Creator of Douyin / TikTok: How We Created A Product with A Billion Views A Day in 18 Months: Part I
Tech Buzz: The below speech by Kelly (Nan) Zhang, original head of Douyin (domestic China TikTok) and as of March 2020, the CEO of ByteDance’s China business, took place in late February of 2018 as part of the offline expert sharing curriculum of popular professional development company Hundun University (not to be confused with Jack Ma’s Hupan University). Kelly joined ByteDance in 2014 after the photo-sharing startup she co-founded was acquired. She was in charge of Neihanduanzi, ByteDance’s earliest product and its first success before Toutiao. Launched in early 2013, the 200 million registered users jokes and memes app was shut down permanently in April of 2018, however, for inappropriate content. It was also, coincidentally, the very first piece of China tech news that we covered here on Tech Buzz China. She has quickly become very respected in Chinese tech circles for her product acumen, the reasons for which I think will be made evident by the presentation below. In her new capacity as CEO of ByteDance China, reporting directly to Zhang Yiming, she is “responsible for the company’s products in the China market, leading product management and operation, marketing, and partnerships for Toutiao, Douyin, Xigua, Toutiao Search and other products and services.” Given that 6 of the top 10 most popular apps in China were owned by ByteDance as of the beginning of 2020 and that substantially all of its alleged $28Bn in target revenues this year will probably be from China, that’s an impressive amount of responsibility. She is also still in her (late) thirties: born in the early 1980s, there’s very little biographical information on her publicly other than that she’s from Xi’an.
Hundun University is focused on business and management, specifically innovation and entrepreneurship. You pay a few hundred dollars in annual membership fees to get access to their exclusive content, such as this talk, and to network with others of a similar mind, of course. They’re well-known in China tech circles, very active and used to lead regular tours for entrepreneurs and investors to visit Silicon Valley and other tech hubs.
Finally, just as we have done with ByteDance CEO Zhang Yiming’s extensive interview at Tsinghua University (broken into Parts 1, 2, 3, 4), we’ve taken a transcript of the video and translated and annotated / contextualized it for your learning and understanding. Not only is it a rare firsthand account of the rise of Douyin, but it also struck me as being very frank and honest. Enjoy!
PS Mega thanks to Pandaily editor Lu Zhao for the translation! You can learn more about her work here.
FULL TRANSCRIPT, PART I
Kelly: Douyin has been around for more than a year. We started it in 2016. Today we are going to talk about how to become a short video product with over 1 billion daily views. In fact, it is far more than 1 billion now. Thanks for your love for Douyin.
Tech Buzz: Douyin was officially launched in China in September of 2016, so at the time of this presentation, it was just under 18 months old. I estimate development & beta testing to have been about 6 months. This is because in past interviews it has been revealed that the idea to make a short video product was finalized on the company’s management retreat in Japan around Chinese New Year’s 2016, which would have been around February.
Kelly: Back to the second half of 2016 when we decided to enter the short video market. You probably know that 2016 was the heyday of the short video industry, and the competition was very intense. But we still decided to join this battlefield because we can see that there’s great potential here. So we thought it was worth trying. At that time, we also made a bold decision. Some of you may already know this, but we made two short video products at the same time, one called Douyin, the other was Huoshan.
TechBuzz: If you’re not familiar with Huoshan (which has since then been folded into the Douyin brand), don’t worry, we’ve got you covered on this episode of Tech Buzz, which covers all of the video products (long and short) that ByteDance has launched over the years. But the TL;DR is that Huoshan was targeted at rural China and can be considered ByteDance’s answer to Kuaishou, the leading short video app at the time and still a very strong competitor to Douyin in China. (And if this is your first time encountering Kuaishou too, don’t worry, we’ve got an episode for that too.) Huoshan actually preceded Douyin by a few months and was launched in April 2016. As mentioned above, it was renamed Douyin Huoshan in January of 2020 and has effectively become a sub-brand to Douyin. At the end of 2019, Huoshan had about 50mm DAU. By the way, ByteDance’s nickname in Chinese is Super App Factory, so having two apps in development and launching at around the same time (but targeting different demographics with a different value proposition) is totally normal for the company. And as the following years have shown, the two different demographics were really sufficiently different to need two different products.
As for the competitiveness of the market, see the slide below that shows basically everyone who is anyone in China had (has?) a horse in the race.
Kelly: As crowded as the market already was at the time, we started thinking how to find our user base, how to define our own characteristics and win the market. We struggled a lot to figure out how to make people like our products. We felt that only new and good experiences can win the hearts of users. But what is a new and good experience? We spent a very long time doing research and data analysis. We tried almost all the short video products available, both domestic and international ones. We downloaded roughly 100 products onto our phones. Everyone on our team used these short video products every day. As the saying [from Sun-tzu’s Art of War] goes, know yourself and know your enemy, and you will never be defeated. After we tried all the short video products, we realized none of them impressed us. So we thought, this may be an opportunity for us. If we wanted to do it, we had to do better than those products, we had to impress our users. I don’t know how you felt when you first used Douyin. Did you feel pleasantly surprised? We made it based on a lot of data and user feedback. Next we’ll talk about our pursuit of the ultimate user experience. How did we do it?
