The operation logic of Zhihu and Douban communities!

Let me start with the definition: a community is a social platform that uses content as a carrier. This definition is just my opinion, there are bound to be different opinions. For example, when I was talking with a GM from Tencent about "what is a community", he found it difficult to understand my definition.

He has always felt that, according to the logic I mentioned, WeChat’s IM function is also a community. Because two or more people can communicate through text, video, pictures, and audio. But no one in the industry thinks that WeChat’s IM function is a community, so he questioned my definition.

Actually, I don’t want to discuss this issue because it is meaningless. Even in scientific research, many concepts are just codes, and it is good enough as long as they can support the research. It is a misunderstanding to focus on ultimate questions. For example, "life" is just a code name in a certain research, and we cannot focus all our energy on the description of the topic itself.

Obviously, whether WeChat’s IM function is a community or not has no impact on the development of WeChat, nor does it interfere with our understanding of this product.

Going back to the definition that community is "a social platform with content as the carrier", we can derive the three key elements of a community: people, content, and platform.

The key to building a good community is to make choices. First consider what people to choose and what kind of content to create, and then build the platform.

Let's go into the text and talk about these three key elements in turn:

1. Who to choose

1. People with similar attributes: self-identification and mutual attraction within the group

There are many dimensions for dividing human groups, and simple coarse-grained division is meaningless. The first is to have a certain understanding of specific groups, and to divide the intersecting groups into multiple dimensions.

• For example, young women in high-tier cities who are highly educated, have high incomes and pursue quality of life, young straight male sports fans, pregnant women who are preparing for pregnancy, and mothers of newborns, etc.
• Secondly, it is the division of a certain industry field or profession, and the characteristics of self-identity are more obvious. Doctors, lawyers, product managers, developers, designers, drivers, etc.
• In addition, there will be division based on interests, and the characteristics of mutual attraction will be more obvious . For example, books, movies, music, sports and fitness, technology and digital products, games, and the second dimension.

The above division method seems simple, but it cannot be simply divided according to demographic characteristics (gender, age, location), but requires an understanding of the common characteristics of the group.

I have seen many cases where demographic characteristics are used to divide user groups and conduct analysis. This kind of superficial look without seeing the internal commonalities is of no value for subsequent practical operations, and also shows that the operators do not have enough understanding of the target group, or lack the ability to understand.

What other divisions are unreasonable? Take a startup company I worked for before as an example, it served the e-commerce practitioner group.

The e-commerce practitioner community existed more than ten years ago and was called "Paidai.com". The CEOs of the best B2C e-commerce platforms at the time were all in this community, and some were even early moderators. Liu Qiangdong has also posted on it before, and Li Guoqing, Chen Nian, Li Shubin and others are frequent guests at offline events.

But as more and more people join Taobao, the community is far larger than B2C, and the people and content in the community are greatly divided, because everyone discusses different things. In addition, there are people working in traditional retail, warehousing and logistics, education and training, etc., which makes the management of the community increasingly difficult and the atmosphere is also affected.

Although this was not the most important reason for the subsequent closure of Paidai.com, at least this phenomenon did occur, that is, the groups did not have the same attributes.

In addition, there are many such "groups". For example, market practitioners, the middle class, college students, office workers, etc. are all overly broad divisions.

These groups seem to have a way of dividing and categorizing people, which isolates people. But individuals feel that they are different from others, and without group self-identification and mutual attraction, it is impossible to form a community.

2. People with production capacity: They must have the ability to produce content continuously

The community is a content platform and also has the characteristic of content supply driving consumption. If there is no content production, it will be like a river without a source of water. It will slowly become stagnant water and dry up. Therefore, the ability to produce content is very important.

It can be concluded that a large amount of effective content has positive value to the community. Therefore, having production capacity is one of the prerequisites for building a good community.

Capacity refers to the ability to produce content, including frequency and quality.

① Of course, the higher the frequency, the better, but it will also be restricted by the content form and quality.

② For short videos taken casually, the threshold for content format and quality is very low, so the frequency can be very high, such as daily updates. Academic papers or in-depth reports require a lot of research and argumentation, and have high requirements for the authenticity and rigor of the content, so the frequency will be very low, such as monthly updates.

The definition of quality is more complex and requires standards.

