#013 - Notes on Alex Zhu of musical.ly
11mins Read - 18 Product Lessons by Alex Zhu, co-founder of musical.ly
👋 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!
(This issue is best viewed on Desktop)
Hey there,
As usual, I like to take a break after focusing on one thing for a while. Wang Huiwen is about to shift gears in a bit so this feels like a good time to take a break.
Well, to continue our other thread, which is ByteDance, I’m also fascinated by the start of ByteDance’s another highly successful product. Not Tiktok (or Douyin), because it’s developed within a large company, so it’s hard for a starter to emulate.
Yes! musical.ly. It’s said that Douyin copied musical.ly. While short videos as a future trend had been on people’s minds even before musical.ly, Alex Zhu is certainly a top-tier product visionary that brought things together.
You can skip straight ahead to the product lessons part. My favorites: No.1 - Go With Human Nature and No.5 - Building A Community Like A Country.
Bio
Co-founder of musical.ly
Head of Product at Tiktok (International), reporting to Zhang Yiming
VP of Product & Strategy at ByteDance (Product Strategy, Company Strategy, M&A)
Previously Enterprise Software at SAP, designer at WebEx.
Wanted to do EduTech (Cicada, 3-5mins educational videos, “Twitter + Coursera“). musical.ly was a pivot.
Journey
2014, Alex Zhu and his co-founder left their previous companies to start something in education.
Their app, Cicada
Raised US$250K from an angel friend.
Developed by a 5-people team in Shanghai.
Allows industry experts to make 3-5 mins educational videos.
It didn’t work because education is going against human nature. — People want to be entertained, not educated.
On a train to Mountain View, Alex saw a bunch of rowdy teenagers singing, taking selfies, and making videos. So he got the idea to make an app to make short music videos - combine social networking, entertainment, and short videos.
July 2014, musical.ly went live, after one month of development. It allows users to make 15s short videos based on a library of music - lip-sync, dancing, etc.
After the launch in July 2014, musical.ly got 500-1,000 daily downloads and tons of user emails giving feedback and suggestions, and its App Store rating was 4.5. In contrast, after one year+ for Cicada, there were <50 user emails.
One hack they did at the time was to make the app name really really long so it contained all the keywords. The App Store algorithm at the time rewards this behavior.
Originally targeted at the high school and college students, musical.ly gained surprising popularity among younger teens.
There’s some traction, but not great. musical.ly was passed by 20+ VCs mainly because no Chinese company has ever succeeded overseas. They did get an investment from Cheetah Mobile (who’d play an important role later during the acquisition).
March 2015, the turning point came. musical.ly’s numbers grew fast.
April 2015, Cheetah Mobile’s RMB5M (~US$800K) investment came in. Apple App Store ranking ~1400.
Around the same time, musical.ly made a design change to automatically attach its logo to the videos when users share. Previously, they noticed that some users cut out the logo when they share their videos on Instagram or Twitter. This small adjustment allows the new audience to know where the videos came from. App Store ranking jumped to 100+.
8 weeks later, on 6 July 2015, musical.ly reached No.1 on App Store in the U.S.
Start of 2016, musical.ly DAU 2M+, Registered Users 20M+, about 6% of the U.S. population.
May 2016, musical.ly Registered Users reached 70M. Secured US$100M Series C funding from GGV, Qiming, Greylock, DCM China at US$500M valuation. Prior, they got $16.4M from Greylock and GGV.
July 2016, introduced live streaming with Live.ly.
April 2017, musical.ly MAU 55M, Registered Users 200M.
(Updated: I was just introduced to the book, Attention Factory, by Matthew Brennan. The above story on how the musical.ly idea came about is what Alex tells the press. The actual story - I won't spoil here - is in Chapter 5: From Paris to Shanghai, Musical.ly.)
Context
The “Home” Market
Douyin was launched in Sep 2016.
Early 2017, musical.ly started hiring roles to prepare for (re)entering the Chinese market.
February 2017, ByteDance acquired Flipgram, a mobile short videos creator community in the U.S. for US$50M.
Early 2017, Tencent led a US$350M investment in Kuaishou.
Short Videos Vertical
Pre-musical.ly Age
Vine (6s videos), acquired by Twitter in 2015, announced its closure in 2016.
The original founders of Vine struggled internally at Twitter. Vine withered due to Twitter’s lack of vision in the video space.
Facebook’s Riff (2015), “Ice Bucket Challenge“.
“Stories” - Snapchat, Instragram.
Tiktok Age
ByteDance acquired musical.ly on 9 November 2017, for a rumored US$1B.
The story on this was musical.ly’s investor, Cheetah Mobile’s CEO, Fu Sheng had a veto right on the deal and he wanted to package in Cheetah Mobile’s other investments News Republic and Live.me as a combo deal. Kuaishou and Tencent wanted to buy musical.ly first, but they were turned off (probably pissed off) by the combo deal. But Zhang Yiming had the stomach, so he took the pricey deal, and we live in this reality.
