Bytedance CEO Zhang Yiming at Tsinghua University: Part 4

Heads of Meituan, Bytedance and Didi

Hey Tech Buzzers,

Welcome back to our series on ByteDance, maker of TikTok (and many other things!) and the most richly valued privately-held tech startup in the world.  

The below is a translation of a 2-hour interview on March 20, 2018, with Zhang Yiming, ByteDance founder and global CEO.  It is the last of four parts and was conducted by Professor Qian Yingyi, who was the Dean of the School of Economics and Management at Tsinghua University at the time.  For greater context and understanding, we have annotated it with our comments in italics. Part 1, Part 2, and Part 3 are here.

PS It’s been interesting to watch ByteDance take an even bigger mindshare than when we first began writing about the company in earnest a few months back.  It’s become a point of contention globally.  In late June 2020, it was one of dozens of companies banned by the Indian government.  Barely two weeks later, the US is rumored to be considering similar measures.  We don’t know what the future holds either, but as far as Chinese consumer internet companies go, it’s the first to receive this kind of attention and treatment abroad.  

PPS Mega thanks to Pandaily editor Lu Zhao for the translation! You can learn more about her work here.

Professor Qian Yingyi (L) and Zhang Yiming (R).

*BAT below refers to Baidu, Alibaba, Tencent.  TMD refers to Toutiao (ByteDance), Meituan (local services, including food delivery), Didi (ridesharing).

Qian Yingyi: The next problem is the similarities and differences between the two generations [of entrepreneurs]. Among the “BAT”, Jack Ma, Robin Li and Pony Ma were born in 1964, 1968 and 1971 respectively, and Tencent, Alibaba and Baidu were founded in 1998, 1999 and 2000. For “TMD”, Wang Xing was born in 1979, you and Cheng Wei were born in 1983; Meituan was founded a bit earlier than the rest in 2010, Toutiao and Didi were founded in 2012.

On average, the year of birth of the two groups of people is separated by 14 years, and the time of founding your companies by about 12 or 13 years. This is roughly the gap. A decade or so is not quite a new generation, but in terms of how quickly technology develops, it can be seen as two generations. Robin Li, Jack Ma and Pony Ma, two of them studied technology: Robin Li and Pony Ma. Jack Ma studied English. “TMD” is the same. You and Wang Xing [of Meituan] studied technology. Cheng Wei [of Didi] studied management.

What are your thoughts on the two generations of entrepreneurs? What are the similarities and differences between you? Of course the times are different, but please analyze and compare. 

Zhang Yiming: I have contact with them. We have learned a lot from “BAT” entrepreneurs. For example, I have read the book on Tencent. I think the similarities between us are: focus on the user experience, have users be the center, create value for users, and value our talent.

Differences: first of all, the entire domestic entrepreneurial environment around 2012 was relatively good, the internet penetration rate was high, and users adapted to our products and services quickly. Second, it was much easier for us to make and receive investment and also to recruit. I believe that things weren’t too bad in 1998 and 1999, before the dot com bubble burst, but recruitment was probably quite hard in 2001 and 2002. Nowadays internet companies are very competitive as places of employment. The college entrance test scores of the computer science department is said to be higher and higher each year because of good employment prospects. 

The TMD CEOs, from L to R: Wang Xing of Meituan, Zhang Yiming of ByteDance, and Cheng Wei of Didi.

Qian Yingyi: They [the CS department] mainly compete with our school.

Zhang Yiming: Yes. Third, there is a greater supply of talent. Fourth, the new generation of entrepreneurs are trying to globalize earlier. It wasn’t only us that attempted globalization, Meituan and Didi also did. This is because we did not start behind overseas companies; we began at almost the same time. Fifth, we are more willing to develop new businesses. The two of them (Meituan and Didi) are developing new businesses that compete with each other. We are less afraid of failure, because of this high tolerance and overall good conditions, we’re more willing to take risks in general, so we can develop business faster. Everyone can see that this group of companies founded around 2012 is growing faster.

