Founder Yang Bo Says Douban Is Going to Become a Simple Place

Douban, which started out with books, movies and music, is trying to use yet another redesign to turn itself into a general-purpose platform for the mobile era.

Regular article

A long-time Douban user has habits he cannot shake: tagging the books he has read and the films he has seen, reading posts in groups with odd names, and posting on the Broadcast feed about how confused he is by Douban's latest redesign. If the confusion gets bad enough, he may click "My Account - Delete Account" in the top-right corner. Most people hesitate somewhere along the way, hit the back button, and go back to tagging books and films, reading posts in groups with odd names, and, a while later, posting on Broadcast about how confused they are by the newest redesign.

They are experimenting. Douban is experimenting. Abei is experimenting.

"Before, we built a few cities within the city. Now we are tearing down the walls of those cities within the city." That is how Abei, as founder Yang Bo is known on Douban, explains the site's past changes. All of them, he says, were preparation for a goal that has never changed: to become simple. Simple does not mean offering only one kind of content. As books, films and music were joined by travel, weight loss, games and cosmetics, all of it came to be grouped under one word, "interests," and Douban can lead users to "discover" more. That is what the site wanted to do from the start.

For more than an hour before we interviewed Abei at Douban's office on Jiuxianqiao Road in Beijing, he was in a product decision meeting with his product managers, one of the company's most important weekly meetings. Each product line used to have its own product managers; over the past year, Abei has taken more direct control of the product. "The company's main product is clear now. There is just this one product, and many product managers meet about it together."

It is also a sign that Douban's thinking is getting clearer. The product Douban is investing most in, and the one Abei is most involved in, is its mobile app. Abei has chosen to make it a general-purpose platform, carrying out the intention already visible on Douban's 2005 homepage: to discover shared tastes. In Douban's current words, it is to "gather the life interests of 100 million people."

Abei has been trying every possible road in this direction. When we showed him screenshots of Douban's homepage across ten years, we could see how tangled those roads were, but the direction was clear. A long-time Douban user also has questions he cannot shake: What does Douban actually want to do? Why does Douban keep changing? Why is Douban still not making money? Why did Douban build Alpha City? Why does Douban keep its distance from offline, on-the-ground sales pushes? We put these to Abei and listened to his answers.

Why did Douban's app homepage become Doulists?

QDaily: Let's start with what everyone cares about most. Why is the Douban app now pushing Doulists so hard?

Abei: With 2.0 we started from Doulists, but that doesn't mean the whole product is built around them. A Doulist takes something users were already doing on Douban, organizing the things they had discovered, and presents it in one unified format.

Before, users collected what they found in various places and put it together. But there were too many ways of organizing, and it was messy, so now it is unified in the Doulist format. Doulists are one part of Douban's whole discovery system. The Douban app isn't built around Doulists, although they are the biggest change in 2.0.

The more important change behind Doulists is that they break down the walls that used to separate Douban's three vertical areas, books, movies and music, from its user-generated features such as Groups, Notes and Albums.

A Doulist is a title with anything inside it. It can hold a book, a film with a review, a photo, or a note with travel tips. In effect it gathers the many kinds of content on Douban in one place. That is the biggest change.

Say you want to travel to Taiwan. What might help you on Douban could be a book about traveling in Taiwan, a group about traveling in Taiwan, a note a user wrote about a trip to Taiwan, or a Taiwan Doulist another user made. Before, all of this was scattered around Douban. You would go to Broadcast, then to Notes, then search through different Groups, then look under books for books about Taiwan. It was spread over different places.

I think when a person is interested in something, their interest usually covers all of these things. In 2.0 we began presenting Douban's various services together in a more consistent form. Doulists are one way of doing that.

QDaily: In 2009, Douban first opened to Broadcast after login, then it became Douban Guess, and after that it changed quite a few more times. Now the app opens to Doulists. Why does it keep changing?

Abei: Basically, because a general-purpose discovery service is hard to build.

Users find good things by many routes, and none of them is perfect. Take Broadcast: if a new user doesn't quickly follow a lot of top-tier users, their homepage is empty.

