this model has also resulted in production cycles and the mismatch between production and sales - excess inventories during off-seasons and out of stock during peak seasons. Selling last year's September stock during March this year was a frequent occurrence. Happening too often, distributors would have a lot of complaints. The usual solution to that is to send an extra batch of stock during the distributor's pickup, which would result in Genki Forest charging a lower than expected price to the distributor.

As such, Genki Forest decided that it's time to build its own factories.

Phase 1 of the first factory in Chuzhou has an annual production capacity of 30M bottles and a total production value of ¥850M RMB. It can manufacture "Ran” Tea, Soda, Milk Tea, functional drinks, and other beverages. Chuzhou phase 2 and a new factory in Zhaoqing, Guangzhou have already started construction in August 2020.

A new factory in the Northern-China region has also begun preparation. With the new factories, leveraging on the data team's optimization efforts, we believe the manufacturing costs will go down significantly.

The other problem with Genki Forest is its reliance on hit SKUs.

From the 2020 data, Soda Sparkling Water accounts for 60% of sales, and Ran Tea 30%, while Milk Tea and all the rest make up the remaining 10%.

Within Soda, the peach flavor as well as a few other SKUs account for the lion's share. By estimation, the entire soda market won't be bigger than ¥10-20B RMB. Relying on a few popular SKUs to hit its ¥7.5B RMB sales target in 2021 may prove to be a challenge.

As such, Genki Forest needs to accelerate its pace to discover new hit products, and broaden the product mix.

Nevertheless, Genki Forest never wanted to be walking on one leg. In Tang's own words, he wants Genki Forest to become "China's Coca-Cola", or even an FMCG empire. Therefore, expanding the product lines is something always on his mind.

Inside the company, Genki Forest is divided into several departments, including beverages, functional drinks, dairy, etc. Genki Forest has also been making plays with its subsidiaries - starting in 2018, Genki Forest began selling yogurt through the child brand, Beihai Ranch.

Externally, through investments by Challenjers Capital or Genki Forest itself, Genki Forest put even more horses on the tracks.

For example, we found that Guanyun Baijiu's controlling share (51%) lies with Genki Forest, and Genki Forest has started working on some liquor-exclusive sales channels inside the company.

What does the future hold for Genki Forest? Can they turn the finite game into an infinite game in the FMCG industry?

Let's witness!

Subscribe now

This is part 2 of the Genki Forest Case Study series. Part 1 is here.

If you want the original slide deck, share this newsletter on Twitter/Linkedin/Facebook or with your friends/colleagues via Slack, etc, and send a screenshot of the share to chinaplaybook@substack.com

Share

03 Data-Driven Product Testing

Without market validation, even the best product idea is moot. The following product testing section is the most critical link in the chain.

As Focus Media's CEO, Jason Jiang, shared publically, in the early days, Genki Forest tested dozens of ideas before eventually deciding on the "Ran Tea" product. The soda product has even gone through more than one hundred different varieties. All that took slightly over a year.

How is Genki Forest able to conduct such rapid new product testing?

Let's look at one of the most frequently used methods by traditional consumer product companies - focus group discussion (FGD).

In FGD, 6-8 people are carefully selected from the target audience to form a focus group. A highly trained moderator will lead the session. The data is gathered through natural conversations with the interviewees and/or a questionnaire to find out how interviewees feel about the product or the brand.

The FGD is usually conducted in a conference room with video recording devices and a one-way mirror so that participants won't be aware that they're being monitored.

It's a practice that has been used in the FMCG industry for decades and naturally, it's proven to work. However, there are two problems with FGD:

1. It's expensive. Outsourcing to top marketing consultancies like Nielsen would cost tens of thousands of dollars.

2. It takes too long. From recruiting a suitable target audience to end results, the whole project could take 2-3 months.

As such, for traditional consumer companies to do all these, plus the bureaucratic approvals and whatnot, developing a new SKU could take up to a year.

As put by Genki Forest's former R&D Director, Ye Suping, the modus operandi of Genki Forest's R&D is fast trial and error.

Take taste as an example. Internally, Genki Forest conducts a taste test every day or two, then makes quick adjustments. The whole R&D cycle is controlled to be within 3-6 months. They can release a new product as soon as 3 months.

Again, Genki Forest uses the Internet playbook to conduct a "dimensionality reduction attack" in a traditional industry.

They straight borrowed how they used data and rapid testing when making games to experiment with new drinks. Data is the truth. The way is cheap and fast.

(Note: 降维打击 "Dimensionality Reduction Attack" is a popular term in the Chinese Internet parlance. Originated from the sci-fi "The Three-Body Problem". Literally, it means a high-dimension, say 4D, alien civilization can use a dimensionality reduction weapon on a lower-dimension, say 3D, civilization, to reduce its dimensionality to 2D, thereby catastrophically altered its environment so they can't exist anymore.

Metaphorically, it means using knowledge from a more advanced field to overpower your opponents or to change the competitive environment to your advantage.

It's a good book. Bill Gates and Mark Zuckerberg read it. I've not read it yet, but will get to it.)

If you work in tech, you've probably heard of "A/B test" or "multivariant test" - basically having N versions of different user interfaces or processes, in the same time duration, let similar groups of the target audience randomly visit/use the different versions to see which one performs the best. The best version is carried over in the next iteration. Rinse and repeat, and your product gets better.

The independent variable could be an ad banner and you invest a small amount of budget for each to see which one has a higher click-through rate; or a button in the app whether to be green or red.

In China or the West, it's already a very mature methodology. For example, Facebook probably has 10,000 tests running at any point in time, even if a test only improves 1% of a product metric.

Genki Forest adopts the same mentality to test packaging, selling point, concepts, etc. From what we've gathered, Genki Forest has hundreds of SKUs in the "inventory", constantly testing and comparing results. Once validated, the SKU is put to scaled production and promotion. In Tang's own words, "95% of our SKUs this year haven't even been released."

However, this data-driven testing methodology can't get you the results overnight. In fact, it's a slow discovery process. But you're constantly iterating and multiple products can break out concurrently.

Chronologically, these are roughly the phases:

1. Taste Test

2. E-commerce Test

3. Convenience Store Test

4. Ad Placement Test

5. DTC Channel Test

1. Taste Test

At the very beginning, there wasn't any method, the only requirement was speed.

For packaging design and recipe R&D, Genki Forest outsourced to the experts. For example, "Ran Tea" was created through partnering with an R&D center in Japan. After the new drink was developed, they'd first let the salespeople try it. If they feel it's okay, then it's rolled out.

Not long after, they found that this was not a reliable way at all. After all, salespeople are quite different from actual consumers.

Here came the first iteration - they instead gave the products to college students (born after 90-95) for the first round of internal testing.

2. E-commerce Test

Obviously, online tests can provide a much richer and accurate data set for evaluation.

Genki Forest would list the products that passed the first round of taste tests on e-commerce platforms, mainly Tmall and JD, to see if they can actually sell. If the backend data indicate a new product passes the standard for scalability, then Genki Forest would then distribute it via offline channels.

This is a method still being used by them today.

For example, we believe that you probably have never seen the following product in supermarkets. However, in the Tmall shop, it records a monthly sales volume of 20K cartons. Exactly because it's still in the testing phase and the data doesn't seem to pass the standard, it's never rolled out offline.