This standard is given by the main beneficiaries, which means that the "high" in "high quality" is in line with the standards of the beneficiaries and meets the demands of the beneficiaries. Therefore, good content is definitely not content with exquisite production in a narrow sense. Whether it is exquisitely produced or costly is not the core criterion for measuring the quality of content.

Therefore, frequency and quality cannot be achieved at the same time, and one must increase while the other must decrease. To ensure production capacity, we must balance the relationship between the two and find the point that maximizes profits.

Generally speaking, the vast majority of people in China do not have the ability to produce content. For general community products, content producers account for about 5% of all users, and 10% is considered excellent.

Even though the current short video platforms have been sufficiently penetrated to allow many ordinary people to participate, only a small proportion of them can maintain a frequent update. Even if they are shooting family matters or emotional stories, there is a threshold.

Therefore, if you want to increase production capacity, choosing a user group with production capacity is a shortcut.

Let me explain in detail below.

Production capacity can be divided into two aspects: the ability to express and the desire to express.

① The highest priority is those with strong abilities and desires.

For example, literary young people were the early users of Douban Books, Videos and Audios, media people were the early users of Weibo, and Internet people were the early users of Zhihu.

For example, many entrepreneurs also belong to this group. Because they have been in a difficult and lonely state for a long time, they need to vent their emotions and give themselves a boost. They also have the desire to PR themselves and their entrepreneurial projects, so they can often be seen on social platforms such as WeChat Moments and Weibo.

There are certainly many other reasons for the success of these community products, but the early selection of a user group that has both the ability and desire to express itself was of decisive significance.

② The next priority is those with strong abilities but weak desires

For example, experts and scholars in various fields are in the position of "professional users" in the user pyramid.

For such users, their mere existence is their value. Although the output is relatively small, the professional identity itself is an endorsement to the community, and just having people is enough. The characteristics of professionals determine that their production frequency and desire are often lower, because these factors are inherently mutually exclusive.

③ The last priority is for those with weak abilities but strong desires

This type of user is usually a die-hard fan of the community, and the real value lies in the interaction itself, not the content production. Die-hard fans are more like people who support the show. If no one reads the content, there is no interaction, and no one follows the content producer, no one will post content anymore. This is the importance of having loyal fans.

So, how do we determine the production capacity of a user group?

The simplest way is to look at the situation in existing communities. Find a community that is similar to your audience, observe who their content producers are, how often they produce, and how the production and consumption of the entire community match, and you will have a rough idea.

In theory, you can find a corresponding group for the community you want to create in existing products. Especially the top few large content platform products have already achieved this level and are definitely covered.

Other methods of judgment are not so simple, and they test one’s understanding and insight into a certain group. You should go to where they gather, communicate and feel, so that you can have your own understanding.

Never assign this task to your subordinate colleagues or fellow researchers.

It’s not that my colleagues are unprofessional, but I believe that any research report has a purpose. Because the researcher has received this task, he must come to a conclusion, so in the process of research and analysis, he will inevitably lean towards a certain conclusion.

In addition, any research report will be influenced by the researcher's subjective understanding. No matter how objective and professional the research and analysis process is, the conclusions drawn are still based on subjective cognition. Otherwise, there is no need for research, and AI will give the conclusion directly.

Therefore, the subjective cognition of the researcher or team will affect the conclusion.

Subjective cognition includes understanding of the business, as well as other knowledge and social experience, which will have an all-round impact. From my experience, understanding users is something you have to do yourself. You can use data and user research resources, but you must never rely on them or leave them alone.

In short, choosing a productive user group is extremely important and can even determine the success or failure of a community. The problems of many community products that have died or are sluggish are due to the user groups they target.

Without production capacity, where will the flow of content production and consumption come from? Without this flow, there will be no activity in the community. Without activity, no matter how strong the community culture is, it will not be able to survive.

3. Seed users are the community culture

The first thing to make clear is that we are not discussing the "community culture" itself here, but rather the choice of seed users, which has a decisive influence on community culture.

There is an old saying about a community: The first group of people who come into the community determines its success or failure. Although extreme words are always hard to believe, based on past experience, there is some truth to them.

Every step of community growth depends on the foundation of the previous step. The status and rules of the previous step are the premise of the next step.