The biggest difference between "following" on Douban and on Weibo is that on Douban the people you follow are experts in all kinds of everyday interests. Their ability to discover things and the range of what they see are far wider than an ordinary person's, but they are not public celebrities. So you can't start by recommending a pile of famous names, saying this is so-and-so, you can follow him. Weibo can do that, so your homepage has content right away.

Many of the users most worth following on Douban are like this. They have no fame outside, they aren't a recognizable name to the public, so you didn't know them before and couldn't follow many of the right people on day one, and the homepage is of limited value.

The most effective approach is for a new user to get good content quickly when they arrive. That may come from trending topics or algorithmic recommendation, so there has been some back and forth. Douban's understanding of this has built up year by year, and I'd say it is fairly mature now.

That is also why Broadcast is no longer on the app homepage but inside Circles. Different types of users feel differently about Douban. Many think of it as a place to read content; many think of it as a place to follow interesting people. These are different needs, and they have to be balanced.

QDaily: What will Doulists become?

Abei: A big part of Douban's value is helping users discover good things in every field. At the very beginning, that meant algorithmic recommendation.

There are several ways of discovering things. The main ones right now are rankings of all kinds and trending topics. Media is about trending topics; for films and books, users look at rankings. When they aren't looking for anything very specific, many people who want to know what good books have come out lately will look at a ranking. A bookstore has a shelf that says these are this week's bestsellers. That is one way.

Another is what Douban has always done, algorithmic recommendation. Its purest product form is the personalized radio inside Douban FM, the private channel, which picks songs for you by algorithm. That was the earliest thing Douban wanted to do. Later we realized it was part of a bigger problem, not the whole of it.

A third way is users: relying on other users to help you discover things. That is what Doulists do. A user who is interested gathers the good things he has found into a Doulist and puts it out to the whole world, saying: whoever is interested, I don't know whom this will help, but here it is.

Users had always been doing this. What 2.0 did was make Doulists subscribable, so others can follow them easily, and now they are part of the homepage. In fact, beyond Doulists, most of what users see on the app's homepage right now is what they get from the fields they are interested in. If a user is interested in home living, for example, content under home living is recommended to them. That part of the homepage comes from trending topics and Douban's traditional algorithms. At this moment most of it comes through these methods, in combination.

QDaily: How many people at Douban work on the Douban app, and how is the work divided?

Abei: More than half the people in our company now work on the Douban app. The largest group is the engineering team. We have a big editorial team of 20-odd people, and then product managers and designers: a dozen or so product managers and fewer than 10 designers. The rest are engineers. Together these are the product and technology teams.

Everyone else is in company-level teams, the kind every internet company has besides support teams, plus the few independent product lines I just mentioned, some of which are subsidiaries.

What does Douban plan to do this year?

QDaily: What else does Douban plan to do this year?

Abei: That is fairly clear. One thing is ratings and reviews. We have always wanted them to be a horizontal, general-purpose service, so that not only books, films and music but other things too can be rated and reviewed. We hope to get there step by step.

The other is discovery on Douban, including the Douban community and everything beyond books, films and music. We want it organized around personal interests, around life or interests: weight loss, for example, is that kind of topic. It will gradually become clearer, with broader usage and wider coverage.

Douban actually has content for all kinds of interests, and a lot of it has been produced. We just haven't organized it well or delivered it to users well, so this is a second undertaking: to cover, fully, the varied interests of the broadest range of users.

Each of these is a huge undertaking, and both are things we very much hope to achieve on Douban. So this may take more than a year.

There is one more thing. Our app, our website and Douban in mobile browsers should be consistent. They differ now. The homepage, for instance: on PC it is Broadcast, in the app it is content. They still differ in many places.

Taken together, what we will do next is make Douban general-purpose.

QDaily: You mentioned that Douban may become simpler. What does "simple" mean here?

Abei: Say Douban had only books. That is simple. Books and films, a bit more complicated. Add music, more complicated. Add games, more again. Then cosmetics and so on, and it gets very complicated.

But once you have a dozen or so categories, in the end it all becomes interests. And personal interest is a fairly simple, clear thing. Once it is general-purpose, it becomes simple.