3. Convenience Store Test

As offline is the primary purchase scenario for drinks, the actual behavior of consumers at shops is the one that should be tested the most.

To overcome the difficulty of getting data offline, Genki Forest chose a very straightforward method - they'd place the new SKU near competitor products, and through human monitoring or camera recording, to see what consumers would do when selecting products.

Through data such as "head-turning rate" (Note: I think it’s self-explanatory.), they can get a clear representation of consumer reactions and whether this new SKU will be popular.

Of course, to let the product be as close to the target audience as possible, convenience stores are the best choice. In the earliest days, Genki Forest tested at Bianlifeng stores. (Note: Bianlifeng 便利蜂, a tech-driven convenience store chain.)

4. Ad Placement Test

We discovered that Genki Forest places information stream ads on Taotiao, to test product selling points.

For example, Genki Forest may release a soy milk product in the future, so they placed multiple different ad copies, each emphasizing different selling points, such as "high protein, high calcium", "double protein" (Note: animal and plant), "zero cane sugar, low fat", "low sugar, low fat". Click on the ad will lead you to their Tmall page, but no actual product is being sold. (Note: i.e. Fake door testing.)

With audience groups being controlled, the click-through rates on different selling points can tell which messaging resonates better with consumers. Testing this way provides rich data and is very cost-effective.

5. DTC Channel Test

Be it a questionnaire survey or a tasting, it's a commodity method used by everyone. And with these traditional methods, the inescapable problem is recruiting the right people to participate in the tests.

Therefore, in 2020, Genki Forest started getting into "private domain traffic" on WeChat.

The main purpose is of course to conduct tests in a cost-efficient way. They gave a name to the people in their private domain - "Experience Officers".

To illustrate, Genki Forest publishes new product testing campaigns through their mini-program on WeChat, "Genki Member Shop" (previously, "Genki Research Institute"). The people who are active on Genki Forest's mini-program are mostly loyal customers. They'd apply to be an "Experience Officer". Once chosen, they'd receive the free products shipped to them. (They just pay for shipping.)

After the experience officers received the products, they'd discuss them in the tasting chat groups and the product managers from Genki Forest would guide them to fill out questionnaires.

From our observation, the products that have been tested so far include:

• "0 Cal Candy" - a candy that uses erythritol to replace white sugar.

• Chick-breast Sausage - low fat

• "Jewel Cup Yogurt" - a new product from Beihai Ranch 北海牧场, a subsidiary of Genki Forest

• "Pop Pearl Bobo Yogurt"

• "Alien" functional drink

Recently, Genki Forest has also been experimenting with big data tools to analyze feedbacks to unearth deeper insights.

You may ask: how come Genki Forest is able to inherit its Internet DNA, while the traditional incumbents can't learn from Internet companies?

We think the most important observation is the model is so new that it's hard to form a consensus on the methodology or experience.

Traditional companies have been the same methodologies for decades. No doubt, these methodologies have been time-tested and proven effective. If they were to use a completely different logic, it would be very hard to convince the stakeholders. Stable income is almost always the highest priority for them.

To draw an analogy: why do finance people have had such a hard time accepting bitcoin? The crypto market is so new that the hundreds of years of experience from the stock market hardly applies.

The second observation is about the flexibility of the organizational structure.

In the early days, Genki Forest's product department accounted for 10% of the company's total headcount. Even in 2020, they still have more than 50 people on the R&D team. (Note: Can't find the total headcount now.)

Every day, they're designing and experimenting with new product concepts, followed by testing. And they have high resource access rights.

In addition, their product development is organized in small groups, and they operate on the "horse-racing mechanism" prevalent in gaming companies, to inspire innovation through competition.

Conversely, in traditional companies, most of the R&D work is being outsourced. From product packaging to brand strategies, the marketing side is being outsourced to professional agencies. The product managers are usually few and they do mostly coordination work.

The upside of this model is that a few people are able to handle a very big operation, and the company sits on top of the value chain. The pursuit is maximum scale effect.

Obviously, if you fight your wars with hired mercenaries, you may improve your odds of winning some battles. But you would never have your own special forces. [that can save you in dire situations or is a sustainable competitive advantage]

04 Customer-centric Subsidization Model

A successful beverage product must overcome the inevitable challenge - taste！

Take Genki Forest's killer product - "Soda Sparkling Water" - it seemed to have solved a chronic problem of the industry.

Why?

Sugar-free healthy drinks are nothing new. Coca-Cola released its sugar-free version, Diet Coke, in 1982. Pepsi even led Coca-Cola by 18 years with the Diet Pepsi.

However, even today, almost all of the big brands' sugar-free carbonated drinks use artificial sweeteners like aspartame and acesulfame potassium in place of sugar.

The problem with artificial sweeteners is that it can be hard to meet consumers' tastes.

(Note: some history to illustrate the point on the importance of taste.)

Theoretically, the closer the taste of the sweetener is to the taste of sucrose, the more natural it feels and the better the "taste".

Aspartame is 200 times sweeter than sucrose, so it's hard to make a mix that tastes natural. And it contains calories.

Genki Forest solved the taste problem with erythritol - it's 60-70% as sweet as sucrose, so naturally, much easier to approximate natural sugar. More importantly, it's almost noncaloric.

From the patent of Genki Forest's soda, they first found that 7-8% sucrose concentration is the optimal taste, then used erythritol and sucralose to recreate the taste.

(Rough Translation:

The case listed is a soda drink, its ingredients include erythritol, sucralose, sodium bicarbonate, citric acid, food flavor, and water.

The case is intended to use ingredients to replace sucrose. Before deciding the substitute for sucrose, we conduct multiple tests to find out the optimal drink sweetness. The process is as such:

Using sodium bicarbonate, citric acid, food flavor, water, and sucrose as raw ingredients - sucrose as the source of sweetness - to testing for sweetness ranging from 3% to 11%, with an increment of 0.5%. Based on a tasting group of 50 people, with 3 tasting sessions for each test, the consolidated results report an optimal sweetness for this product at 7%-8% sucrose concentration. It’s lower than the similar products on the market with a 10%-11% sucrose concentration.

Thereafter, a variety of sugar substitutes are tested to approximate the 7%-8% sucrose concentration sweetness to find out if there are suitable low-calorie or noncaloric candidates. The process is as follows….)

(Note: Somehow I felt the need to figure out all the sugar substitutes from a product perspective, as summarized in the below chart. Obviously, it’s not an exhaustive research and doesn’t account for the taste, recipe implications, other health effects, etc.)

Here's the kicker. Erythritol was discovered in the 1850s, and it's commercially manufactured in China more than 15 years ago. Why don't the beverage giants use it?

Simple - the cost.

Based on the market data that we gathered, erythritol is 100 times more costly than aspartame.

This reveals to us the truth about the traditional beverage industry - the number one focus has always been profits and scale, not the customers.

Beverages are actually a highly lucrative business. The biggest moat is not the product itself, but the business model which brings high margins.

Coca-Cola has been consistently producing a gross margin of ~60%. The newly listed Nongfu Spring (SEHK: 9633) posts a margin of over 55%. The functional drinks brand, Eastroc Beverage, which recently submitted its IPO prospectus, also has a nearly 50% gross margin.

To scale this high-margin business as much as possible is what the big brands really care about.

