字节跳动七周年演讲

ByteDance 7th Anniversary Speech

> By Zhang Yiming · internal speech at ByteDance's 7th anniversary meeting · first published on ByteDance's official WeChat account 字节范儿 · 2019-03-14 · [Original link](https://mp.weixin.qq.com/s/gRoBPD6GaFvm7OSs-PBQOQ) ([archived](https://web.archive.org/web/20200327193650/https://mp.weixin.qq.com/s/gRoBPD6GaFvm7OSs-PBQOQ))

Welcome, colleagues from every country, gathered in this small lecture hall for our annual meeting. In 2018 we achieved a lot, and also went through quite a few setbacks and difficulties, and many colleagues worked very hard for it. Thank you all! The past year was an extraordinary year not only for our company but for the whole industry, and even the whole world. In 2015 we held our annual meeting in Okinawa with the theme "An Era of Great Change". At the time we felt that ever-developing technology would bring big changes to society and the world, but we didn't foresee that when technology brought huge change, it would also provoke a huge counterforce, and that tech companies would go through many challenges in the process.

In the past, I've stood here and shared with you the company's business and the industry landscape; I don't want to talk about any of that today. To prepare for today's talk, last night I went to a place to look for inspiration, and I'll take you to see it too.

It really brought back a lot of memories. In many interviews over the years I've often been asked whether building the company was hard. I don't think it was. Although we were crammed into a small residential flat with simple furniture, in our memory those days were pretty happy: it was a new beginning, every day was full of hope, and most importantly we were creating something valuable. Even without the hardship of living in a basement or a story of ups and downs, I still recalled quite a few things worth sharing with you. So today I'll mainly tell a few stories.

Space has a shape; dreams have no limit

Everyone knows we got our first funding in a little café near here. Actually two other rounds also happened in this flat. About half a year after the company was founded, an American entrepreneur visiting Beijing as a tourist came to see the company through a friend's introduction and talked a lot about product and technology. He was amazed and said the technology of the team in this apartment was on a par with Silicon Valley; could he make an investment? Later he became our investor.

At the end of 2012, right in Jinqiu Jiayuan, we started discussing internationalization. When we chose the name "字节跳动", we also settled on the English name ByteDance. At the time hardly any of our colleagues had been abroad. But we felt the opportunity mobile internet brought us existed all over the world.

> [Note: ByteDance's early office was at 「锦秋家园」 (Jinqiu Jiayuan) on Zhichun Road, Beijing. The Huxiu and Jiemian reprints of the 2015 annual meeting speech (03) have 「锦绣家园」, probably a transcription error in that version; 「锦秋家园」 here matches public reports.]

Over these years we kept discussing internationalization. Last year we said we would speed up internationalization. One colleague was very keen, and one day told me he was going to India next week for research. A few days later I asked him which city he was in, Delhi or Bangalore, and he said he was still on Zhichun Road... He'd been stopped at Indian customs. I asked why. He said his document wasn't valid. I asked what document he had. He said an APEC card. He'd taken a card from the Pacific Rim organization to a country on the Indian Ocean. He said, it says India on it. I searched for ages and found an abbreviation, IDN. That's Indonesia, OK?

Of course this didn't stop his enthusiasm for going to India; a week later he set foot on Indian soil after all. He even posted a short video on Douyin, sitting in a "tuk-tuk"; after a long trek, it took him two tries to finally get in, and he looked pretty cheerful. This colleague did a lot of industry analysis and research in India and recruited many candidates; I don't know how he did it, since his English isn't that good. That made me think of Tang Seng (唐僧) going west to fetch the scriptures. I won't say who this colleague is; just think of him as Tang Seng, since he did fetch the true scriptures anyway. [Translator's note: Tang Seng, the monk Xuanzang as portrayed in Journey to the West.]