You'll see many people compare Douban with other places, Douban versus Tianya, Douban versus Weibo, Douban versus Zhihu, and in the end find that none of these services is like Douban.

Douban has always been general in nature. It's just that there has usually been a vertical counterpart alongside, and putting the two together makes things complicated. What we are doing now is taking the general things all the way, and then it becomes simple as a concept.

QDaily: Douban started with movies, music and books. Doulists have also existed separately within each section for several years. How did you come to decide to merge these things and relaunch them?

Abei: Douban started with books, films and music, but its own vision, what it has always wanted to do, is to expand into more fields, so over these 10 years this has been a major theme. In 2007, the influence of Douban's books, films and music sections was growing, gradually spreading from a niche crowd to the general public.

At that time we had just formed a product manager team, and we were discussing how to go into more (vertical) fields. We gave three product managers one field each and asked each to study how to approach it and what was worth doing. I remember clearly: one had digital products, one cosmetics, one games.

But in the end we concluded that it wasn't right to go into (more fields). There were two reasons. One was that Douban was still small then, a company of about 20 people, and the job seemed too sprawling for us to be sure we could do it. The other important reason was that we were still obsessed with personalized recommendation, which didn't seem worth as much in other fields. So at the time we didn't push Douban's service into more fields.

But at the same time, Douban Groups were attracting more and more topics beyond books, films and music. Users were creating groups and discussing cosmetics, games and digital products in Douban Groups. So instead of treating those fields the way we treated books, films and music, with ratings and reviews, we put our attention on the community and watched whether Groups could grow into something general. So our conclusion then was that recommendations, ratings and reviews didn't suit those fields.

We brought this up again recently because the question never had a good answer. One direct result was that on mobile, Douban got split into many apps: books, films, music, Groups. That is a big problem for us, because users can't tell which one they should use. So it is still not an ideal way to tell users clearly "what Douban is." When we integrate, the integration has to happen at the level of the concept, not just by putting several apps together. So the biggest change in Douban App 2.0 is that, as a concept, it makes Douban a general-purpose thing.

Doulists, interest tags: these things are general. Look at what was added to Douban App 2.0, Broadcast and Groups: they are all general, horizontal things, so within one product we get good things from every field quite naturally.

QDaily: There's a question I've had about Douban for years. If I search for a film under Movies, the book with the same title doesn't show up. Was that technically hard, or purely a product decision?

Abei: It was a product decision. For a while we really did treat them as independent services. The Douban Books and Douban Movies logos that still exist, and the standalone Douban apps, are all results of that.

By that logic, if you use Douban Movies, just use Douban Movies; you don't need to know Douban has anything else. That was once our thinking.

QDaily: Can I put it this way: when we talk about interests, what Douban wants to do has no boundary?

Abei: We call it general-purpose. Whatever your interest, it exists on Douban. If you are interested in Tesla, for example, there is already content on it; it's just scattered across Douban, in mini-sites, in Groups, in users' albums and notes. You just don't see it right away. It's buried rather deep.

QDaily: Put it this way. A place where I can follow everything I'm interested in and everyone I'm interested in. That sounds more like a public social network, and in the WeChat era China has nothing like it, never mind Weibo.

Abei: We once called this discovery. Discovering personal interests means finding the things most useful to you, which may include content and may include people. But that is still a rather broad thing. Objectively, many products help you with these things, just in different ways. For example, if you subscribe to a few official accounts on WeChat, you also get new content in those fields every day. (These products) can help you in many, many places.

The key is how Douban does these things. Douban has always been a social service. At the simplest level, most of the content users see on Douban is made by users. It is neither produced by editors nor scraped from other sites. So the social gene has always been in Douban.

Second, we still put the stress on personal interest, on what matters to the individual. Plenty of things, current affairs for example, are followed by everyone, but how much they touch or affect an individual's actual life is limited. So I think there is a very big difference here, and the way of doing things will be very different.

QDaily: In your ten years running Douban, what was the hardest stretch? Was there ever a time you were so anxious you couldn't sleep?

Abei: I've probably always slept well (laughs). Every year there are problems and challenges, sometimes big and sometimes small. But Douban as a whole has kept making progress, sometimes fast and sometimes slow.