Even if consumers may prefer the taste of the erythritol recipe, these companies are disincentivized to comply as it lowers their profit margins, no to mention the risk that the new taste may cater to too small a target audience.

Looking from their perspective, the chief concern for public companies is their earnings and the shareholders. No one would want their reporting numbers to look bad.

In a startup sharing organized by Sinovation Ventures, Tang repeatedly emphasized: in the current age, product is king, not distribution.

This is what he learned from paying much "tuition" in his web directories and gaming days. Pay attention to the second line on the slide.

Does this mean Genki Forest would practice what they preach and invest disproportionately more in product quality?

To verify whether Tang walks the talk, we gathered data from multiple sources to come up with an estimate for the manufacturing cost of Genki Forest's Soda Sparkling Water - it's about ¥2RMB.

Wholesalers (with underwriting) usually get a 15-bottles carton at the price of ¥38-42 RMB, so ¥2.5-2.8 RMB per bottle.

Essentially, the gross margin per bottle is ¥0.8 RMB (28%), much lower than what the traditional incumbents charge.

The extra cost is precisely spent on raw ingredients. From public data, the cost of raw ingredients of a 550ml Coke is around ¥0.2 RMB. In contrast, the cost for Genki Forest is at least ¥1 RMB.

In the early days, the cost of erythritol would even be higher for Genki Forest, as they import from the American company, Cargill.

If the marketing overhead is attributed to each bottle, the unit economics can barely break even. Plus operations, channel expansion, and other costs, we speculate that Genki Forest should be operating at a loss currently.

Nevertheless, Genki Forest is seizing the opportunity - put consumer demand as the top priority, subsidize the customers with cash-burning, and live by the idea that product is king.

Find the article insightful? Please subscribe to receive more learnings from Chinese Internet companies and I’d appreciate it if you can leave a comment or help spread the word!

Subscribe now

Subscribe now

One of the more interesting things about the China tech scene is startups are everywhere - not just in software. As Wang Huiwen covered a lot about marketing and product strategy in the last section, I think this in-depth case study on Genki Forest helps to deepen our appreciation. It’s a long article and took a lot of work (nothing compared to what the Growth Box team has done), so I’ll split it into 3 issues because of the email length limit and people’s attention span.

I’ve also obtained the original slide deck from the Growth Box team and translated it. If you want a copy of it, share this newsletter on Twitter/Linkedin/Facebook or with your friends/colleagues via Slack, etc, and send a screenshot of the share to chinaplaybook@substack.com

Share

Founded in 2016, the beverage startup Genki Forest has become a national sensation with the "0 Sugar, 0 Fat, 0 Calories" slogan on its bottles.

In just 4 years, its valuation soared from $0 to ~US$2B. (Note: now valued at US$6B with its latest funding in March 2021 led by Sequoia China.)

In this article, the Growth Box team comb through the remarkable achievements that Genki Forest has accomplished in the past 4 years, and share the learnings of our research & analysis.

Foreword

In our conversations with beverage industry experts, we've noticed some interesting observations:

1. The beverage incumbent brands are perplexed by the rise of Genki Forest - with their dominating grasp on channel distribution and extreme cost control, how did they let a small startup with a gaming background slip through?

2. Based on what Genki Forest's operators shared with us, Genki Forest is still a small company but with big ambitions - they've set their sights to become the next Coca-Cola. At the same time, they are dissatisfied with many of Coca-Cola's practices and they want to advance a different set of values.

These are some interesting questions that set us on a two months journey digging into hundreds of industry reports, consulting industry experts, analyzing millions of sales data points, and even going undercover in their community groups.

We discovered some counterintuitive characteristics about Genki Forest:

• Almost all of Genki Forest's products are the result of combining old elements in a new way. Each SKU can trace back to an original inspiration.

• Even as a new consumer brand, Genki Forest is not afraid to set a high price point, and is willing to pay for more expensive ingredients. The average cost of a Genki Forest soda is 1RMB, much higher than what traditional incumbents pay for their products. The sugar substitute, erythritol, chiefly accounts for the higher cost.

• Genki Forest's SKU development speed and R&D scale are several times higher than its competitors. But they only released 5% of their SKUs to market.

• The beverage industry is inherently B2B. The other brands usually only have 5% of their channels being self-operated, but Genki Forest has 22%.

• Genki Forest starts to invest in other brands in other categories as soon as their main products get on track.

To explain these characteristics, we will structure our discussion into 6 areas:

1. How did a game developer successfully crossover to make drinks?

2. The "Arbitrage Mindset" behind the consistent hit-making

3. Data-driven new product testing

4. A consumer-centric subsidizing model

5. "Flat" channel management

6. Building a DTC channel with Private Domain Traffic

We will quote some of Genki Forest's founder's thinking around product and business building. These are objective references to provide context, and we won't comment on them. Pick what you think is relevant.

01 The Game Developer Who Decided To Sell Drinks

Before Genki Forest, Tang Binsen has already made a name for himself in the Internet and gaming industry.

(Note: In 2008, his game "Happy Farm" was operating in 20+ countries, with 500M+ registered users. The game studio company, Elex Tech, was sold for ¥270M RMB in 2014.

In 2010, he took investments a ¥30M RMB investment from Kai-Fu Lee's Innovation Works to start the game publisher (among other things), XingCloud to help Chinese companies to expand and localize overseas.

In 2012, he took inspiration from Qihoo 360's free antivirus software model and started a web portal and antivirus software company serving the overseas markets.

In 2015, ELEX's self-developed game "Clash of Kings" became a hit, reaching No.6 on North American top-grossing charts. It set a revenue record for a Chinese-developed game in international markets and it's selected to be Facebook's 2015 Game of the Year.

Source: Baidu)

Despite that, Tang Binsen had zero domain knowledge and connection in the drinks industry. How did he create a beverage brand from scratch with zero prior experience?

Through learning, of course. Prior to founding Genki Forest, Tang Binsen had already researched into the consumer retail industry and even had some operating experience with soft drinks. He came prepared.

In 2014, after selling his gaming company, Tang started a VC, Challenjers Capital, which to date, has a cumulative asset under management of ¥8B RMB.

Looking through Challenjers’ portfolio in detail, we can find that other than supporting his former employees to venture out in gaming, Tang's another focus is in the consumer retail space. Challenjers is an early investor in Ramen Talk 拉面说, Panda Brew 熊猫精酿, as well as many liquor and drinks brands such as the new drinkers' favorite, Guanyun Baijiu 观云白酒.

Every successful investment in a company necessitates in-depth research into the industry as well as comparing hundreds of competitor and comparable companies. After the investment, the investor also has the privilege to be deeply involved in the company's operations. This is an excellent way to learn and accumulate experience.

Following this thread, we found something even more surprising - Tang had already invested in a company very similar to Genki Forest as early as 2015.

In Challenjers’ portfolio lists the beverage company UG Food 优选固本. (Note: not there anymore as when I accessed the site.) From the two companies' descriptions, they are almost identical - the operating model, and the "born after 90-95s" target audience.

UG Food’s Company Description:

Youxuan Guben (Beijing) Food Co., Ltd. was founded in 2015 and is a comprehensive fast-moving consumer goods company integrating research, production, supply, operation, and sales. Since 2015, the company has developed and created the first brand-new concept beverage product - "plus a bit" (nutrient fruit-flavored beverage). The core product consumer group is positioned at the young generation born after 90-95 who are young and vigorous and have strong self-expression.

