Before I get into this article… I, tongue in cheek, talk about a plausible direction where TikTok and ByteDance might head. First, The Washington Post, next- our food. I do this to get all the jokes out of me before I ruin this article with a meme.
It remains without saying how prolific TikTok/Douyin has become to all audiences, east, and west. It began as a startup founded by Alex Zhu and Luyu Yang intent to make an educational video platform. It didn’t precisely disrupt education. Over three years, the team was able to take advantage of the mobile wave while growing to 200 Million users internationally, reporting 60 Million Monthly Active Users by August 2017. In this article, we’re going to talk about how a pivot and very intentional thinking led to one of the largest social media apps of this decade.
Alex Zhu and Luyu Yang met while working as Directors of Product at eBaoTech, an insurtech company in Shanghai. Luyu Yang was a serial founder before teaming up with Alex. According to his LinkedIn, he founded 2 companies. He also co-founded one consultancy, before all that co-founded one while studying at Central South University in Changsha, Hunan, China. Alex Zhu is a designer by trade, working as a designer for Chinapages, Cisco WebEx, and SAP before meeting Luyu. After briefly coming back at SAP, Alex becomes enamored by the idea of the future of education and MOOCs around 2012. He tells longtime friend Luyu about the concept, and they get to work- founding Cicada Education in Shanghai.
The idea was to make a platform for short-form educational videos. After raising $250,000 from China Rock Capital Management and a quick sprint of six months. The team had difficulty finding traction trying to make engaging educational content that can fit with the expectations of a 3 to 5-minute video. The problem: educators had trouble adjusting their lesson plans in bite-sized chunks. The videos took too long to make, and the target demographic of teens didn’t engage anyways.
One time on a train in Mountain View. Alex observed young people recording themselves with the front-facing camera, some were lip-syncing. This got the gears going, they realized that it was tough to create functional, short videos. A problem they encountered with Cicada.
Around 2014, Vine showed that many content creators had to resort to custom video creation stacks to create high-quality videos. Musical.ly’s founders posed a question after going through their iterations… What if music could democratize high-quality short video creation?
They decided to take whatever money they had left and put it to what is now known as Musical.ly.
Due to the founders initially designing for the original demographic of 13–18-year-olds, most of their research about the audience still applied. This was a demographic that indexed on over-sharing. This allowed them to keep acquisition costs low while building a feature set that favored discoverability.
In 30 days, the first prototype of Musical.ly was born.
However, in China, the app didn’t catch on due to a demographic mismatch. Most internet users were slightly older, and the app ecosystem in China was not yet as mature as Baidu and WeChat still held dominance. Because of this, they were initially rebuffed by investors who unable to see other case studies of Chinese apps gaining a foothold abroad. The company had to look towards lower-cost marketing funnels. Luyu credits the rejections for their frugality and effective acquisition funnels.
Gaining a Foothold
Despite as appealing as the idea sounded, this idea would not enter a blue ocean in the western market. Applications like Dubsmash and Triller were making waves. Even before, apps like Mindle and Eyegroove were startup incumbents. Triller was founded by veterans of the music industry, and Mindle was a product influence of Vine. Dubsmash was pivoting to short-form video after unable to breakthrough as an app that was supposed to help make music videos. Needless to say, there was a Facebook-owned Instagram to be worried about. The founding team viewed this as validation. The team used those applications as a baseline.
The founder’s optimized for one thing, and that was being intentional about building a community. The app shipped with hashtags to try to spur one. Users could post videos with any of those hashtags and follow them to watch specific videos of a particular topic or genre. Alex Zhu likened this process to founding a country. There was an understanding that in the short term, that the app would be a tool for video creation, but they had larger ambitions. The hashtags gave them a foothold to become a platform. The team thought of the environment as an ecosystem where consumers and producers were equal parts to the whole of the application. This led them to be mindful of their feature investment.
They eventually did an official launch of the application in July 2014. It was a hit with 500 downloads on the first day. The next day, with another 500 downloads. It resonated in the app store. The founders wondered why. They deduced that it was because of a cultural phenomenon happening in the U.S. At the time, Spike.TV, a now-defunct television channel, aired “Lip Sync Battle” and suddenly searches for “lip-sync” on the app store kept on showing the app. Curious teens downloaded and tried it out. This was enough for them to get into the “Photo and Video” charts for the iOS app store.