QDaily: What are this year's problems and challenges?

Abei: Let's talk about these two years. Last year we decided to move all of Douban in a horizontal, general-purpose direction, integrating the whole Douban app. That was quite a big decision. Before that, we had spent a year trying new mobile internet products, new things that had nothing to do with Douban.

You could say we were experimenting, which has always been how Douban works, but every year brings new problems: how do we do this? Douban has always had one problem: there is no precedent. It isn't as if there is a company in the US and Douban only has to copy it. Douban has had no precedent from its first day until now, 10 years.

So we made independent attempts, and then decided to integrate Douban completely, to make it fully general-purpose. That was quite a big decision, and not an easy one to make. But once it was made, it was made.

As for this year, the next step is more on the user side. Douban's product direction is fairly clear; there are always decisions to make on details, but overall it is clear. The question is whether this (product) matches the Douban that the broad mass of users have in mind. If I have to name this year's problem, that may be the one we are putting a lot of thought into.

Why is Douban splitting into more companies?

QDaily: Douban has quite a few products now. Maybe a dozen or so in total?

Abei: No. It's actually quite simple: right now there is Douban, and we focus on Douban. Most products have been merged into Douban.

Apart from that, Douban FM is an independent product, and so is Douban Read. The offline work our musicians team does is also run by independent teams, some as subsidiaries and some as independent products. I don't get involved much in those, maybe once every few months. The rest is Douban as a whole: the Douban app and the Douban website, which are one thing. It's just that many things haven't been brought into line yet. For example, some people still use Douban Broadcast as a separate app. We will merge all of these step by step.

QDaily: You just mentioned that Douban Read and Douban Music have been spun off as subsidiaries?

Abei: Douban Read is not a subsidiary yet; that is in progress. It works on e-books and self-publishing around the Douban Read product. There is also a team working with musicians, which is now independent too.

But they are not the same thing as Douban Books and Douban Music. Douban Books and Douban Music remain inside Douban.

QDaily: When did Douban start setting up subsidiaries and running things independently?

Abei: All of it happened within the past year.

QDaily: The advertising team also seems to have been spun off. Why do that?

Abei: Douban as a whole is now defining and forming our core product. As a company, Douban needs to be more focused, on its main business.

At the very beginning there was just one Douban. After 2009 it evolved into four products: Douban Movies, Douban Books, Douban Music and the Douban community. Internally there were four teams; outside users saw and understood them as four different services.

In that setup it was hard to say which was Douban's core product. Many users thought it was movies; others thought it was Groups.

Now we are forming our core product, which is simply called "Douban." It includes ratings and reviews and interest discovery. The rest falls outside the main business, but those things have room to grow on their own, and a large enough user base considers them valuable services. So in management terms, we say they should be run more independently.

I don't get much involved in that area. Over time, I may oversee it from the board rather than as CEO; that's the biggest difference. In the end it comes down to the idea that a company should be focused. Focus doesn't mean doing only one thing; it means your main thing is clear. And Douban's main thing is clear now: the Douban app. The Douban app and website are consistent, and on many things the website will follow the app a bit later.

There is also a reason that has to do with the company's core capabilities and culture. Douban grew through product innovation, so the whole company is a team built around product. That is our strength and, we think, an important part of our genes; by the same token, there are many things we aren't good at.

Take musicians. If you are doing things on the ground, such as handling copyrights and putting on concerts, those are things we aren't good at right now, but they have to be done if you want to do the musician business well. So we felt it didn't need to be inside Douban. It can be a separate company whose capability isn't product innovation but these things. Its culture may also be different. That approach is more consistent and purer, and more effective.

QDaily: When you made this change, how did the team react?

Abei: I remember I even gave a talk to the whole company about why we were doing it, but there wasn't much opposition. For the team it was a good thing, because it created more opportunities: once each is an independent product, what each person is responsible for goes from a feature to a product.

How does Douban make these decisions?

QDaily: In 2005 you were one person, talking with users and then making the decisions yourself. Ten years later Douban has more than 200 people. How did decisions like Doulists and going general-purpose come about?