Genki Forest’s Company Description:

Genki Forest was founded on 1 June 2016. It is an innovative beverage company with a complete industry chain, integrating research, design, procurement, production, logistics, sales/e-commerce, and services.

UG Food's flagship product is "plus a bit" 加一点 is built on the idea of "say no to sugar, say no to high calories, and say no to added fat", which highly coincides with the "Zero Sugar, Zero Calories, Zero Fat" selling proposition of Genki Forest.

If the previous two pieces of evidence may still be coincidental, then the following finding is the smoking gun.

UG Food's cap table lists Tang Binsen's wife, Xu Xiao, as a shareholder with 6% of the company's shares.

This is an unprecedented case in Challenjers’ records. As when UG Food started to fade out from the market, Genki Forest was born.

All these are to say, Tang was very serious about plunging into the beverage industry and he actually understands the industry.

Of course, aside from experience, another major requirement is capital.

Unlike other fast-moving consumer goods, the beverage industry has a very high market concentration. A few household names each cordoned off their own territories. For example:

• The carbonated drinks category is duo-polized by two international brands - Coca-Cola and Pepsi, with respective market shares of 59.5% and 32.7%.

• In the tea category, Master Kong 康师傅 and Uni-President 统一 have 60% of market share between them.

To build a new well-known brand would require a significant sum of marketing spend.

In addition, the beverage industry is essentially a B2B business and there is a heavy reliance on the wholesaler-retailer distribution model.

With the instant, compulsive consumption nature, the vast majority of drinks are purchased offline. E-commerce only accounts for 4% of beverage purchases.

Therefore, it's necessary to open your purse string to pay for the channel fees to make sure your drinks end up on the shelves and other retail endpoints. For convenience stores like FamilyMart, the slotting fees can be punishing.

As for recruiting talents, building distribution channels, and establishing a sales system, well, you can't skimp on those either. Indeed, in the early days, Genki Forest poached many top talents from the beverage giants.

Tang Binsen already had an illustrious career in tech, so capital was not an issue.

13 years before founding Genki Forest, Tang founded the gaming company, Elex Tech, after graduating from Beihang University with a master's in degree computer science in 2008.

With Xiaonei's support, his first game "Happy Farm" took over the country by storm. Shortly after, he brought the game to Facebook and it became an international hit.

Following this first pot of gold, Elex subsequently developed web portals, antivirus software, and another hit game, Clash of Kings.

After selling the company in 2014 for ¥2.6B RMB, Tang had already achieved financial freedom before he turned 32.

Although all the necessary conditions lined up perfectly, as to why Mr. Tang decided to make drinks, it's anyone's guess.

Here is the word on the street. When he was making games, he saw how his fellow game developers toil during crunch time, and they had to drink strong drinks like cola and Red Bull to pull them through. At the same time, these people have a strong demand for sugar-free drinks.

An idea formed in his head and he decided to "switch career" to make better drinks. As soon as the idea was clear, he started recruiting a team to help him do that.

02 The "Arbitrage Mindset" Behind Consistent Hit-Making

For every industry expert that we've consulted, we ask this one question: What is the core driving force behind Genki Forest's hyper-growth?

And their answers are surprisingly similar: they make great products.

So the conclusion of our research may disappoint you - Genki Forest doesn't have a "one-trick" strategy to consistently make hits.

In the beverage industry, the making of a hit is complex combinatorics of marketing, channel distribution, pricing, timing, and many other factors.

There is no fixed "hit formula". Whether you're an MNC incumbent or a new upstart brand, the only way to make something people love is through constant testing. Fair, at least in this regard.

As for Genki Forest, what they did was to drastically accelerate the "testing" process through relentless optimization and innovation, thereby increasing the probability of creating a "hit".

From the conception of a new product idea to market validation and feedback, and to making investments to improve product quality, Genki Forest imbued the Internet mentality into everything they do.

Let's analyze from the start - the conception of new product ideas. We think the core product strategy of Genki Forest is Geographical Arbitrage.

"The world is not and will never be so-called 'flat'. For successful businessmen, the competition is who can first extract profits from the information asymmetries."

Alexander Tamas, the founder of Vy Capital and a former partner at the Russian investment firm, DST Global, was the one who came up with this theory.

(Note: He may not have said these exact words. It's frequently quoted in Chinese tech media, and these are my translation.)

(Note: A 2010 Economist article (paywalled) which the origin of the "Geographical Arbitrage" theory traces back to. With the benefit of hindsight, we can see how it all played out. The fact it's from 11 years ago and I think most people still don't know/appreciate it highlights the point on information asymmetry and arbitrage. You're doing yourself a favor by reading this newsletter. 😉)

In the past decade-plus, DST has been utilizing and refining the strategy to perfection. Their first overseas investment was Facebook. Then they shifted their focus to less mature markets like China and invested heavily in JD.com, Alibaba, Xiaomi, and many more. (Note: ByteDance too). They also discovered unicorns in Russia and Brazil.

(Related Readings: 1. DST's non-census and right investment in Facebook in detail. 2. Alexander Tamas)

Interestingly, DST was an investor in Happy Farm's overseas rival, Zynga.

In an interview in 2015, Tang Binsen declared his love for the Geographical Arbitrage theory - "If a product can achieve great success in one market, it means it has potential to be promoted in other markets. People's usual concerns with cultural and habitual differences are hardly the decisive impediments."

Recall Tang's early Internet experiences, we can't help but notice whether it's web portals, free antivirus software, or mobile games, it's all about bringing the traffic models proven in China to less Internet-developed markets.

For a long time, half of Elex's revenue come from countries like Russia and Brazil. In these countries, for every 10 Internet users, one of them is using Elex's products.

Evidently, the same playbook that Tang used 13 years ago, is being used again today.

For example, one of the earliest investments made by Challenjers was Tiger Brokers. It basically copied the U.S. company, eToro's successful model to China. Now, Tiger Brokers is a public company (NASDAQ: TIGR).

Back to Genki Forest, the arbitrage opportunity is to copy mature, popular consumer concepts from more developed markets to China.

Based on reports, the logic behind DST's investment decisions is to look at the GDP and Internet penetration to evaluate whether a mature Internet business model can be replicated to the country, and what would be growth potential be.

If we were to superimpose this logic to the consumer goods category, the market size of the beverage industry determines the profit potential while the penetration rate of a product concept determines the growth potential.

In categories such as sugarless tea and sugarless carbonated drinks, Japan and other countries are clearly way ahead of China.

By CICC's estimation, in 2019, the penetration rate of sugar-free drinks is over 80% in Japan.

No doubt, the highly developed Japanese consumer market offers plenty of arbitrage opportunities.

Before Genki Forest's sugar-free tea product - Ran Tea (燃茶 "Burning Tea") came to market in China, Japanese consumers were already familiar with Suntory's sugar-free oolong tea, and Ito En's sugar-free green tea.

Genki Forest's Milk Tea may have also drawn inspiration from Ito En's milk tea and Fujiya's marketing.

The key factor at play here is either:

• Many foreign big brands' products were never promoted in China despite doing well in their home markets.

• Or they've been introduced to the Chinese market but just never caught on.