Of course, there’s interest from the U.S. side as well. Zuck was said to have met Alex in Facebook’s office. I guess Facebook either wasn’t sincere (doubts on the short videos vertical or because they have their own video products) or they couldn’t agree on the price.
Less than a year later, musical.ly was merged with Douyin. The international version is called Tiktok.
ByteDance put its powerful algorithm and an enormous marketing budget behind it. (More on that later)
Copy Tiktok Age
Facebook’s Lasso
Tested in countries like Mexico. Discontinued to make way for Reels. Arguably failed due to not betting enough.
Facebook’s Reels
YouTube Shorts
Netflix Fast Laughs (15-45s clips from its popular shows)
Quibi
Triller
Product Lessons
Early Days
Go With Human Nature
People want to be entertained more than they want to be educated. That’s why their EduTech startup didn’t work out and musical.ly took off.
If you want to build a new user-generated content (UGC) platform or social network, the content has to be extremely light. The content creation and consumption need to happen within seconds, not minutes.
Choosing Your Audience
They went with a young audience as the early adopters because they have the most time on their hands and have the most experience making videos. They also like to talk about cool stuff in school and on social media. Zero distribution cost.Participatory Design
Get users in early in the design process. Building with users, for users.
They have hundreds of users on musical.ly WeChat groups and they have daily conversations. Not just conversations about the product, pop culture, anything really. That way, they can be immersed in the American teen culture.
For every design, they always share the ideas and wireframes with these users before doing any coding.
Getting The First 10 Users (American teenagers no less)
It involved a lot of experimentation.
The key is to make sure there’s a very visible link that allows users to send feedback and ideas. They had a “My Ideas” button where users can submit the top 3 things they like and the top 3 things don’t like about the product. This increases product velocity.
They also spent a lot of time looking at what users created. Alex would create fake accounts to comment on people’s videos. He’d try to understand why they posted those videos, try to build empathy, and really try to understand their underlying reasoning.
Building A Community/Platform
Building An Influencer Community Is Similar To Building A Country/Economy
In the early stage, building a community from scratch is like discovering new land.
Imagine you just discovered a new land. Let’s call it America. Now you want to build an economy. You want to grow the population and you want people to migrate to your country.
Let’s say musical.ly will be America in this analogy, and YouTube/Instagram will be Europe. How do you convince creators from other regions (i.e. social platforms) to move to America (i.e. musical.ly)?
The problem with Europe (YouTube/Instagram) is that the social class is already well-established. The average citizen of Europe has almost zero opportunity to move upward in the social class. We saw an opportunity to leverage this. We will build for the average citizen in Europe.
In this new land, you have to have a centralized economy in the early days. This means that wealth distribution is accruing to a small percentage of people in your land. You make sure they succeed in building an audience and wealth. This will make them role models for the country (platform). You effectively create the American Dream. People in Europe (YouTube/Instagram) will start to realize that this normal person went to America (musical.ly) and became super-rich. Maybe I can do the same? This will lead to a lot of people migrating to your country (platform). (Baby Ariel was the example).
However, you have to decentralize the economy at the same time. Having an American Dream is great, but if it’s only a dream - people will wake up eventually and say, “It won’t work for us.”. You need to give the opportunity to the average person. You have to decentralize your traffic model. Give all users satisfaction in creating content and create a middle class.
From “Tool” To “Community”
Prerequisites
The content created by the “Tool” has to allow user creativity. (Scattered, varied content)
e.g. Prisma makes people’s photos look like artwork. It’s a great tool. But the content created has no personality. It’s the tool/algorithm’s creation, not the user’s creation.Network Effects (number of links that can be formed between nodes)
Content is to help form connections between people. You have to build in many discovery mechanisms, leveraging on any possibility to help people discover content or vice versa.
Examples of discovery mechanisms: tags, friend’s friend, geo-location, the same song, the same question.
Community operations need to create a cultural undertone that creates a sense of belonging. (More on this in No. 8)
Engagement & Network Effects
Early on, they focused on the “supernodes”, i.e. accounts with a huge following, songs that are popular.
As the network grows, you need to decentralize more and pay more attention to the medium nodes, small nodes.
At the late stage, focus on small nodes, create a distributed network.
Personalization allows users to find which nodes or clusters to go.
(Note: For more, read Eugene Wei’s essays which I’ll link to at the end)
A Community With Attitudes
“Don’t Judge Me” campaign - users wear ugly makeup and then in one swoosh, they return to their normal appearance.
Three Creeds for musical.ly
Forever Young
e.g. "@GanstaGrandma", a 86-year-old grandma with 2M+ fans.
Everybody Is Beautiful
Be Creative & Positive
Empowering Creators
Live tour in 20+ U.S. cities performed by the influencers on the musical.ly platform
TV reality shows
Partner with record labels: 1. Help record labels to sign new talents. 2. Promote record labels’ new artists
Working With Record Labels
Don’t cut into their revenue stream
Help them make more money
For musical.ly music is not the end product, it’s a raw material. In contrast, the end product on Spotify, Pandora, etc. is music.