Abei: Many things work like this now. Take the decision to do Doulists: while building Douban App 1.0, we gradually saw it emerge in the course of practice and execution. Douban's 1.0 was mostly built around ratings and reviews. That was our first step, the simplest thing: integrating Douban's ratings and reviews.

With ratings, we kept asking what users look at and what they use to look. In answering that question, we gradually arrived at Doulists. Then, while building Doulists, we realized how important tags were and integrated tags.

So this discovery wasn't a case of me climbing a mountain one day and suddenly having a magical idea that we should do it this way. It was consensus that came out of week-by-week iterations. Mostly it is consensus; sometimes I need to make the call.

We now have a product manager team of a dozen or so people responsible for all kinds of things, and many decisions are made together.

QDaily: It sounds as if your job right now is mostly that of a product manager?

Abei: Yes, half of my time. That includes talking with product managers, thinking about and looking at which product decisions to make, talking with designers about many details, product-level matters. The other half is what I'd call company matters, the things any CEO has to do.

QDaily: Is the product managers' meeting an important one for you?

Abei: Yes. I have two important meetings. One is the management meeting on Monday morning, which all the company's executives attend, covering the week's important company business. After that meeting there are some people I may not see again that week. Douban Market, for example, is a commercial product; they discuss things at this meeting, and the rest of the time it is mostly email.

That's one important meeting. The other is the product managers' meeting on Tuesday morning, a product decision meeting that takes the whole morning. Today, for instance, it ran past one o'clock and still wasn't finished.

This arrangement is recent, from about the past year. Before, there were many product lines, each with its own head and product managers under them. It turned into a lot of big departments, and I managed those departments.

Not anymore. The company's main product is clear: just this one product, with many product managers meeting about it together.

QDaily: Does every product decision come up at that meeting?

Abei: Not quite. But it really is very different from before. When there were many product lines, how involved I was varied a great deal.

QDaily: I hear you like reading books about urban planning. Are you still reading them?

Abei: That was when I was working on Alpha City. I read less now. And I've already read everything worth reading. For more than half a year I had a Doulist called "Alpha City Founding Reading List." If you finish those books, I'm sure you could earn an undergraduate degree in urban planning.

QDaily: Which urban planning book influenced you in particular?

Abei: Probably The Death and Life of Great American Cities, a classic. It takes a critical approach, and it was probably the biggest influence.

There is also The Timeless Way of Building, which is closer to execution. It has hundreds of design principles, from the city all the way down to the glass in a window, all gathered into one big library. The same approach can be applied fully to designing an integrated community.

"Community" is of course an old word, but you can give it a new name now. For Douban, you might say something like interest-related lifestyle services. Its biggest difference from a communication tool like WeChat is that Douban is not a lifestyle service for person-to-person communication, so a big-picture view matters: you have to plan at the macro level. A communication tool like WeChat usually doesn't need much macro planning, or if it does, it isn't a fatal problem. A community does need macro planning, which is why it can be compared with urban planning.

For example, a street corresponds to a Group, and a district to an interest. Take an interest like health: Douban may have hundreds or thousands of such Groups, and that still matters to most people. On WeChat, I have hundreds of friends I message one-on-one and a few group chats, and I don't need to care how the other hundreds of millions of people use it; it has nothing to do with me. In an integrated community, it does.

QDaily: Alpha City is more or less shut down, but the move to go general-purpose is settled.

Abei: Right. I think you could say Douban today carries a shadow of Alpha City, or you could say the two are entirely different things.

At bottom, Alpha City's biggest idea was that users live in clusters by group, and people with different interests follow different things according to those interests, say each street is one interest, but inside it really is general-purpose.

Alpha City did have things related to books, films and music, but much more that had nothing to do with them. Back then, for Serie A and the Premier League, every club had a fan group that took over a street and named it after their stadium.

At bottom, Alpha City was a community platform, and the city was a metaphor. Covering every kind of interest and organizing by interest: the new Douban still does that. And we reached that conclusion slowly in the course of execution, not because I slapped my forehead and suddenly decided how it should be done.

QDaily: If running Douban is like planning a city, how far along do you think you are?