As the legendary American adman, James Webb Young, put, "An idea is nothing more or less than a new combination of old elements".

Of course, to create a unique brand, merely transporting ideas is not enough. That's why we postulate the second arbitrage - one based on brand positioning: the "Faux-Japanese" style is intentional.

In 2017, Luo Zhenyu (Dedao - "I get it" app founder, show host) invited Tang Binsen for an internal sharing. In the sharing, Tang recommended the book, Positioning, and put forth his "Neuro Connections" theory: arbitrage through human minds.

He gave an example - when "Japan" is mentioned, our brains' neuro networks make the association with "quality", as well as Japanese brands such as "Muji" that are known for good quality.

The brand positioning of Miniso is to intentionally create an association with Muji. (Note: The Chinese names look/sound even more similar.) Subconsciously, as you associate "Miniso" with "Muji", you associate "Miniso" with "quality".

There are two requirements here:

1. Find an empty space in consumers' minds and form a connection there.

2. That empty space must have value.

Evidently, all of Genki Forest's products to date center around "zero-sugar", "sugar-free" or "no cane sugar".

Under this "sugar-free expert" brand positioning, the tonalities of all its products are highly congruent with the positioning. The consistency also helps Genki Forest to focus on the sugar-free drinks category and the health-conscious core target audience.

Looking at the market from that vintage, it's both empty and valuable.

Find the article insightful? Please subscribe to receive more learnings from Chinese Internet companies and I’d appreciate it if you can leave a comment or help spread the word!

Subscribe now

Subscribe now

The Internet 4Ps in this issue is the second most mind-changing idea that I get from Wang Huiwen. Hope you enjoy it as much as I did!

See other articles in this Meituan PM series here.

4Ps in the Internet Age

The 4Ps theory that we covered in the previous class was developed in the 1960s, so it's mostly built on the context and constraints of traditional industries. The 4Ps theory is now less frequently brought up in the Internet age as compared to the previous merchandise age. Some people would even question whether it's still valid. In this class, we'll discuss whether the theory still stands and how to apply it in the Internet age. 

Bits vs. Atoms

The unit of measure in the Internet age is bit, whereas in the physical world, it's atom. Most of the peculiarities of Internet businesses have to do with the fundamental differences between bits and atoms.

Firstly, bits travel a lot faster than atoms. Bits travel close to the speed of light (i.e. fiber optics). Atoms travel much slower - moving one atom from the U.S. to China would probably take a minimum of 12 hours. 

(Note: If you’re interested in how the tech works: 1. How Does Light Carry Data? 2. The Global Fiber Optic Network Explained.)

Secondly, the cost to replicate is much lower for bits than it is for atoms. The cost of making a copy of a 1GB movie file is close to 0. To make an extra BMW car would costs at least tens of thousands of dollars. In some areas, it's not even possible to make an extra or an exact copy of a physical good. 

Thirdly, bits are programmable, atoms aren't. 

These three characteristics result in some fundamental changes to the 4Ps theory. Of course, some things stay the same. 

Take the design cost in Product as an example. Compare the R&D costs to develop the Windows Operating System and a BMW car. It may actually cost more to develop the Windows OS. But after the R&D is done, the cost of replication for Windows is much lower than that of a BMW car. 

In short, in the age of bits, the fixed costs are higher, but the cost of replication and the cost of transportation are much lower. 

How It Affects the 4Ps

Downloading a copy of the Windows OS may cost just a few cents or a fraction of a cent, even if you factor in broadband costs and electricity. But the cost of transportation to import a BMW car from Germany to China can be quite high. (Note: one number from Statista is logistics account for 3% of the total cost of a car.)

This difference in the cost structures leaves a lot of room for imagination for pricing. Due to the significant costs of replication, transportation, and storage, a BMW car cannot be priced lower than its production cost. Furthermore, you also have to factor in fixed costs as well as marketing costs. All these results in the price of a BMW car being quite expensive. 

In contrast, for software, the costs of replication, transportation, and storage are extremely low. This allows you to infinitely compress the price, even to 0. When the price is 0, it's a quantum leap for Promotion. 

In Internet companies, especially the pure online ones, from an organizational point of view, most of the heads of marketing come from a PR background, as opposed to from marketing or sales. 

In consumer companies like BMW, P&G, or Unilever, the heads of the marketing departments usually come from a marketing background. Their heads of PR usually report to their heads of marketing, which is the exact opposite of Internet companies. 

This is because after you have set the price to zero, the act itself has an inherent marketing effect. Unlike traditional companies, you don't require as much advertising. You can keep the marketing costs extremely low for quite a long period of time. 

I was in charge of the marketing department in Meituan. For a very long time, we didn't do much advertising and we mainly did PR. 

Case Study: Antivirus Software (Qihoo 360 vs. Rising)

In China, in the antivirus software space, there was a revolutionary war between Qihoo 360 and Rising. 

Rising (Box retail software)

Rising sold their software via a boxed retail model. The product is replicated on physical CDs and it's sold copy by copy. They had to advertise as well as establish retail channels.

This model resulted in quite a high cost to sell software. Advertising costs, sales channels, commissions, etc. Salespeople would go to many stores in many cities to talk deals - how much each copy should be sold, and how much to split with the channel owner. This resulted in high costs in Place.

That's on top of a relatively high R&D cost - at that time, antivirus software were considered to be rather technically complex. 

All these factors considered, the price of the product couldn't have been low. It's a price that the consumers felt that it's too high while Rising felt it's too low. 

Rising was doing well, but not great. 

Computer viruses were a scary thing for most consumers. And a lot of people used pirated software. Some virus makers would make their pirated software freely available on the market. (Note: Whether it’s via CDs in the early days, or downloadable files later on.) When consumers installed these software, their computers would get the viruses as well.

As such, consumers had developed a mindset that since pirated software are usually laden with viruses, but at the same time, they still want to use the pirated software, so their antivirus software cannot be pirated. This consumer cognition allowed Rising to set a rather profitable price point. 

Qihoo 360 (Cloud software or SaaS)

Qihoo 360 flipped the market upside down. Their product was free. Of course, that caused some changes in Place. Since the price was free, there's no channel willing to help Qihoo 360 with distribution. 

Qihoo 360 set up an official website to allow users to download the software for free. The product being free itself is a very compelling proposition. As such, there's no need for advertising. 

For a very long time, Qihoo 360 was the best in terms of PR among Chinese Internet companies. When I was in charge of Meituan's PR, I spent a lot of time learning PR from Qihoo 360. Their PR costs were very low. 

Firstly, they hired a bunch of copywriters. These copywriters were good at writing on the Internet to begin with, so they all had a solid knowledge base and good writing skills. Some were connected in the industry. 

Secondly, Qihoo 360 delved deep into what goes on in people's heads when it comes to antivirus.

If you think your computer is affected or possibly affected by some virus, what would you do? Usually, consumers would go on the Baidu search engine to search the error code or to post a question attaching a screenshot of the problem.

Qihoo 360 would consolidate all the symptoms of a problem and write an article about it. Towards the end of the article, they'd mentioned something like "Hey, there's this software called 360, and here's the download link. Install it and it'll solve your problem."