What I mean is that very often, even though conditions are limited, whether the office is small or you don't speak the language, the dream can still be vast, and you can still pursue creating very big things. I once saw this line on a construction site, "Space has a shape; dreams have no limit", and I felt it fit perfectly.

A couple of days ago I looked through my Weibo and found a private message. It reminded me of the story of how we first built the recommendation engine. At the end of 2012, not long after the Toutiao app was released, one evening in the 6th-floor office at Jinqiu Jiayuan, I called all the PMs and RDs to a meeting. The core topic: if we want to build an information platform, it looks like we must get the personalized recommendation engine right. Do we start this now or not? At the time very many people worried that we didn't have the "genes" or the ability, and expressed concern one after another. I said, we don't know recommendation, but we can learn. So I found the author of the book 推荐系统实践 (Recommender Systems in Practice) and asked him for an electronic copy to read. He said the book wasn't published yet and wouldn't give it to me, seriously delaying our company's recommendation capability. I could only look for material online and wrote the first version of the recommendation engine from my own imagination. He has now joined our company.

At the time, building a recommendation engine was still very hard for a startup. Many people were building similar apps; some relied on unscalable operations, some tried to personalize through simple customization; very few companies truly committed to building a recommendation engine, and many failed.

But we felt that if we didn't solve personalization, our product would just be micro-innovation; we might pick up some of the mobile internet's dividends, but we couldn't achieve a fundamental breakthrough or truly create value. At all times, we must strive to solve problems at the root.

Speaking of mobile internet, at the time the whole industry also had no confidence in mobile internet advertising. People felt the screen was small and unsuited to ads. Ad formats were all banners, offer walls and the like, with very low conversion and a very bad user experience.

But when I asked Zhang Lidong, he said he was especially confident. I asked him why, and he said because people often wrote in to the user feedback mailbox wanting to advertise. Even though these users' individual ad spend probably wouldn't exceed 5,000 yuan. We thought it over carefully and felt that personalized-recommendation feed ads were what could solve the problem at the root. We had to find an advertiser to test it. In September 2013, Lidong finally brought in a deal: Gome's Beitaipingzhuang store. At that time we didn't even have an ad system. What to do? We hardcoded it directly in the feed, writing the ad creative data and the geographic targeting range into the business code. To prove the ad worked, we also designed a closed loop: after seeing the ad, click to bookmark the article, take the article to this Gome store, spend 200 yuan, and get free cooking oil.

At first we set the recommendation radius to 3 kilometers. A whole morning passed and no users went, so we expanded the radius to 10 kilometers, and got a dozen or so users. We kept expanding, and in the end the coverage reached the North 5th Ring Road at its northernmost and the South 2nd Ring Road at its southernmost. More than a hundred people came, and we finally gave the gifts away. Although the result could hardly be called good, we were very excited: at least we'd achieved a mobile internet targeted closed-loop LBS ad.

It reminded me of something in the Steve Jobs biography: on his 17th birthday, Jobs's father gave him a car, the following passage is all about how beat-up the car was, but at the end Jobs says: but still it is a car. That's how I felt about our first personalized-recommendation feed ad. [Translator's note: "but still it is a car" is in English in the source.]

Do the right thing, not the easy thing

Our Toutiao Missing Persons Search (头条寻人) has now found 8521 people. On average it pops up 80+ missing-person notices a day, and on its best day it found 29 missing people.

We also had a lot of discussion about doing this. You've probably all seen 404 missing-person pages, which many companies have done: putting missing-person notices on a page that basically can't be opened, so by the time users see it, the child may have been missing for a month, and isn't in the same city either. Colleagues at our company proposed doing that very early, but I firmly opposed it. That method basically can't find anyone; it's all a gimmick, better not to do it at all. If we were going to do it, we'd seriously find a way that could solve the problem.