Abei: I think before, we built a few cities within the city, and now we are tearing down the walls of those cities within the city, the internal walls. We really did build several of them. Books, films and music were cities within the city. Some people won't be happy about tearing them down, but overall I think it will be livelier and will open up many new possibilities.

QDaily: Douban has made so many changes and launched so many new things. Did any of them cause a big internal dispute?

Abei: Not really. Though everyday discussions often turn into arguments.

Many decisions really don't happen all at once; there is usually a fairly long discussion. Take the decision to do Doulists: in the half year between the launch of 1.0 and 2.0, they kept turning up in all kinds of places, until someone stood up and said, hey, this thing is a Doulist.

If you're talking about disputes, movie ticketing involved a lot of people, so when we finally decided not to do it, it did affect some people. That was a hard decision too.

Product innovation is another word for product exploration, and exploration necessarily includes successes and failures. As for failed explorations like Alpha City, people who've been at Douban a long time take them in stride. It's no big deal; this is how Douban has grown, year after year.

For me, the hard part is usually deciding to shut down a product, or deciding not to do something, because colleagues have already put in a lot of time and effort.

When a product is shut down, I go and explain in person and apologize. I explain why we aren't doing it, why it isn't launching, or why we are closing it.

When we finally decided not to go on with Alpha City, some people did cry. The team had feelings for it.

If you really claim to be an innovative company, you have to tolerate failure. It is part of what you are made of, and the whole process matters. For me, the occasions I actually have to rehearse in my head are, if anything, ones like today's.

People who have been at Douban a long time find all this quite ordinary, nothing special. People newly arrived from other companies panic.

An old question: how does Douban make money?

QDaily: Here's a question many people must have asked you. Does Douban make money?

Abei: We are close to break-even now. Last year we were profitable in two quarters and not in the other two. Brand advertising is our main source of revenue and the largest.

QDaily: What other projects could bring in profit in the future?

Abei: We keep it fairly simple. From the earliest days, Douban has been watching the relationship between user services and e-commerce. Douban's first revenue was actually not advertising but e-commerce.

On Douban's first day online, we earned about 20 fen, a fifth of a yuan, from a revenue share with Dangdang. So Douban's oldest business model came from e-commerce: not from us selling things ourselves, but from e-commerce. We have always paid attention to this. Brand advertising came (to Douban) several years later, and it was the first revenue model to reach a certain scale.

We have also kept watching and working on the relationship with e-commerce. Douban now has an experimental product, Douban Market, that is directly tied to e-commerce, but we aren't the ones selling; other people sell (goods) on Douban.

We keep watching this; it just hasn't reached scale yet. We will keep trying and keep at it in this area.

We also have teams working on revenue, commercialization and Douban Market. They are going all out, always moving forward, always making progress, and our revenue is growing year by year. For the company as a whole, we are now close to break-even; last year we were profitable in two quarters and lost money in two.

On the revenue side, the thing to watch is brand advertising, which is also the largest now. Then there is e-commerce-related revenue. Those are the two.

But at the company level, revenue is not what Douban cares about most right now. Maybe it will be later, but not now. Nor is it what I care about most. My colleagues handle it.

QDaily: Speaking of e-commerce, Douban once offered online movie seat selection and later built products like Dongxi. But Douban seems to have always kept a certain distance, especially from offline, on-the-ground sales pushes.

Abei: The main reason is that Douban is still an internet product company. The logic that decides what we do and don't do sits on the user side: how many users, what services they need, how we provide them. The revenue model should follow. First the product's positioning is settled, then we consider which revenue model is most effective. If you were an e-commerce company, the logic might run the other way: first I decide what kind of e-commerce to do, then where the traffic comes from, then what product to build to reach users.

QDaily will continue to follow companies of this kind: ones that may have been around for a while and have earned real goodwill among users and standing in their industry, but cannot yet be called big companies. These companies keep a startup spirit yet have begun to reach scale, and we will look at what they plan to do next and what they are thinking about now.

Founder Zhou Yuan on Zhihu's 2015 and its future

Founder Liu Chengcheng says media is only one of his tools, and 36Kr wants to do something else