It’s a big topic. We were even discussing what “achieving the ultimate” means with a few friends last night. You may have heard your boss say you’re not pursuing perfection, or you are not doing work in a perfect way. I think “achieving the ultimate” is actually an attitude. If you have the attitude, you’ll have your own method of improvement. If we measure perfection on a scale from 0 to 100, while 100 should be perfect, I think everyone actually has a different understanding of what is the “ultimate”. Some people may think 60 is already the best, some may say 80, some may think 100 or even more is the ultimate. Therefore, I think the ultimate state is subjective. So next we will use Douyin as an example to talk about how we pursue perfection. I’ll show you several cases later.
In terms of this ultimate experience, Douyin focused on four keywords. The first one is full screen and high definition. You might have thought Douyin is sort of different when you first used it. Why is the video on such a big screen? Well, I don’t know what you think of full screen and high definition. But we made an experiment with many short videos and phones. We uploaded videos to different apps. Some videos were displayed in a rectangular format, some square, but at least on the domestic short video apps, they were rarely in full screen mode. But when we uploaded the same short video and viewed it in different sizes on different mobile phones, it was clear that the full-screen video impressed us the most — it was most visually impactful and immersive. We know this because we did a lot of blind testing with many people. As for high definition, we all know it costs money. To do high-definition short videos means we need to pay more for content delivery network / bandwidth costs. It was a hard decision to make but in retrospect, we think it’s 100% correct.
Tech Buzz: Most of you probably know that the full-screen functionality that Douyin “copied” was actually from another Chinese app, musical.ly. Even though it seems inconceivable to have it any other way these days, at the time, it was truly pretty novel. I mean, to get a good sense of what the competition looked like at the time, consider these 2016 screenshots of Dubsmash (Western), Miaopai, Meipai (both Chinese). All of these were very successful at the time, by the way, and all were known for surfacing viral content — usually dancing or singing — that made its creators into legitimate stars much like we see on TikTok today. But definitely in the Chinese market there were not any other full-screen competitors who were mainstream, probably because they were already doing well, so they didn’t think of “fixing” it. Anyway, despite its nifty design and success overseas, musical.ly, which Douyin was modeled after, failed to gain much traction in the Chinese market. And it was late to the game, but not that much later than Douyin. It entered what should have been its home market, China, in May 2017, eight months after Douyin’s launch, under the name “Muse.” There have been many reasons cited for its failure over the years — clunky registration process, poor choice of marketing partners, weak recommendation engine, lack of dimensionality in its content, and even faults with its generally much-admired UI that did not appeal to Chinese users. If you ask me, it was all of those things, yes, but the main thing really was that they weren’t sufficiently localized. It’s true that the musical.ly team was Chinese … but that does not mean they understood Gen Z urban Chinese youth that well. That’s something the Douyin team had to go and discover for themselves as well.
Kelly: I don’t know if you have noticed but actually the video shot via the phone’s built-in camera is in full screen. So now that we’re making a short video mobile app, why don’t we showcase videos in its original style? So we thought that may be the most suitable, though the previous standard feed system can’t present full screen very well and we had to make a new UI design.
Then it was the music. Our slogan used to be Douyin, a short music video app. People may think we’re just trying to differentiate our product from others. But actually we don’t think so. If you take a look at short videos with music and without music, you’ll feel totally different. In fact, music is more like a filter in short videos. It helps the short video become more expressive. We made two apps [Douyin and Huoshan] at the time and we wanted to make one more special and personalized while the other was more general. So the original idea of Douyin was to be personalized and expressive, so that it can be more attractive to our target users. We also conducted in-depth research on our users. For example, we found that many young people walk on the street with their headphones on, especially students. Their life is actually quite simple, going back and forth between home and school. Music keeps them company the most. Everyone has a need for music. Also, music goes very well with short videos, so we decided that full screen HD plus music is what we wanted to achieve.
Tech Buzz: Douyin did indeed position itself as a “music short video community for young people” for the first two years. Before it officially changed its slogan in March 2018 (right after this presentation) to “record your beautiful life” 记录美好生活, it was “让崇拜从这里开始，专注新生代的音乐短视频社区,” which translates roughly into “let adulation begin here, a music short video community focused on the new generation.” Pretty specific, right? Very different from the competing products at the time, whose positioning was much more ambiguous and generic. You could say this was really just copying from musical.ly, but at the time, this would have been also risky: very few consumer products worked both domestically and abroad. You could argue that Instagram might have, if it wasn’t banned, but very close (and well-funded) close clones like Niceapp failed to make it, so I don’t know that that supposition would necessarily be correct. In fact, before TikTok, it’s pretty hard to imagine a product that could bridge the China & rest-of-world divide, so differently did Chinese consumers seem to behave. Thus it was not so crazy that this seemingly obvious combination — young people, short video, music — which had achieved some success abroad, wasn’t that self-evident in China. Sure, since then I think we’ve all discovered that Gen-Z’ers globally may have more in common with each other than they do with us old(er) folk, but five years ago? Not everyone could see it that clearly. So for Kelly and her team, I can understand that this was still a “discovery” of sorts.