Qihoo 360 wrote all kinds of articles to be published on high domain authority sites on Baidu. When consumers searched for answers, they saw all the articles written by Qihoo 360. They wouldn't even consciously realize that they downloaded the 360 software, they were just following the tutorial. 

Their copywriting team was about 20-30 people, and each person had a KPI of writing 2-3 articles per day. The product and tech team have collected a lot of problems that happen on consumers' computers. The copywriters would just read a problem and write an article. It's like an assembly line. 

This method resulted in a much lower cost than other means of marketing, while Qihoo 360's market share grew fast. (Note: Basically, content marketing.)

Why did Qihoo 360 win?

Qihoo 360 was a late entrant in the antivirus industry, so their tech wasn't on the same level as Rising's in the beginning. However, Qihoo 360 understood deeply about consumer psychology.

Regular consumers can't tell whether their computers are really affected by viruses. They don't have the technical knowledge. There aren't any 3rd party institutions that provide this assessment either. Even if there are, consumers may not trust them. So how do you let the consumers know that your antivirus software is good?

It's a case where perception is more important than reality. 

Just telling the user that his computer isn't infected by viruses is not enough to make him happy. Any time the user feels that his computer is being hard to use or it's being slow, he would think that the computer is probably being affected by some virus. 

Qihoo 360 has a very conspicuous button that allows the user to actively kill the viruses. It may not actually kill any virus; it may just be deleting some irrelevant files that are dragging the computer down. But the user did feel that the computer has become faster, so he would think it must have worked. (Product managers absolutely need to understand consumer psychology and marketing.)

Sometimes, there may really be some virus. For antivirus, it's all about active defense. How long it takes for an antivirus software to recognize a new virus and find a solution for it is very important.

But you also must let the user know that your software can indeed solve the problem. If your antivirus software quietly kills the virus in the background, the user can't feel anything. 

Rising is a software company, who sold antivirus software via boxed CDs; it's not an Internet company. Their new software that can kill even more viruses has to be sold through the sales channels, and the consumers have to buy and install a new software. 

Qihoo 360 updates its software online in real-time. So even before Rising’s new software was sold, 360 had already solved the problem (i.e. new virus). This formed an impression in consumers' minds that Qihoo 360's software can kill new viruses as soon as possible. 

Aside from that, Qihoo 360 later released a feature that shows the user how long it took for the computer to start and how many people it beat. (e.g. “Your computer took 26 seconds to start. You beat 97% of others in the country.”)

Qihoo 360 understands consumers' mindset very well and it also has a very deep appreciation for "bits travel faster than atoms".

Rising makes more money by delaying updates, but Qihoo 360 played the reverse to let consumers feel that Qihoo 360's product is better. 

The Internet 4Ps

Qihoo 360 did many things right, but since its product is free, Rising had always thought that Qihoo 360 would go bankrupt sooner or later. Why Qihoo 360 didn't and in fact beat Rising? We can explain it using the new Internet 4Ps theory. 

All the people who think that the 4Ps theory doesn't work in the Internet age, as well as the people in traditional companies who think Internet companies don't have any operating methodologies don't realize this - bits are programmable. 

In traditional industries (like FMCG), all the 4Ps have their job to do and they're well separated. For example, Focus Media is doing Promotion, 7-11 and Walmart are doing Place. So only for Product and Price, you do yourself. 

This necessitates that for Place, there are channel fees, and for Promotion, there are advertising expenses. Every dollar that P&G gets from selling a bottle of shampoo to a consumer, some part of Price is being taken away by Promotion and Place. 

In Qihoo 360's example, the costs in Promotion are dramatically reduced. (i.e. via content marketing and word of mouth.) There is no Place. (i.e. The product is distributed by the official website/The producer owes the channel, and the replication & distribution costs are approximately zero.).

Zero terms cancel out. As such, the Price is back to 0. 

P&G's shampoo, when sold and arrived at a consumer's hands, is still a shampoo. Atoms are not programmable. 

In contrast, Qihoo 360's software arrived at the consumer's hands as a Product. But because bits are programmable, after some time, this Product becomes Place (channel). And Qihoo 360 starts to earn the money from Place.

Game publishers would use the Qihoo 360 channel to promote their games. (Note: Qihoo 360’s antivirus software is not one to sit there quietly and running in the background. It pops notifications frequently and even serves as a news portal. Users hate it now, but it was how a crop of Chinese antivirus software operated.)

As such, Qihoo 360 changed the money it's supposed to earn from Product to 0, in exchange for future revenue as Place. 

(Note: “If you aren’t paying for the product, you're the product.”)

Different Ways to Slice the Cake: From Retail to eCommerce, eBay vs. Taobao

In traditional industries, the 4Ps have a clear division of labor, and you earn what you're supposed to earn in that department. P&G has no retail shops, so it only earns from Product. Walmart has no in-house products, so it earns from Place. (Of course, Walmart now has private labels, but it primarily earns from Place.) 

There is a category of products that can serve multiples roles. For example, Taobao to P&G is simultaneously Place and Promotion.

When eBay was competing with Taobao in China, eBay made a mistake - it charged merchants a listing fee. For every product that the seller lists, eBay charges a fee, similar to the listing fee model in traditional retail. 

However, why traditional retail charges the listing fee is because there is limited shelf space in the physical stores. For eCommerce, there is unlimited shelf space. Taobao didn't charge the listing fee so it's an advantage for them.

Today, Taobao charges advertising fees, which means to the merchants, Taobao is doing both Place and Promotion. When a consumer sees a product on Taobao and is able to directly order, that's Place. Being able to rank in all sorts of ways (Note: including AI-powered recommendation), that's Promotion. Taobao combined Place and Promotion into one, and it made the fees on Place zero, and instead, it charges for Promotion.

4Ps & Business Model Design

Many powerful business models can be simplified as making one of the 4Ps 0 and instead charges from another P.

A more regular example would be Costco. If you want to buy some stuff from Costco, you have to pay a membership fee to become a member first. After becoming a Costco member, the more you buy, the more you save. 

Why Costco can achieve a lower price than other retailers? 

Many products sold in Costco are white label and private label products. Private label products don't need much marketing expenses. The costs that Costco saved from Promotion can be passed down to consumers in form of lower product prices.

Even in the face of the relentless Amazon, Costco remained one of the few retailers that are relatively unaffected, exactly because Costco reduced Promotion to 0 and instead, earns from membership fees. 

(Note: The membership and the services and experiences that it entails is the Product.)

The 4Ps theory on the one hand is a framework for us to design our business model. On the other hand, it's a way to define the profit distribution along the industry chain.

I think it's still a powerful theory today. Some of the best business models of today are essentially compressing one of the 4Ps to 0 to gain significant advantages, then to earn from another P. The Internet greatly increases the possibilities to do so. 

Subscribe now

Subscribe now

Let’s get you up to speed:

Wang Huiwen is the co-founder of Meituan, who recently retired at age 42 with an estimated net worth of >US$2B. He opened a Product Management course at Tsinghua University in September 2020. This article and others in this PM series are my translation and edits based on a compilation of the content of his course shared online. You can see other articles in this series here.

4Ps (Marketing Mix)

Some people think that the order of the 4Ps matters while others don't. I'm leaning towards it does matter and I think the order is Price > Product > Place > Promotion.