Tech Buzz: It’s important to emphasize here that as far as Zhang Yiming is concerned, he is not “behind” any of the US tech companies that started at the same time.  He is not alone in thinking this.  While this interview took place in 2018, it had actually already become the dominant narrative as early as 2016.  By the beginning of 2017, I remember attending conferences in Beijing where Chinese entrepreneurs began their keynotes with statements like “as we all know, China leads the world in consumer internet,” a sentiment that hasn’t changed and has only deepened since then.  Also by 2016, numerous experts proclaimed that the time for Chinese apps to globalize (in Chinese: 出海, or go abroad) had come.  To their credit, lots of Chinese companies such as APUS (software) and Xiaomi (hardware) were expanding abroad very quickly and doing so successfully.  APUS in particular was founded in 2014 but was already investing in India by the following year, which was quite aggressive for the time, but a great example of the optimism that infused the industry back then. Another worthwhile highlight is Zhang Yiming’s observation that the new crop of internet companies are more willing to “develop new businesses,” even if they encroach upon the turf of an existing category leader.  If there is opportunity, there is nothing that is off limits.  We have seen this from ByteDance’s own expansion into education, gaming, and more.  As for the Meituan and Didi battle, that has been ongoing since at least 2018, and there were even suggestions of an acquisition recently, given the overlap of services offered.  Lack of focus or pragmatic opportunism?  I am now increasingly of the belief it is the latter.  What do you think?

Meituan’s nonstop expansion into more and more product categories.

Qian Yingyi: I totally agree. The six of you have close contact, but it is true that TMD leaders are different mentally. In your opinion you guys are more willing to try something new. 

Zhang Yiming: Our group of entrepreneurs are not likely to “run out of money,” since as long as the business is good, we can always find investment. The environment is better.

Tech Buzz: While there has been a steep drop in venture capital funding in China since this interview took place, that should not affect top players like ByteDance.  But is it a problem overall?  Well, if you are to believe DCM, storied global (China + Japan) venture capital fund with Silicon Valley roots and their managing partner David Chao, this is what he said upon the close of their latest $880mm fund in July 2020: while there are some pension funds pulling out of investing in China, it’s probably not going to make a dent in the money being raised.  His argument? China is still going to grow faster than the US in the next decade, so capital will continue to flow. Your thoughts? 

Qian Yingyi: Some people metaphorically refer to “BAT” as the “first half” of the Internet, and “TMD” as the “second half”. What are the essential differences between these two halves in your opinion? Can artificial intelligence be used to define the so-called post-mobile era and post-internet era? There was “BAT”, followed by “TMD”, and then there will be other “XYZ” in the future that many people sitting here today want to be part of. This is the dream of many students born after 1995 and 2000. In your opinion, using time as a marker, what are your thoughts on the first half, the second half and the future overtime game?

Zhang Yiming: My impression is that the phrase “first half, second half” came from describing the disappearance of the “user dividend.” You won’t have many new users when most people have smartphones and have installed all the apps. Using this perspective, we see that typical internet applications work like that. 

After that, you won’t be able to exploit the user dividend through simple means such as marketing, but there are still several directions you can go for breakthroughs. You can go deeper, deepen the scenario / use case, provide a deeper service with a stronger value chain for the same group of users. You can go wider, either go global or open up new dimensions, such as Mobike or ofo who put a chip on bicycles to connect them to the Internet of Things, so you can think of that as a sort of extension. And now there is the Industrial internet of things (IIOT). That’s not part of the traditional consumer internet or corporate internet as we know it, it’s an extension.

There is another very important dimension, which is what I expect from Chinese internet technology companies. In the past, we have all worked with customers, but actually the to-B business is more difficult. The databases, cloud computing, chips, or even the payment systems that C-end [consumer] products use is actually the lower level of the ICT industry. If we can enter the enterprise infrastructure upstream after we are done with consumers, the entire Chinese technology industry will have been upgraded.

Tech Buzz: The “user dividend” (Chinese: 用户红利) pops up often in speeches by Chinese internet entrepreneurs, like this one by Wang Xing back in 2016. It’s a similar idea to the concept of “demographic dividend” (Chinese: 人口红利) that explained so much of China’s astounding growth in the past two decades, including its internet sector. (Basically, there are more young people than old, resulting in a boost in the economy because there are more folks working than there are not.)  You’ll see this analogy of “dividend” applied to all sorts of other internet concepts too, like “time spent dividend” (时长红利) and “traffic dividend” (流量红利).  In my opinion, this is an example of the Chinese mindset, where the environment matters as much as the product.  In the West, we talk about the entrepreneur’s biggest challenge as “solving a real problem,” but in China, it’s just as important that the founder is [also] taking advantage of “favorable macro conditions.”  Dividends are shorthand for these favorable tailwinds.  