The founders realized that most of all, the users kept coming back. Quickly, they added features like duets where two users can collaborate on a video. By adding demographic-specific social features like BFFs also helped with trying to build said community.
Throughout the 10 months, however, the team was worried, short on cash they couldn’t engage in any ad spend. But in this period proved to be formative, they were able to build out the company and build out the application to maturity. The user base was slowly growing but not to the scale that could have saved the company. Right on time, the company was ready to unveil a new design out into production. They previously invited their early users into Messenger and WhatsApp group-chats with other users so they can directly get feedback on the new designs. They were able to incorporate its tight feedback loop into a redesign that would prove fateful for the company.
Initially, the app had a share feature. But the company noticed that users kept on trying to crop the watermark from the video. They saw that some users would make videos to similar topics, and to spur new video creation, they added challenges to increase content diversity.
Since teens didn’t need skeuomorphic design, simplifying the interface would shorten time-to-action to make the product stickier.
They revamped the home screen to create a more welcoming interface.
But the key was the new video watermark. As soon as that launched…
It would move from the photo and video charts and finally enter the US overall app charts. Despite no significant technology innovation, Musical.ly would innovate and break conventions on the business of video.
Upon breaking out on the Top 40 charts in 2015, they were hovering around 1,000,000 Monthly Active Users. This was enough validation for them to close around $17 Million in funding in a Series B. Featuring Greylock, GGV, CRCM, and Morningside as notable investors. At this time, high-quality applications were starting to proliferate in the Photo and Video categories in the app store. As mentioned, Musical.ly shipped video challenges that operate very similar to topics on Twitter. Around the time when teenagers are out of school, one topic was trending. That was the “don’t judge me” hashtag. Where users would purposely obfuscate their appearance to look “ugly” and then reveal themselves.
Instead of trying to pivot away, noticing that users were tuning in to these “channels”. The company leaned into this. This assisted with the growth of the app. The same demographic that got Snapchat to its lofty heights were using Musical.ly beyond lyric videos. The critical thing about the app’s design was the fact that the video would auto-play. If you wanted another video, all you had to do was swipe up for the next video.
The team built creative tools that enhanced its network effects. They shipped a leader-board, and community challenges. To help with safety and protecting their users, they updated their safety standards. This only served to deepen their content ecosystem. The challenges being posted to other platforms was a testament to its influence beyond the app. After that moment, it never left the Top 40 charts in the Play Store or the iOS App Store.
This also attracted a new demo of users that initially wasn’t in their target. Still, they started expanding the type of content on the platform. They made sure they turned into evangelists and considered their needs on the roadmap while catering to them over time. Musical.ly later started laying the foundation for a creator’s program and partnerships with brands to help bring in ad revenue.
A Chance At Stardom
New and growing platforms also tended to attract those looking for an audience. As the leader-board started to become critical, this social game started drawing a new generation of teen influencers. Luckily for them, Musical.ly, along with other apps in the genre, had and maintained robust sharing integrations with other social networks. This easy video export allowed the application to take advantage of an organic growth method. Another event happened (wherein my opinion) led to a new exodus of users from other platforms. Vine was already dead but announcing it made users conscious about a 6-second sized hole in their lives.
The largest repository of teen marketers was left to languish. Many already went to Instagram. Many tried to pivot to YouTube. The smart ones went to Musical.ly. Because they got the cheapest customer acquisition funnel on the planet. The YouTube video compilation.
The export native behavior of Vine users was to export en masse short 6 second videos into genre videos on YouTube. Vine meme compilations titled, “Best Vine Memes 2014” were some of the best sources of content on YouTube because it’s algorithm favored personalities and long vlogs. Meaning, YouTube snuffed out any independent creator who wanted to create short form content but enabled those who wanted to aggregate content. This already mirrored the challenge behavior on Musical.ly.