Price > Product

A whole book can be written on the topic of pricing, so I won't be able to talk about everything about pricing. However, I will focus on why I think Price comes before Product. In other words, price determines the product and not the other way around. The relationship between the two is similar to that of demand and supply. Generally speaking, in industries with scaled production, demand is more important than supply. Pricing affects demand, and product is the supply. 

For example, it is that the iPhone X should be priced at $1000, or the No.1 smartphone should be priced at $1000? 

iPhone's share of profit is higher than its share of shipments. Consumers are willing to pay for the best smartphone for $1000, regardless of whether the manufacturing cost is $100 or $500. That's the price premium of being the best in a category. 

Whether it's iPhone or Tesla, the product is not always satisfactory. But since they're the best brand, it doesn't matter. Consumers can pretend there's nothing wrong with the product. The bragging rights alone affords a high price premium. In the example of petrol cars, the high-end and low-end cars may have the same chassis. The low-end cars may merely be a lite version. 

Place, Promotion

The third component is Place, which we usually refer to as “channel”. You can have multiple channels. For P&G, Walmart is their channel. Now Taobao and JD are also their channels. For a game producer, their channels would be mobile app stores and Tiktok. 

The best scenario is that you have synergy between your STP and your channel. For example, if you're developing an ACG game (anime, comic, game), and you can get your source material from Bilibili as well as use it as a promotional channel. As such, your target audience, game title, and promotional channel are highly concentrated around Bilibili, and your chance of success will be greatly increased. (Note: Real-life example, miHoYo.)

The last component is Promotion. It's complex and highly practical, so I won't expound here. But I'll use an example to illustrate STP and 4Ps comprehensively. 

STP & 4Ps

Before Xiaonei ("On-campus Network"), we worked on Duoduoyou ("Many Friends"). However, Duoduoyou didn't get much traction, mainly because Duoduoyou was a social network meant for everyone. Of course, today, many social networks such as Facebook practically have everyone as their users. But that's not the case in the beginning. Even if your end goal is to create something that is used by everyone, in the early days of the product, you should only serve a narrow, well-defined segment of users. Remember that segmenting is not to choose the final market, but to choose an entry point. 

In the spring of 2015, we were discussing whether Duoduoyou had chosen too broad a market. At the end of 2015, we've decided to go deep and launched Xiaonei to serve a segmented market - college students. The product was of course designed to meet students' needs. In the autumn of 2006, the beginning of a new academic year, many freshmen came in and thought Xiaonei was the school's official website. 

The names themselves explain a lot. The reason we chose the name, Xiaonei ("On-campus Network"), was precisely because we segmented and chose the college campus market. Imagine if we marketed the mass market Duoduoyou ("Many friends"), would the students still mistake our website as the school's official website?

This is the magic of market segmentation.

In addition, in order to strengthen the recognition of our Positioning, you could only register for Xiaonei if your IP address is on the college campus or if you have the school's email. You could also declare your school and major, hometown, high school, etc. so we could aggregate the information and recommend relevant people to you. 

What this means is that, after we've done our selection for the STP, in the Product part of the 4Ps, we again made product design decisions to match with our STP. There's nothing too complicated or insightful with these product decisions, but they did indeed strengthen users' recognition of our product. 

We also hired hundreds of campus ambassadors to do promotions. The population density of students on school campuses was very high, so the promotional costs were very low. Alternatively, if you wanted to have a nationwide promotion at that time, your only choice would probably be advertising on national TV, which would be unrealistic for a startup. This step helped us to achieve high effectiveness and efficiency in Place and Promotion of the 4Ps.

If you recall the Diffusion of Innovations theory from the previous class, you would know that college students have a high willingness and ability to try new things - most of them would be Ealy Adopters or Early Majority. 

As such, don't worry about you've chosen too narrow an entry point in the beginning. As the Chinese saying goes, a needle can pierce the sky (一根针扎破天).

(Note: Figuratively, it means applying enough force at the correct point can crack any problem open or if you do one thing exceptionally well then nothing can stop you. Similar wisdom: 1. “It's better to do one thing well than ten things poorly.” - Heather Hart. 2. “Do not try to do everything. Do one thing well.” - Steve Jobs. 3. “Better to do a little well, then a great deal badly.” - Socrates.)

(The second half of the Chinese idiom is 一招鲜吃遍天 - if you are good at one craft then you can earn a living anywhere you go. English idiom - "When you have the skill, you can eat your fill.")

Doing STP right has a profound impact on product design and promotion.

4Ps are a combination

I'd like to emphasize that the 4Ps are a combination. For example, hotel reservation is a product. The corresponding service for this product is that if you have reserved a room on Meituan, but when you arrived at the hotel, the hotel tells you that they no longer have any vacancy, then our platform is responsible for helping you to find a room with guaranteed quality. Therefore, the price for this hotel reservation product will be relatively higher. 

Hotel group buy is a different product, and it will be priced relatively cheaper. It requires the consumer to call the hotel beforehand to confirm there is room availability. If it turns out that there is no vacancy when you arrived at the hotel, then Meituan is not responsible for that.

This is the relationship between service/product and price. 

In the hotel industry, hotels will do their channel management. Their first channel is walk-in customers. The second is tour groups. The third is corporate customers. And the fourth is online traffic. Among them, there are some big price differences. Corporate customers have the lowest price point but the demand is the most stable. Tour groups are affected by seasonality and the price fluctuation is big. The hotels have to do channel isolation to make sure the walk-in customers don't become corporate or tour group customers as walk-in customers have the highest price point. 

These channels are combinations of the 4Ps. A company that simultaneously manages several 4Ps combinations faces significant operational complexity. For example, P&G operates multiple shampoo brands. It's not an uncommon occurrence for a company that operates multiple 4Ps combinations to mess up their price, product, place, and promotion across their different products.

Pay attention to the matching relationship between STP and the 4Ps. Do the selection in STP first, then apply the 4Ps. 

Find the article insightful? Please subscribe to receive more learnings from Chinese Internet companies and I’d appreciate it if you can leave a comment or help spread the word!

Subscribe now

Subscribe now

Let’s get you up to speed:

Wang Huiwen is the co-founder of Meituan, who recently retired at age 42 with an estimated net worth of >US$2B. He opened a Product Management course at Tsinghua University in September 2020. This article and others in this PM series are my translation and edits based on a compilation of the content of his course shared online. You can see other articles in this series here.

Targeting

The second part of STP is Targeting, but it's definitely not an assembly-line-like process where you do Targeting after Segmenting. Sometimes, you'll have a difficult time doing Targeting that you have to go back and rethink your segmentation.

If you have trouble finding the right target market, it may be that you didn't choose the right dimension or the right level of granularity during Segmenting.

When Targeting, you must be very clear whether you are looking for the entry point or the eventual target market. There is a big difference. For example, the entry point of Xiaonei (“On-Campus Network“) was the college campuses, but the target market was all the internet users. 

Long-Term

When choosing an entry point, you should choose the one that is most advantageous to you from a long-term perspective. 

By long-term perspective, I mean that when there is a market opportunity, you saw it and so must the others. After you've gone ahead with this particular market opportunity, and there's someone else who had chosen this market too, and the two of you spent 5 years working on it respectively. Then you discovered that when you two have overlapping competition, you face unfair competition since you're on the disadvantageous side. As much as possible, you should avoid that situation from happening.