Using this perspective, it should not surprise you ByteDance’s expansions into various other sectors.  One goes where the dividends are.  Especially when it comes to B2B products, which is a well-known growth opportunity in China.  No wonder that Bytedance has not only launched Lark and other productivity tools, but also a whole suite of enterprise products in its recently unveiled Volcano Engine.  I would not expect it to stop there. The (productivity? automation?) dividends are calling!

ByteDance unveiled its cloud services platform called “Volcano Engine” officially in July 2020.

Qian Yingyi: Compared to Chinese companies, companies in other countries still have an advantage in the to-B [enterprise software] business. 

Zhang Yiming: Amazon excels in cloud computing, and Oracle and Salesforce do a good job as well in the to-B business. No matter whether it is to obtain the so-called user dividend, for marketing or social growth purposes, becoming more globalized will help in going more upstream and doing the more difficult work there.

Qian Yingyi: Upstream is deep technology, both of these are difficult. Or maybe you go beyond the traditional internet, which is also hardcore. From a broader perspective, the two countries with the most unicorn companies in the world are the United States in first place followed by China. If “BAT” was behind American companies 10 years ago, today “BAT” is up to par with those leading American companies, and “TMD” is even ahead of American companies that started in the same period.

The valuation of the three “TMD” companies might be even higher than the valuation of the top three unlisted unicorn companies in the United States. The valuation of the three “TMD” companies is now $100 billion, higher than Uber, Airbnb plus any other company in the U.S. What do you think of this? On the other hand, we also have our own characteristics and weaknesses. What do you think of that?

Zhang Yiming: We can’t say that we have too big of an advantage. Take ride hailing as an example, many countries have localized the services using regulations. If you look at single markets, yes, China has a large population, the overall growth of the economy is fast, the value of each customer is high, and China does not have so many car owners.

Airbnb has done a great job in globalization, but it is a relatively slow business. To eventually achieve greater results, a competitor needs to make a breakthrough either in the value chain or globalization. In the same few years, we haven’t surpassed them but we have at least achieved equal progress. A few years ago, “BAT” was basically one fifth of the market value of comparable American companies. Baidu’s listing price was about one fifth of Google’s market value, and this was considered quite high at the time. Now China’s entire economy has improved, as has the basic infrastructure, so these are all helpful factors. 

Tech Buzz: I don’t know how Robin Li will feel reading this portion of the interview but as of July 2020, Baidu’s market capitalization is only about 4% of Google’s ($42Bn vs $1Trn). Alibaba is at $480Bn, while Tencent is at $630Bn, versus Amazon at $1.4Trn and Facebook at $670Bn, two companies that are often cited as (very) rough Western analogues. Just using market capitalization as a proxy, it looks like BAT is about one-third now of Google / Facebook / Amazon. This is a highly flawed measure, but nonetheless one data point that is often used, and so one I thought worth updating.

Similarly, the TMD also has one notable underperformer, Didi, which has fallen by more than one-third from its peak.  As mentioned earlier, it’s even considered an acquisition target by Meituan, who is now trading at around $160Bn.  The main protagonist of our story, ByteDance, however, is rumored to be worth at least $100Bn.  While the US still exceeds China in terms of number of unicorns, its most valuable startups have fallen in value, not grown, such as Airbnb and the finally public Uber.  For this most recent batch of top post-2010 internet companies, China has done significantly better than be “one-fifth” of the value.  It will be interesting to see what the next 10 years brings. 

Part 4 (of 4) Youth is more important than maturity, don’t take a negative and critical attitude towards new things

Qian Yingyi: So far you’ve objectively analyzed globalization and deep tech companies. The last few questions are related to personal growth and insights.

I read a report about you once and was very impressed. You like to describe concepts using coordinates and matrices. You think that mathematics is the description of the basic relationship between material things. Even the advertising profit model and entrepreneurial strategies of ByteDance you have a set of sophisticated algorithms. Mathematics is a way of thinking. I used to study mathematics. What role does mathematics play in your way of thinking and working? We attach great importance to mathematics here, but students can’t see its usefulness, so please introduce what that might be.