By early 2016 our users had reached a certain density, and we discussed again whether it was possible to go from LBS ads to LBS missing-person search. After the first successful case in Yanjiao, we carefully did the calculations: how much it would cost to find one person, whether at 100 million DAU we could meet the need at scale, whether we had the chance and the ability to carry missing-person information for all of China. Today, Toutiao Missing Persons Search is already the most effective platform and the one that has found the most people. Later I found that the US has a system called AMBER Alert, very similar to the system we worked out ourselves. I hope all our philanthropy projects seriously pursue sustainability, broad reach and high efficiency, and don't do superficial work.

Some time ago we opened a new office, Zijin Digital (紫金数码), and its renovation was widely discussed on Toutiao Circle (头条圈), with many colleagues complaining the office smelled bad. Because our team grows so fast, we always have to move to new offices, so we've always had this problem. I asked the real estate team why it couldn't be solved. They said that tested against the highest standards, it absolutely doesn't exceed limits. We did try all kinds of ways, but this is an industry-wide problem. Some companies put out more plants, but that's really only psychological comfort.

So I said, have we really tried the most effective methods in the world? Could we get involved in the production stage, could we contact materials experts and start from the craft and process to solve the problem completely? Although it's hard, and maybe it can't be done in a month or a quarter, or even solved within a year, we still need to find ways to try. I hope next year we can have a breakthrough in this. It would not only solve the air quality problem tens of thousands of employees face when moving, but might also be a good business opportunity.

One day I saw a JD our HR wrote for hiring PMs and was really angry. One line said: five or more years of internet product experience, with experience in product planning and product iteration for products with ten million-plus daily actives. I told this HR, by this requirement, Chen Lin, Zhang Nan, a whole batch of our company's PMs, not one of them could get in; even I couldn't get in. Never mind ten-million-DAU products; before they joined, they hadn't even done a product with a million, or even a hundred thousand, DAU.

Many colleagues didn't have a glamorous background or a great résumé when they joined our company. The company's product managers include people with design backgrounds, operations backgrounds, and engineers who couldn't write code well and switched roles. Maybe some prefer to hire glamorous-background "gods and goddesses", but we prefer unpretentious "fresh young things" (小鲜肉). The idea we've always held in hiring is to find the most suitable person: whether their traits really fit, focusing on a person's fundamentals. School, relevant experience and title aren't that important. Writing a JD like that is easy; essentially it's laziness. Discovering a person's traits is what's hard.

We may not be able to do it best for now, but we must always keep the mindset of solving problems at the root, whether it's philanthropy, renovation or hiring.

Embrace uncertainty; great force works miracles

Toutiao Hao now has great influence; from articles, video and Q&A to live streaming supporting poverty alleviation, it has become a very important content ecosystem on the mobile internet. In 2014, when we decided to build the Toutiao Hao platform, progress was very slow. The product was very crude, and inviting authors was very hard. At the time, beyond their own jobs, Toutiao Hao's product and operations people had to invite authors to join every week, leaving messages for creators via Weibo private messages and WeChat official account back ends, explaining from "what is intelligent recommendation" and "what is Toutiao" onward; often they'd approach several hundred people and only a few would be willing to join. There were also many doubts internally about whether we could make it work. But we still felt it was very important for building the platform's content ecosystem, and as I recall Chen Lin was especially insistent. In this situation, the company mobilized all employees to invite creators to join. Many colleagues started by inviting self-media friends around them, and all colleagues doubled as customer service for Toutiao Hao authors, answering questions and solving all kinds of issues through internal communication.

Looking back, many of our methods at the start weren't good, but we worked very hard and were very focused, and great force works miracles (大力出奇迹).

The short video business also went through some twists. In early 2015, at the Okinawa annual meeting, I took everyone to an izakaya to discuss for the second time whether to do short video. The first time was 2014, but at that time the whole Zhichun Road subway was covered with ads for Tencent Weishi, and Weibo's Miaopai was also promoting all out, so we hesitated. Plus we'd been through a lot of turmoil and couldn't spare the energy. By the end of the year Meipai and Kuaishou had already taken off, and we felt we had missed it.