Those who went to Instagram had their videos and content behind a walled garden. Their fans were happy but, it made any new discovery of fans subject to IG’s algorithms. Around this time, Facebook was ramping up monetization and focused on the war against Snapchat’s Stories. The feed would be de-prioritized. Vine influencers who moved to Musical.ly got free press on the new platform because their content was shared elsewhere, even if they didn’t share it themselves.
A New Proposition
As the app started to grow into the youth-focused video platform of that time. The team realized to foster growth, they needed to move beyond its demographic. At the time, it was assumed that the app was a social network. However, when they dug deeper into the metrics. It wasn’t a social network, it was a content platform. The user’s session lengths lasted, on average, up to an hour. The founder’s credit this discovery to the consistent discussions they had with users. IG and Snapchat were social-graph led, and after user interviews, users loved staying on the app to watch videos. Because it wasn’t just another social media app, it meant that either purposely or by accident- it became TV for the smartphone generation. To age up, they needed to broaden the content offering.
Now, all described above seemed like table-stakes to western applications. The genuine differentiating factor was thinking about the content ecosystem in economic terms.
Building a community, especially from scratch, is like setting up a country.
In the beginning, you have to allocate all your resources to a very limited number of users. For example, now that you have discovered a new continent, I hope that people in Europe can move to the New World and build a country on your continent. What do you do? When there are few people on this island and there is little GDP, if you distribute them evenly, everyone will actually live a miserable way, so no one wants to move from Europe. So, at the beginning you need to let This group has a very high Gini (co-efficient), which allocates all GDP to a small number of people. The GDP here is actually the flow, which is distributed to a small number of users, so that these users can get rich first.
The app, initially, was re-distributing views on purpose to a small selected number of users to spur on the social game. Since Musical.ly wasn’t limited by a social graph, the flow of attention was a key product factor behind its decision to inform it’s algorithms. The promise of viewership being continuously fulfilled instilled confidence in budding users within the app.
A Unique Advantage
At this point in 2016, the team was busy trying to age up the audience. They looked into camera features and product ideas to try to expand the base. However, from the initial thesis on views. They realized it was all in the algorithms. The Follow feed was sunset. All the content was already in the app, but they just needed to showcase it better. By removing user-driven personalization, content diversity grew once the app showcased what was already on the platform.
Around this time, they still were aggressively pushing opportunities to expand their user acquisition funnel. The company started selling shirts on Amazon, the user base began throwing parties (independently of the company I may add). The users took to calling themselves “musers.” Best of all, by having the app feature videos beyond the usual user palette. The app stayed relevant to its user base by always being in the cultural moment.
At this time, Facebook acquires Musical.ly competitor Eyegroove in 2016 but to be used for a different roadmap… The battle for AR and camera. Snap and Facebook were too busy at each other throats to realize that Musical.ly was about to push their advantage.
With the new round of funding, they ramped up hiring and partnership development. Musical.ly hired veteran executives from media companies in Los Angeles. They created best in class mobile video editing tools and effects. This product direction scaled beyond ephemeral messages and group chats that led them to the next milestone. At this time, they were able to reach 60 Million users with the Chinese market. But to be clear, they had around 10 Million American youths on the app. A very sizable amount of teens.
The user numbers allowed them to secure better-licensing deals with music labels to feature their music on the app. Eyegroove was already was running on a similar thesis. Still, FB instead applied the technology to its closest competing offering, Instagram fueling the camera effects instead during its storyification of its platforms. Meanwhile, FB did have a compelling thesis for becoming a content platform with Facebook Live and Facebook Video. However, a few distractions arose. The company was dealing with the fallout of Cambridge Analytica and public perception issues around 2016. Moreso, the content on those platforms weren’t compelling and didn’t take advantage of the organic nature of people who wanted to create content. Even more so, content creators were heavily moderated and would get their accounts removed if they didn’t use “pages” properly.
However, the success of Musical.ly wasn’t blind to Facebook. Their success led to high-level discussions with Kevin Systrom, co-founder and then executive of Instagram. But Facebook later passed on two concerns, aging up the audience (something the founders were already concerned about) and complications related to Chinese ownership.