Zhang Yiming: I couldn’t see it either when I was a student (laughs). Mathematics is important, but at the same time I think language is also very important. Both language and mathematics are tools for describing things. Sometimes things are suitable to be described by mathematics, and sometimes they are suitable to be described by language. Language and mathematics are both operations. For example, legal issues also have causal analysis and nested concepts, which is very similar to coding or mathematical proofs. When describing things using language, different adjectives can also be mapped with coordinates.

However, compared with mathematics, the description of language is more difficult to quantify. Many problems can not be solved well if they cannot be described accurately, so mathematics is needed. Take our recommendation system as an example, we make a user profile for each user. Many people think we describe each user with a lot of labels and words, but in fact, we regard profiling as a vector, they are relationships in a space. (Mathematics) is important in describing the advertising system, including the variety of descriptive content, the degree of convergence, and the degree of generalization, which all need to be quantified before they can be improved.

Qian Yingyi: You are born in the 80s, a very young successful entrepreneur. Many users of your product are even younger people, like Wukong Q&A, Xigua Video and so on. How do you make sure people born after 1990, 2000 or even 2010 like your products?

Zhang Yiming: This is actually quite challenging. I attach great importance to this matter myself. You said that user demand is changing and the product is evolving, so it is very important to stay young, our products are used by mostly young people.

Qian Yingyi: We’re all growing old, how to stay young?

Zhang Yiming: Spend more time on Douyin (laughs). We have an all-hands staff meeting every two months, and an employee once asked, how does the company become more mature? My thinking is the opposite, we should always stay young and don’t take a negative and critical attitude towards new things. There must be a reason for the emergence of new things. We must experience it, try it, and observe its development. We have many requirements for the team. Very early on, I found that the management team didn’t use Douyin and I was very anxious. I told them to shoot two short videos every month, they had to get a certain number of likes, as a way to force everyone to stay young. Our company’s corporate culture is to always be entrepreneurial. Similar to Amazon’s Always Day 1, we should always think like we do on the first day of the company, always wondering about what users are thinking.

Tech Buzz: Amazon (along with Nike, Netflix, Facebook, Google, and a host of other leading businesses globally) is one of the companies Zhang Yiming looks up to and “copies” from.  When I visited their Silicon Valley office, foil balloons in the shape of the letters and numbers that make up “Day 1” greeted me at the door.  The company unabashedly takes great ideas from others and adopts them as their own.  With attribution, too.  Not only are employees treated to posters proclaiming “Day 1” in the office (and sometimes bathrooms, too), they are told it sprouts from the wise mind of Jeff Bezos and are encouraged to read The Everything Store

A version of the Day 1 poster that hung in ByteDance offices in China. As you can see, The Everything Store is a recommended read at the company.

Qian Yingyi: These words are the best publicity for Douyin. For sure the number of users will increase greatly; everyone wants to stay young. The following question is from my point of view. I teach at a university. What do you think college students should do in order to meet the needs of future technology companies, especially when it comes to training? There are also some teachers here in the audience.

Zhang Yiming: As an employer, we prefer people who can learn multiple skills and abilities. For example, many of our product managers were engineers, some of them were designers. The leader of the human resources department studied electronics, and the leader of the administration department studied computer science. Your major doesn’t matter that much. What is more important is the ability to learn a variety of knowledge, and maintaining that ability to learn may be more important than accumulating knowledge. Now you can get any information from the internet, so I think your ability to organize your knowledge structure and update it may be more important.

Tech Buzz: In countries like the US, this seems obvious: what you major in often doesn’t have much to do with your ultimate choice of vocation or employability.  But in China, there is still a lot of inertia behind the thinking that your field of study dictates your future career prospects.  Although this mindset is quickly receding, it continues to exist. 

Qian Yingyi: When you were in college, you were interested in biology. You studied electronics, computer science, and business administration. You also liked reading biographies.

Zhang Yiming: If I had the opportunity to learn again, I would learn more.

Qian Yingyi: Tell me two things you would want to learn.

Zhang Yiming: I want to learn more mathematics.

Qian Yingyi: Math is still valuable. Anything else?

Zhang Yiming: I want to learn more physics.

Qian Yingyi: We just opened a new physics course called a brief history of physics. When he was here, [Elon] Musk said that he double majored in physics in addition to business administration, which makes sense. Another question that many people wonder, there was no mobile internet when you were a student, and you just said that you read a lot of books systematically at the time. Now that everyone is reading pieces of content on the mobile Internet, what are your suggestions for students? In this case, everyone only reads some short messages, articles, and 15-second Douyin. Should we still systematically read paper books or e-books? Do you have any suggestions?

