China's Quora: How to start from zero, and solve the cold start problem


New marketplaces and social products have a common cold start problem: The network is not useful until you have enough useful content. The network increases in value exponentially as you have more content and users. But, at the beginning, you have nothing to start with. A classic example is a new product that has no users yet. How do you solve this problem?

Zhihu may not be as well known outside of China as WeChat or ByteDance's Douyin, but over the past years, it has cultivated a reputation for being one of the country's most trustworthy social media platforms. Originally launched as a question-and-answer site similar to Quora, Zhihu has grown to be a central hub for professional knowledge, allowing users to interact with experts and companies in a wide range of industries.

Even, in the three or four years of Zhihu's growth, Zhihu founder Zhou Yuan has always been asked how the community started from zero and how to solve the cold start problem. Recently, in a public talk, Zhou Yuan raised this question again and shared with entrepreneurs the process that he started the community from 0 to 1.

 

Four key points that influenced the community product

In his speech, Zhou said that Zhihu is actually a platform that helps users from different industries share their knowledge and insights, while the community is a very low-cost platform to help users communicate and create new relationships and connections between them. In the community product, there are four key nodes that affect the product as a whole.

The first is input. No one defines input method as a keyboard input method, if what you think in your mind is typed into the internet, it can be printed as a 3D printing model. When talking to Sogou Input Method, the people at Sogou thought that if a little change was made, the information in everyone's mind would have a better chance of being passed on to the Internet. And in Zhou's opinion, with the community shape changing greatly, there must be a great combination on this side.

The second is thinking collision. Why is it that when people talk about things they will find a café? Because you can have a collision of ideas and experiences with others in a café. What about on the Internet? For example, on Zhihu, you see many people asking questions every day. If you don't do anything, it basically amounts to seeing what questions people are more interested in and what kind of questions they are seeking answers to. Every question that arises is actually a node of information. There are all kinds of community products in the world, but which are constantly generating opportunities for thinking collision.

The third is mobility. There have been many thoughts about why blogs do not do well as a product. In addition to the high cost of writing, the main reason is that blogs do not have an internal circulation of information flow mechanism. In the past, the blog wants to get traffic, it has to take advantage of the search engine. If the search engines couldn't find it, basically it is no use. In the process of evolution, if the product can cause a better flow of information, it can pay attention to the things produced more, it distributes things more accurately, and can hold a great deal of say in the information node.

The last one is the closed loop. The closed loop in a community is whether the role each person plays in the engagement process can lead to a long-term transformative relationship. When you go back to traditional media, the relationships are never transformed. In the case of Zhihu, the questioner and the answerer can be transformed.

 

Three questions must be answered clearly before start

1. What exactly is the common value of this community that you are trying to make? 

No matter what kind of content it is, when you see it, you know that these people are the same as you and want to connect and communicate. This common value will have strong and weak differences. If you open WeChat, you will have many groups in your contacts. You will find that some of these groups have a lower frequency of communication and desire. However, if it is about investment, about stocks, about a certain interest, or even related to your personal interests, it will be very active. The common values will determine how strong the motivation is for this group to interact in the community.

2.  How do you find your seed users? 

If you don't even know where these seed users are, or what kind of platform they are gathered on in some way now,  it's basically hard to solve the cold start problem either.

3.  Can your product form some kind of closed loop? 

We've seen a number of breakout apps and great products in the WeChat Moments last year, are people still interested in them? It may have a very large flow of traffic in a short time, but in the end the traffic does not turn into real users and does not really flow back. After three days, or a month, we may forget about these products. You have to think about whether the model is sustainable after a cold start.

When you've answered these questions, it's about how to start from 0 to 100, or 100, or 1,000, which is just the first stage.

And when Zhihu is solving the cold start problem, good reputation is the most effective way to get a community product going. I think this is probably the only way. True communication must be the transfer of information generated by a small circle of people in your life who have already established a trusting relationship. It's not about spreading the word, it's about those people around you who you trust and who want you to send out messages that are word of mouth.

The most important point is also what values this community has in common, it could be a shared belief, a shared interest, or a shared identity. They have a common identity, they are a common group to communicate with.

 

The aspects that can measure the products

The user is the scene. The whole internet has changed in that scenarios are becoming smaller and smaller, and more precise and close to the individual. Before, all scenes were basically big platforms. After four years of development, many users have become scenarios themselves in the platform. In the whole community product, it should be possible to continuously produce a lot of small platforms that are not standardised, it must not be the big platform side, but based on the same small group of users using the scenario.

The founding team must be active users of the product. We were inviting our friends to come and use it. At the same time, we ourselves were the first wave of heavy users of the product. When a user comes in, he doesn't read any manuals or guides, he just sees how the previous users use it. At a very early stage, it is very clear that you yourself, your team, should become its user and should express how it is used and how active it is, with your behaviour.

Make sure you understand the emotions of the users. Zhihu has a feature called "Like the Author". It is an important feature for long-term peer-to-peer communication between users throughout the community, in addition to making objective pro's and con's. 

Maintain restraint in your operations. You will be particularly confident if you have done very well in the previous stages. The result of that confidence is that you are likely to want to do more with methods that have been proven before, and this is the time to be careful. Because it is essential to keep the community at its own pace, whether that pace is fast, or slow. If you are too good at operations, you will turn yourself into the role of God, and this time you will make a lot of very stupid and ridiculous mistakes.

Good reputation is the core that community product starts from 0 to 1. First of all, early on, you must use your product authentically, you can't just hand it over to users after you've made it. This is a very lazy idea. You yourself must become the core user of this product.

Secondly, operational behaviour has to be gradually replaced by user behaviour. You have to become an active answerer. If, after three months, the most active person in the whole community is still yourself, there must be something wrong.

Thirdly, you must make sure that the path of communication exists in the closed loop of the product. We used something that spread very well on WeChat last year, but it didn't form a good cut-off model.

The last point, and the most important, is that at the heart of word-of-mouth communication, it is not the spread of mass traffic, but the ability to have word-of-mouth from people who have actually used the product.