Kelly: And special effect filters. Well, I don’t know if you have seen the 2016 App Store ranking list. There were a lot of special effects camera apps on the list. You’d realize that Chinese people really like these special effects and filters. It’s quite different from foreign users. When we did research on many overseas products, we found that they did not pay attention to these details at all. They wouldn’t think about adding some beauty effects to make people look prettier. I do think there are differences between countries and cultures. When we decided to serve young users in China, we found that they really care about special effects and beauty filters. Who doesn’t like a prettier self? So we also put this as one of the key elements of Douyin.
Tech Buzz: What’s funny is that the concept of a filter was originally popularized by Instagram, although of course the Chinese app beauty filters do far more than change colors and contrast … they change shapes and render whole faces completely unrecognizable, ahem, I mean, more beautiful. This is, I think, a huge reason behind why China’s face-related AI is so good! Anyway, as we all know by now, even the very aggressive filters and effects did eventually make it overseas, such as this example of Chinese app Meitu’s popularity in the US in 2017. It’s true that China still takes it one step farther though, and for better or for worse, how much the filters can beautify you does still remain a main criterion for which apps users select to download.
Kelly: Lastly, personalized recommendation. I think everyone can understand this without me talking about it. Yes, personalized recommendation is indeed what Toutiao knows best. And as a user-generated-content (UGC) community, it’s hard to distribute short videos uploaded by so many users without having a personalized recommendation system. After we decided these four keywords, we started to think how to do our best in each. We’ll talk more in detail shortly.
The next step is extremely crucial: to give the product a good name.
I don’t know what you think about naming. It’s true that many companies don’t pay attention to naming. They may just come up with a few names and choose one casually. But we did massive tests on names. I remember our entire team and even the rest of the company came up with hundreds of names. Using these names, we did a lot of research and tests on users. We found that a name is very important for a product. When a user doesn’t know your product, he judges and understands your product only by its name. So the name seems very important to us. And if you are in the internet industry, you will surely know that names are very important for the conversion rate. So we spent a lot of money on testing names. I think this may be why we ended up with Douyin. In fact, Douyin was called A.me at first. A.me also a great name but it didn’t work very well in China when we promoted the product and spoke with users. After all, we’re a Chinese company, and making a product for Chinese people. Douyin was designed to serve the domestic market at first so we thought it’d be better to have a Chinese name. Though A.me sounds cool, because it means “Awesome Me” in English, it’s not intuitive enough. So we changed the name again after we launched the product for several months. We finally decided to call it Douyin.
The name of Douyin is actually quite interesting. I didn’t come up with this idea; one of our project managers on the team did. I don’t know how you understand these two words. “Dou”抖 is actually a verb [meaning “to shake”]. It gives you a vivid feeling when you add the word “Yin”音 [meaning “sound.” Together, the word “douyin” is a noun that means vibrato, actually]. In the era of A.me, we found that some users added special techniques like cool camera movements in their videos. We realized that our users spent a lot of time, sometimes even hours to produce a 15-second video, to make it cool. Users’ love for camera movements and special effects inspired our product management team. Moreover, Douyin short videos have music. So music is also crucial. I don’t know if you’ve felt this but when you hear a song you particularly like, you can’t help shaking your legs to the rhythm because you really enjoy the music, you have been conquered by this song. So we thought the word “dou” expressed love and acknowledgement of the product and decided to go with “dou”. Some companies named their products with verbs like “pat” but we thought that wasn’t good enough, not special. After we came up with Douyin we did a lot of questionnaires and user research. We found that, as we imagined, the name increased people’s acceptance and curiosity about the product. So we decided it was a good name and have been using it until today. The name issue is particularly important. It is a particularly critical step for you to reach the user, so be sure you pay attention to naming. Finally, why do we call it Douyin Short Video? Because you should not only have a name but also a category. Otherwise people still have no idea of what your product does. You don’t know what Douyin does if you just look at the name. But if you say Douyin Short Video, it’s clear to users that this is a product that shoots short videos, and you can shake and play around with it. It’s a very unique name.
Tech Buzz: And now you know … when Douyin was launched in September 2016, it was actually called A.me. It wasn’t until December that it changed its name to Douyin. I’m actually a bit surprised a more rigorous approach to naming wasn’t instituted from the very beginning, in beta testing, because the story of how ByteDance’s first product Jinri Toutiao — widely considered an “uncreative” and “too direct” name, because it literally means “Today’s Headlines” — was named is via pure A/B testing. In fact, as Zhang Yiming has alluded to in interviews, this story is so well known that the company a lot of people think ByteDance A/B tests everything, when there’s still a good amount of human judgment and intuition that goes into decision-making!
Stay tuned for Part 2 … !
This article was originally published in Tech Buzz China.