At the end of 2016 we discussed it again and felt we still couldn't give up: this would be something that brought a lot of change to the world, and we still had to try with full force. Not only do it, but do two apps; not only do it in China, but do it overseas; not only do it overseas, but also do acquisitions well.

Later we really did acquire Flipagram. At Spring Festival 2017, I suddenly told the team we were going to Los Angeles to do the integration work, and everyone on the team looked troubled, because they had rarely been abroad before, and all said they felt unsure. Now the company already has several products making progress in many countries.

The meeting room in this photo is the office from our first trip to Flipagram. Last year I also steeled myself and did a CEO face-to-face in LA. Later I saw a local media outlet say I gave a speech in halting English, "蹩脚" English. [Translator's note: "halting English" is in English in the source, followed by the Chinese gloss 蹩脚.]

I tell these two stories to say that many things are very hard at first; you have to mobilize resources and go all out trying many times before you can make progress. Right now some of our products still aren't good enough. We welcome active complaints and suggestions, but don't give up hope so easily. I think saying "it's dead" (凉凉) at the drop of a hat is very snobbish. What is snobbishness? Snobbishness is merely echoing the surface status quo, being unable to go beyond the present and imagine things that haven't happened yet. We should absorb the truly valuable complaints, and on important things, in the right direction, have abundant passion, bold imagination and indomitable will, try in a down-to-earth way, and great force works miracles.

Many friends ask me why we're doing social, and there's internal feedback too: don't compete with a certain company, the pressure is huge. Let me show you this picture: a screenshot of a Douyin link being blocked when a user sends a message on WeChat. Last year in the app alone we received 200,000 user complaints. People were asking, why can't I share links via WeChat? Is your software broken? Why can't I send my mom a Douyin video, why can't I send my classmate a Xigua link? We could give up commercial interests, avoid competition and not do a certain thing, but how do we face these 200,000 users' complaints? Should this problem be solved or not?

Yesterday a colleague asked me whether I had expectations for Duoshan. I said my expectation for Duoshan is no expectation. If protecting these users' right to share and communicate is an important problem that needs solving, our expectation is to keep thinking, keep trying, and find a way to break through. [Translator's note: 多闪, Duoshan, ByteDance's video messaging app launched in January 2019.]

We are a group of pragmatic romantics

The outside world describes us in many ways: that ByteDance is an overly rational company, that Zhang Yiming is an overly rational person, an "A/B testing company", an "app factory" and so on. I strongly disagree. Yesterday on the way to Jinqiu Jiayuan, recalling these stories, I felt that we're a very romantic company. A colleague asked me whether we should call it idealism, since "romantic" is a bit pejorative and sounds unreliable. I said no, idealism isn't enough; romance is more romantic than idealism, only ours is pragmatic romance. What is pragmatic romance? It's turning imagination into reality: face reality and change it.

Refined artiness is not romance; rough grandeur is romance; new things are all rough. Showing off your feelings to deliberately move others is not romance; independent thinking that cuts through the noise is romance. Having vitality is romance, facing the future is romance, embracing uncertainty is romance, keeping possibilities open is romance.

What is pragmatism? People often treat doing easy things as pragmatism, but it isn't. Doing the right thing is pragmatism; short-term speculation is not pragmatism. Great force working miracles is pragmatism, getting to the bottom of things is pragmatism, grasping the essence is pragmatism, respecting users is pragmatism, recognizing the world's diversity is pragmatism.

Our team, whether imagining huge things in the shaped space of Jinqiu Jiayuan, or a colleague who can't even speak Mandarin well taking an APEC card to India for research, these are all very romantic things. Even our ByteStyle (字节范) is the most romantic and pragmatic expression. "Always strive for the extreme" is romantic, "open and humble" is pragmatic, "always day one" (始终创业) is romantic, "candid and clear" is pragmatic, "pragmatic and daring" is both pragmatic and romantic.