Zhang Yiming: I think Toutiao, Douyin, Wukong Q&A, or Xigua Video can all enrich and expand your horizons, and give you a lot of information.  However, I still think you have to read books, especially textbooks if you want to build up a basic knowledge structure.

Qian Yingyi: You mentioned two types of books. One is biographies. Tell us why we need to read textbooks. Textbooks are systematic, but may seem boring. Why do we need to read textbooks?

Zhang Yiming: Textbooks have the most concentrated and refined form of human knowledge. Popular science books can serve as general interest reading in certain fields, but in terms of concentrated knowledge, textbooks give you deeper insights. At school, it is still very important for teachers to supervise the study of textbooks.

Qian Yingyi: There is nothing else that can replace textbooks. It is the most systematic, structured, rigorous and refined type of book, relatively speaking. We’ve touched upon a lot of questions. The last question, what do you think is the most important thing for ByteDance in 2018? What is the most critical problem you have to solve as CEO in 2018?

Zhang Yiming: There are three things that are the most important for our company in our seventh year. For the company itself, I hope to continue to improve the company, including the management, culture and systems. As for business, I hope to go global. If (globalization) can succeed, it will be a milestone for us and a landmark for Chinese internet companies. Third, as a member of a society, a platform-based company, we hope to take on more corporate social responsibilities. In the past, we were more concerned about the growth of the business. Now we hope to put forward higher requirements for the management team and business leaders. We should not only care about business and globalization, but also have a broader perspective to understand and assume more social responsibilities.

Tech Buzz: Fast forward two years from the date of this interview, and that’s exactly what Zhang Yiming has done.  While he remains CEO of ByteDance, by creating new leadership roles for the company’s China business, he’s freed himself up to focus on global expansion and new initiatives, which is exactly what he has been saying is important to him all along.  His focus on upgrading the management and culture of the company also remains consistent.  Both of these points were emphasized in his letter to employees earlier this year on the eve of the company’s 8th anniversary in March 2020.  

Qian Yingyi: Business, globalization, corporate social responsibilities. There is still some time left, let’s have the students ask questions. 

Question: I am studying economics at the School of Economics and Management as my second degree. Many alumni in our department became product managers after graduation. Speaking of how to make products, I read a lot of articles that many think Toutiao [ByteDance] is a company relying heavily on data such as A/B testing for judgment. As a new generation product manager, is it better to judge based on data than your own intuition?

Zhang Yiming: This is a big misunderstanding of our company, a very big misunderstanding. We have always emphasized within the company that we need to communicate directly with users, get first-hand experience, and build perceptual knowledge. We think this is just as important as A/B testing. We are opposed to going from one logic to another, because that is easy to make big mistakes. First-hand experience can help us stay close to the truth. We are willing to bring the user’s feelings and experience to the company. The first thing we did when globalizing is to get feedback from overseas users. We provided contact information so that users can directly contact us.

Perceptual knowledge is very important. A/B tests cannot measure long-term goals. For many important judgments, especially in the early stages of the product, product intuition is very very important. You must have empathy with users to understand their feelings and do so through user interviews. As a product manager, the most important quality is empathy, not A/B tests.

Tech Buzz: This jives with what I’ve been able to glean from my research, but the impression of ByteDance as an only data reliant company persists. Part of it is because of the oft-repeated stories about ByteDance’s early days.  Why was the product called Jinri Toutiao? Because it performed the best in A/B tests.  How did the company decide to go after video in 2016, even though the field was so crowded and competitors so well-funded? Because they added a video option to Toutiao as an experiment and saw that users were spending up to half of their time there. 

Question: I am a class of 2021 undergraduate from Tsinghua SEM. We want to start a business, how can we prepare during our four years of college? As finance and management students, what other skills should we master in-depth other than our major in order to improve our future competitiveness?

Zhang Yiming: On starting a business, we must keep an eye on society and industry. Interest is a good mentor. The most important thing in starting a business is to find your direction. To be interested in things, you should read more news and learn more about other companies. 

I don’t have a good answer for the second question. (I think) It is very important to understand an industry, no matter what you do in the future.