Besides, Facebook, despite all efforts, couldn’t take advantage of how to drive growth outside of the usual set of brand name efforts that it was pushing for Facebook Watch. Mainly because Facebook was also heavily investing in video and was heavily prioritizing videos on its feed over photos and static content. This affected the amount of content that was shared on the wall. Not to mention the number of bots and lack of robust moderation. Many pages on Facebook resorted to posting memes and content like videos, even though they, in fact, were images. This affected the social graph. However, young people couldn’t download videos and share them. More so, the “pivot to video” burned established content creators and publications where brands didn’t want to work with Facebook anymore.
It goes without saying, Musical.ly advertised heavily on online social media platforms. Using Google Ads to help drive growth to age up. Quite notable were their ads on YouTube.
The ads were poorly received by the more “ironic” corners of the internet. They signed up in droves to bully the existing content creators by “dueting” responses to them. A feature initially intended to have two users sing together in a “multiplayer co-op” like fashion to the same song- which turned into a creative way to make jokes. (And abuse other users.)
This gave rise to a new genre of video on the platform. Users borderline bullying users who were dancing or posting “cringe” made the platform for a source of humor that later leaked on to Twitter and the stories of users on other platforms. This was one small part of taking them to 40 Million Monthly Active Users (MAU). The other part was signing partnerships with NBC, Hearst, and other big names to start getting content produced for the app.
Meanwhile, in China, ByteDance’s video product Xigua Video released in July 2016 got traction but was unable to achieve greater success in China versus Kuaishou (2011), Miaopai (2012), and Haokan Shepin (2017?).
Each of those applications had different communities and different user bases. Those user groups dictated the focus of the application. Leading to those apps having different types of content within the mainland. Kuaishou was the incumbent at the time, with a $100 Million investment from Tencent. It had wealthy and young users that were a boon to advertisers and luxury brands.
ByteDance later funded a similar app called a.me. It was later rebranded and folded into ByteDance as Douyin. (Literally: vibration/vibrato) It was very similar, if not an outright copy, with all of the features of Musical.ly. The design was sharp and contemporary that was able to win young, affluent users in China. It was hard for a random person to build a following on Kuaishou, so users chasing exposure and a fast-growing video platform flocked to Douyin. In a year, they were able to get to 100 million users in the mainland using the same tactics as Musical.ly without a specific focus on lip-syncing.
An Exit and The Next Set of Users
In late 2017, ByteDance’s ambitions later led to the western market. After securing $2 Billion of funding at a $20 Billion valuation, they were on a corporate purchasing spree. They purchased a News startup Flipagram. It was rare at the time to see a fourth Chinese super company rise from the three mainstays that were Alibaba, Tencent, and Baidu. Douyin quickly grew to 100 Million users in China in 2 years, effectively winning their share of the Chinese market. ByteDance determined that the best way to become entrenched in the market is to expand nearby to local markets such as Japan, South Korea, and Singapore.
According to the press release from ByteDance, they left the Musical.ly team on their own to run the app. But they were to take advantage of ByteDance’s bread and butter while getting all the music licensing deals thanks to Musical.ly’s presence in Los Angeles. ByteDance’s flagship application Toutiao (Headlines) was an app that suggested News articles based on the user’s interests. This was a company that had 6 years of experience building recommendation systems with one of the most enormous data-sets possible. The company later announced it would operate under the original name in western markets.
Then, they rebranded. ByteDance adapted the Douyin logo but applied a new name. It would be called TikTok. The new owners would change almost nothing about the design. (For the record, Douyin already looked very similar) They would remove the Your City and the Follow feed into a For You feed. The idea was to remove the barriers to getting new users up to speed. They eliminated on-boarding registration, the leader-board was removed, and they acquired additional content partnerships. Smart growth hacking was also applied, the app would hide the clock to make users forget about how much time they spent on the app. (I haven’t been able to confirm this independently, they allegedly paid content creators on YouTube to share “Dank TikTok Meme Compilations”)
The rebrand was successful- users didn’t associate the app with teen lip-synching videos anymore. TikTok would assume the role of a new content platform that would have far-reaching effects on western/eastern content consumption.