What else is pragmatic romance: having empathy is pragmatic, having imagination is romantic.

Some people say, is A/B testing all your company knows how to do? I posted a Weitoutiao: empathy is the foundation, imagination is the sky, and in between are logic and tools. A/B testing is just a tool; it can't test user needs; empathy is the important foundation. Without empathy, the products you make certainly have no soul and can't meet user needs. But empathy alone isn't enough; that can only make useful products. To make outstanding products, imagination is very important. [Translator's note: 微头条, Weitoutiao, Toutiao's short-post feature.]

When Toutiao was still very crude and had very little information, we imagined Toutiao's feed connected to an intelligent antenna, the antenna connected to a boundless ocean of information, and every refresh would bring back from the ocean the information you are most interested in at this moment, in this place. So we worked hard to cover all kinds of useful information, from news to pictures, from agriculture and rural life to academia.

Douyin is the same: imagine full-screen video turning the phone into a window through which you see a rich world; Douyin is a projection of this colorful world, which feels wonderful. Without imagination, you might only make a popular lip-sync app or a funny-video app, and Douyin could not have evolved from a cool niche music-and-dance app into a mass platform embracing beautiful scenery, opera and art, moving stories and everyday consumption.

Tracking down bugs one by one, doing user interviews and A/B testing is our pragmatism, but we also have imagination, and romance. Information creating value is our pragmatism; recording a good life is our romance.

In the movie 《爱丽丝仙境漫游记》 (Alice in Wonderland) there's a passage that left a deep impression on me, and I've excerpted it to share with you: [Translator's note: the source calls it a film; the passage quoted below is from Lewis Carroll's Through the Looking-Glass, though the source attributes it to Alice's Adventures in Wonderland. The English quotation below is in English in the source and is reproduced as given.]

"Alice laughed: "There's no use trying," she said; "one can't believe impossible things." "I dare say you haven't had much practice,"said the Queen. "When I was younger, I always did it for half an hour a day. Why, sometimes I've believed as many as six impossible things before breakfast."

——Alice's Adventures in Wonderland

Alice says to think of six impossible things before breakfast. I changed this line a bit: [Translator's note: in the source it is the Queen, not Alice, who says this line; translated as written.]

Romance is all about imagining interesting things that could happen in theory but not in reality yet.

Changing "impossible things" to "things that could happen in theory but haven't actually happened yet". That's what romance is.

As Alice said, every day and every year, think about what things could happen, but won't happen if you don't think about them and do them. Whether small things or big things, whether at work or in life. As a company we should also keep thinking about whether we can do things that surpass the current level of the industry, surpass current products, and are more innovative and more valuable. I hope we will always be a company with imagination.

On pragmatism and romance, there's another related pair of concepts: ego and breadth of vision (格局). The opposite of ego is breadth of vision; pragmatism and romance are essentially about having a small ego and a broad vision, which is also how you can have empathy and imagination.

From the company's perspective: some say that when the business grows fast, office politics doesn't easily appear, because fast growth means the landscape gets bigger, so even if people have big egos they don't easily collide; but once the company grows slowly, people easily get squeezed together. You can imagine electron motion, or planets in the universe: small ego, broad vision, more room to perform. The opposite is a box packed with many inflated balloons.

The English word for 公司, "company", as a verb also means "to accompany". In the company's early days I used a metaphor: building a company is like a journey; let's go see the most beautiful scenery together, not linger and dither halfway, not take shortcuts and go astray. Today I want to add half a sentence: I hope to go see the best scenery together with all of you, this group of people who are both pragmatic and romantic.

That's the end of my talk. One day on Douyin I came across a video of a little boy running happily. At that moment I had a feeling that running with all our strength is our romance. Running together. Thank you all! [Translator's note: "running together" is in English in the source.]

Copyright: © LookMinds. Our own edited text and translation, not a substitute for the original; rights in the original belong to the author and the platform. Please see the original link.