The good thing about using music as a raw material is that they aren’t competing with other revenue streams for the labels or publishers. People will continue to stream. They aren’t hurting other revenue streams for artists either.
Having music on musical.ly actually helps artists or labels generate more revenue. As an example, Jason Derulo published his music on musical.ly. They did a campaign for him. Within 4 days, there were more than 1M videos created by the users for that sound clip. Before the song is even released, they have already generated millions of impressions for that song. That song ended up becoming very popular.
Incentives
The first thing creators look for on social media is fame. But after that, it’s not enough, they’re looking for money. That’s what will make stick around long-term.
Growth
User growth Is Exclusive
Consumers usually choose one brand.
For Snapchat to grow, it has to win over users from Facebook and Instagram. (On the other hand, imagine the pressure on Snapchat when Facebook, Instagram & Whatsapp copied “Stories”.)
Asian countries such as China and South Korea have strong domestic social networks. Western social networks can’t really go into these markets.
Scenarios, Content Formats & Different Social Needs
You want your platform to be as big as possible. To be big, it has to have enough use cases. A monotonous platform won’t become big.How to have more use cases?
Invent new ways for people to play around.
Explore different content formats.
Content Formats:
The same format can help to carry different expressions.
Images are suitable for the visual stuff, but you can also use them to share something about daily life. Images are very inclusive.
However, for videos, the suitable expressable elements, the information, will be a bit less inclusive.
For example, if you want to share some amazing food you’re having with your friend, most people would use pictures. Because the incremental value through videos is little but the cost of production is much higher. Same for the cost of consumption - people can see the picture in one glance, but it’d require at least a few seconds for videos.
Change The Value Proposition In Order To Change Your Growth Curve
They changed their value proposition from “general music video marker” to “lip sync” at the time. What they did in that release was before the users sign up, they’d first be presented some onboarding video and they’d get the idea of the most important use case - lip sync.
Secondly, they put a human-curated list of content (“the best content“) on top of the feed.
Then they made sure people invest (i.e. posting). They send a notification for them to post their first musical.ly and that dramatically improved retention.
Going from Niche to Mainstream - “Plus” Strategy (i.e. adding features)
Live Streaming
After one year of attempt, they realized that live streaming doesn’t do well in the U.S. market as did China.
Long videos
“Party” feature
Considering AR/VR
Importing User’s Contacts For Growth
It’s not all pros and no cons. It requires a balance.
Pros
Content produced by your friends gain more exposure
Cons
Interest-based discovery gets diluted. (Network Pollution)
You can get landlocked by just the content from your friends. (“Bubble“)
e.g. Weibo (interest) vs. WeChat (social graph) | Twitter (interest) vs. Facebook (social graph)
Success / Simple
Most Successful Products Are Simple. They Don’t Succeed Because They’re Simple. They’re Simple Because They Succeed.
The core product is powerful enough that it doesn’t need extra complexity.
Products that “add” become more and more complex and people usually conflate cause and effect.
e.g. WeChat has just 4 modes. (Chat, Contacts, Discover, User Account). They could naturally add Payments, but they refrain themselves from doing so. (Note: A strong directive from Allen Zhang, will cover in the future.)
Users & Markets
North American Users
North American users focus on “me”. They join social platforms to gain fans, not to watch other people. They’re like the “entrepreneurs” in social media.
e.g. musical.ly has rankings by nations. The U.S. users would change their country so they could pop up in other countries’ charts to gain more exposure. The European users would complain that the U.S. users invaded their territory.
(Possible why live streaming is less successful in the U.S.?)
(Note: The above lessons are mostly based on two interviews that Alex Zhu gave - one in English with Greylock Partners, the other one in Chinese with China’s MusicBusiness. The one with Greylock I watched on YouTube several months ago and they have since taken it down. You can find a remnant of it on this tweet or read Blake Robbins’ semi-transcription of it. Oh, found the audio of it!
Alex is probably not gonna give another useful interview anymore - it’ll just be PR as it stands.)
Further Readings
For understanding what made Tiktok successful after ByteDance’s acquisition of musical.ly, I recommend Eugene Wei’s three pieces series on Tiktok.
Abstracting away the cultural barriers with a powerful machine learning algorithm (FYP)
Committed paid acquisition for subcultures to reach a minimum viable scale
Algorithm-friendly design
Remix a remix app.
For understanding the parallel success of Douyin in China, I recommend Kelly Zhang’s (head of Douyin/ByteDance China CEO) speech, translated by Tech Buzz China.
Creator of Douyin/Tiktok: How We Created A Product with A Billion Views A Day in 18 Months: Part 1
Creator of Douyin/Tiktok: How We Created A Product with A Billion Views A Day in 18 Months: Part 2
Since you’ve scrolled all the way here, why not take 1 min to answer 3-5 questions. It’ll help me to tailor the content and write better.
Thanks for the write-up! Very insightful