Out of their LA office, users that were getting substantial traction would be signed to agents. The heavy usage of the application from teens (the demographic with the freest time) would start to have direct influence over the music charts. Songs that were featured on the app would start to make or break songs. Artists like Lil Nas X and iLoveFriday, who at the time were relatively obscure, would have their songs shared thousands of times before they hit the mainstream. A&Rs would start to make accounts to try to find the artists that would eventually be the stars of tomorrow. The rapid growth in and of its self became a boon for publications documenting the rise of TikTok also exposed TikTok to a curious set of adults who wanted to try out the application.
The revamped TikTok app benefited from the corporate structure of ByteDance, who were more than willing to double down on the existing strategy. The app promoted existing creators but more than ready to elevate new creators rather than serve incumbents. (An issue YouTube faced with its fraught relationship with its creators) You had niche creators who were baking cookies, and those who made retainers get famous. Users in its creator’s program would not only get their videos promoted but receive advanced notice of challenges and video ideas from the company. Machine Learning techniques were applied to classify items and the form of videos to determine how each video is similar to each other to pick candidates to play next.
It also didn’t help that western incumbents were busy trying to harvest and monetize existing users rather than try to chase more growth abroad. A pressure faced by companies who IPOed where shareholders were pressuring for profitability. To make matters worse, Snap Inc. launched a botched update that affected it’s perception and would later haunt its stock price.
The news hasn’t been favorable for “The Blue App.” eMarketer reported 16-year-old Americans are now less likely to use Facebook than 60-year-olds. 60% of the teenagers 12–17 reported using Facebook in 2015. Now in 2019, only 39% of teenagers in the previously mentioned age group use Facebook. While Instagram TV hasn’t really hit the same cultural nerve that TikTok has been cultivating.
TikTok managed to build a better content funnel that captured people looking for fresh and funny content.
Open Questions and My Takeaways
TikTok is a success case for many Chinese companies that have international ambitions. Already, the video app Kuaishou has an international version called Kwai that just hit the 3 Million user milestone in Brazil. Facebook is expecting to respond with Lasso and a newer app with the same premise of TikTok. On Instagram, it’s rumored to add a “Popular Photos” product with an infinite photo feed. Other applications are on the same premise but targeting different age groups like Firework. Much was said about Snapchat’s ascendant rise until the market’s sobered up, will that also apply to ByteDance?
With TV and other high production value content, there is explicit fragmentation seen in the “Streaming Wars.” Will short video platforms follow that route? Or is there is a new aggregator that will take all the market share like graph-based social networks? TikTok and its parent company will soon answer if its a fad or if it’s here to stay. Already it is undeniable that it will influence the next generation of products to come as U.S. regulators turn towards big tech with greater scrutiny.
At the time of this writing, both of the founders of Musical.ly are still with ByteDance. In contrast, their efforts and lessons are out in the wild for the founders of today to apply.
- Most breakthroughs are unique executions from existing tech.
- Your users will surprise you. Meet their needs and think about the next set of users.
- Always talk to your users. Build tight feedback loops to respond to feedback.
- The app didn’t change much in terms of design, new features to get Product/Market fit usually aren’t the answer. It is building an ecosystem where everything fits together.
- Be willing to sunset features that might be favorable today, but enable a better tomorrow.
- Don’t be afraid to pivot.
I’m sure this will be a fascinating space to watch, I await the next wave of youth that define future cultural moments. A new set of tastes is sure to turn everything on its head once again.
If you enjoyed this article, please like and subscribe, send me hate mail at my DMs on Twitter (link below) and be sure to thank your server the next time you go out to eat at Applebee’s. What is that button you see on the right? That validates me, so press and hold it as long as you can.
Hailing from Miami, Florida — Angelo Saraceno holds a Bachelor’s degree in Computer Science from Florida International University. He is a Product Manager at an enterprise software company. He spent his time as an undergraduate working and studying entrepreneurship circles, and traveling the country, learning about what it takes to build successful teams and businesses. In his spare time, Angelo enjoys salsa dancing and thinks soccer is the most beautiful thing on this Earth. You can find him on Twitter @ndneighbor.
Special thanks to Tae Hong-Min for the conversation around YouTube and algorithms. Thanks to Raven Mo for providing cultural context. Shout out to Andrea “Hella” Chen and Jessica Zhou for editing and reviewing this article. Thanks to Yuan Wang for finding more typos.