I left an assignment in the previous class - Taobao and JD, which one has the stronger Scale/Network Effects? (Again, Wang Huiwen considers network effects a subset of scale effects. You may pause here and think, before continuing.)

It’s JD. But today, Alibaba’s market value seems to be much larger. (Alibaba ~US$634B vs. JD ~US$118B). Network effects refers to as the size of the business (sales, #users, #customers, etc.) gets sufficiently large (vis-à-vis competitors), it creates significant advantages in:

1. User Experience

2. Cost of Production

However, Network effects is not the only law in the business world, just like how Newtonian Mechanics only hold true under low-speed conditions. Network effects is the law of universal gravitation in the business world; speed is the theory of relativity in the business world.

Taobao was able to grow faster because of its asset-light model. Who will win when one has the faster speed and the other has the stronger network effects? I don’t know. In China, now it looks like Taobao has won over JD, and in the United States, Amazon over eBay. (Taobao is China’s eBay, JD is China’s Amazon). 

In fact, Taobao and JD chose different markets in their beginnings. Taobao chose apparel and Jingdong chose digital 3C. (computer, communication, and consumer electronics). The difference is worthy of our attention.

When you are Targeting, you should not only pay attention to the size of the market but also pay attention to the scale/network effects and speed. 

The opposite of a small truth is falsehood, and the opposite of a big truth may be another truth. I can't give you a specific methodology here, but the impact of choosing the wrong target is huge.

Entry Point

Of the companies just mentioned, only Amazon really thought about the entry point on the first day. Amazon wanted to become the Everything Store from Day 1, but their entry point was to sell books. 

This entry point was very clever. At the beginning of a category-creating product, users must be fearful/skeptical, the industry’s infrastructure incomplete, and the company’s operating methods and systems lacking.

Why is the books category good?

1. Books have a sufficiently large number of SKUs. No matter which city you are in, there will always be books that you can’t get your hands on. Even if it's available in your city, you may not know which store has them.

2. Books are highly standardized in the United States.

3. For books, you have less to worry about in terms of logistics and warehousing. You can stack them, drop them, and they don't expire.

4. In the early days of e-commerce shopping, people were very worried about buying fake goods. But even in pirated books, the content is still authentic.

5. The average order value for books is not too high. (trust, risk, inventory management, etc.)

6. Most early Internet users have a relatively high level of education, and that closely matches the profile of people who buy books. 

From this choice of entry point, we can see that Amazon was highly strategic. (Watch Jeffery Bezos explain)

At the time, China already had e-commerce websites like Dangdang, 8848, etc. 

The entry point for Dangdang was the same as Amazon’s - books. Unfortunately, it did not seize the right opportunities to iterate as fast as possible to become an Everything Store.

8848's founder was thinking big and originally intended the site to become an Everything Store. However, it's difficult to manage too many SKUs as an early-stage business. 

Therefore, if you have ever experienced the confusion and struggle of finding PMF, you will be deeply appreciative of how amazing Amazon's selection of the entry point is. This kind of high-quality decision-making from its earliest days demonstrates it’s destined to be an exceptional company. 

Competitive Environment

The choice of T differs in different market competitive environments. In particular, you must be wary if you've chosen too broad a market.

Smaller competitors who adopt the Focus strategy (Porter's Generic Strategies) can slice off a part of your market segment and build a company there. You run the risk of being "Unbundled". In men's apparel, suits, jeans, shirts, and T-shirts each have their brands. If your target market is the entire men's apparel category, your risk of being unbundled is very high. 

To use an Internet example, there's a company called Craigslist in the U.S., which is similar to 58.com in China. There are many sections on the website, such as property sale, rental, house moving, pet care, making friends, etc. It's essentially the digital version of the classified section of newspapers in the print era. This company is not super big, with less than 100 employees, but it makes a profit of hundreds of millions of dollars, so it's a pretty solid business. (Note: $694M revenue in 2016, 50 employees in 2017).

However, the environment has changed since its founding. Someone created a graph and mapped out all the new companies that have replaced each section on the Craigslist homepage. Every category of Craigslist is being taken away by others. 

In China, Baidu is also facing similar risks. Baidu is a general search engine, but you don’t search for clothes and restaurants there anymore. Some user behaviors have gradually been migrated away.

Considerations

Targeting seems simple, but its applications are ever-changing, and the factors to be considered are very complicated. When choosing T, you must have product design, marketing efficiency, organizational ability, and capital efficiency in mind. 

Craigslist didn't go into segmentation, because each segment requires a team of hundreds or thousands of people to go deep into it, posing new, significant challenges to its organizational capabilities. (Note: Craigslist’s founder, Craig Newmark, is aware that he’s not a great manager.)

Let’s look at Meituan Delivery. When we started, we started from the college campus market. One of the reasons is that there are a lot of students on campus, so there is high customer density and the distribution for restaurants is relatively simple. 

In addition, we can also hire college students who work part-time to support their studies, and they don't cost a lot. We don't have to build a fleet or a fulfillment platform either. Self-operated distribution requires a much higher level of organizational abilities, but more importantly, it costs a lot of money.

As a 4-years late entrant in the industry, it'd be very difficult for us to persuade investors to invest in us if we want to build a self-operated distribution network from the start. Investors would think why I don’t invest in Daojia and invest in a late entrant? Daojia had already had 4 years of experience in the delivery market with a self-operated fulfillment model.

The good thing is, the college campus market can achieve a considerable GMV in a relatively short period of time. From the fast ramp up, we can establish some initial organizational abilities and systems, as well as gain investor confidence.

Positioning

P is a relatively complicated part of STP. Positioning in the perspective of users - what your product is, and why would they choose your product. There is a book called "Positioning" and another book called "The 22 Immutable Laws of Marketing". I won't go too much into these books. 

Positioning is directly related to the consumers’ minds. For example, Patek Philippe's advertising slogan "You never actually own a Patek Philippe. You merely look after it for the next generation.". They're positioning Patek Philippe products as heirlooms. 

In another example, beverages can be divided according to several dimensions - tea vs. carbonated, sugar vs. sugar-free, etc. Recently, sugar-free carbonated beverages have developed very quickly. One of the best-known brands is Genki Forest, which is positioned as sugar-free carbonated drinks.

"Sugar-free carbonated drinks" is not Genki Forest’s T, it's who drink sugar-free carbonated drinks.

Genki Forest’s T should be someone who thinks that high-sugar drinks are unhealthy, but at the same time, they think their body/health is still pretty good so there's no need to drink completely unflavored drinks. 

My health condition is not that great, so I can only drink mineral water or flavorless soda water. So, I'm not their target customer.

In terms of the relationship between T and P, T is more focused on demand and customers while P is more focused on supply and products. Effective marketing is one that can quickly associate T and P. 

If a 2C product manager does not understand marketing, it means that he hasn't even gotten a foot in the door yet, and he needs to build up customer awareness fast and cheaply. 

Closing

Going back to product. One of the most important things in building a product is the why. That's why this course spends a lot of time talking about strategy and the like, and not how to actually make the product.

Only by knowing why we need to develop this product can the Needs be served accurately. And only by knowing what the strategy is can we build products that match the strategy.

Find the article insightful? Please subscribe to receive more learnings from Chinese Internet companies and I’d appreciate it if you can leave a comment or help spread the word!

Subscribe now