For example, a passerby opens the door and enters an unfamiliar room. If everyone in the room is dancing to music, the passerby will think that this is a place to dance, and he or she can choose to join in the dance or leave.

It is for this reason that seed users are of decisive value to the community. The foundation for the growth of this community building is the first few people in this room, whether they are reading books and newspapers, or singing and dancing. More and more latecomers will imitate and follow them, which will have a guiding influence on the subsequent development of things.

Those latecomers who do not agree with the seed users will only choose to leave or be driven away. They won't try to change anything, nor do they have the ability to do so.

The community's head effect and the development trend of centralization will make seed users stronger and stronger.

That is to say, as the community develops, the Matthew effect will appear. Traffic, resources, connections, and voice will be concentrated at the top, and their influence on the community will become increasingly greater. This is a phenomenon in economics and sociology, and the community is similar to the online form of society, so the essential logic is similar.

From these representative and well-known community products, we can see the influence of the Matthew effect:

• Douban had undergone several large-scale revisions, which were boycotted by core users in broadcasts and groups. This put a lot of pressure on Abei, so much so that he finally had to roll back to the previous version. In recent years, the criticism that Zhihu has changed has mainly come from its early core users, bringing considerable public pressure. At least it has more or less impact on Zhihu's brand image.
• Kuaishou , which previously emphasized fairness and inclusiveness, differentiated itself from Douyin with its strong social attributes, but recently it no longer mentions inclusiveness. Because the reality is that the more high-following users are, the faster their fans grow.

I don’t think this is because some products are poorly made. On the contrary, they are successful enough to see such a phenomenon. After all, they have really become a community and really look like a city.

Also, we cannot say that the Matthew effect, which makes the strong stronger, is a purely bad thing. We must also look at it dialectically:

• It is beneficial to the community. Seed users support the community to move forward. They are the most determined group of people and the backbone. If we encounter difficulties, we can get through them with them around. The role of seed users in this regard is not available on any community platform.
• The disadvantage to the community is that since seed users have the right to speak and connections in the community and are the center of power, they may become xenophobic and make it difficult for new users to integrate. There is also the possibility of conflicts with platform officials, resulting in boycotts or collective exodus, which will have an impact on the community.

To put it bluntly, it's a game of thrones.

If the rights are divided reasonably and used in the right place, it will benefit both the community and the users. But if you don't get the balance right, it can have devastating consequences.

The conclusion is that the culture of the community comes from the initial seed users, and these seed users are content producers. Their content tone, language system, expression methods, values, etc. form the community culture, and it is difficult to make fundamental changes afterwards.

2. What content to choose

1. Content quality is inversely proportional to social value

Content quality refers to production costs and standard requirements. The higher the cost and standards, the lower the social value; and vice versa.

There are two reasons:

1. Attention allocation
2. Social pressure

(1) From the perspective of attention allocation

If the content quality is very high, users will focus on the content itself and ignore people who produce or interact with the content.

For example, the number of interactions for copyrighted long video content from Youku, iQiyi and Tencent Video is far less than that for short videos or texts. There is even less promotion of social relationships between people. It is because the content is important enough that users focus on the content itself and cannot pay attention to other elements.

Isn’t barrage interactive information? Isn’t Bilibili’s content high-quality?

The atmosphere value of the barrage form is greater than its social value.

For the audience, posting barrages is a channel to vent emotions and express opinions; and the barrages and content posted in the same field are well integrated into more emotional and human content, which is the value of atmosphere. But the barrage highlights the content and is itself considered part of the video content, so there is no human element in it. Without ID, there is no social interaction.

(2) From the perspective of social pressure

Content that is expensive to produce, exquisite, and heavy puts more social pressure on users, just like entering a luxuriously and neatly decorated room, which makes people feel restrained.

As the pressure on people’s behavior increases, the possibility of socializing becomes smaller. Taking live broadcast content as an example, its characteristics are sense of accompaniment and reality. It is easier to build trust by communicating and chatting online face to face.

In order to achieve the above-mentioned feelings of companionship, authenticity, and trust, live streaming must maintain its "roughness".

From the perspective of professional content production, roughness means:

• The production is not sophisticated enough - the background, lighting, angles, etc. are all very random and imperfect, and the sense of real time is very strong;
• The expression is not rigorous enough - there are slips of the tongue, frequent emotional ups and downs, and even heated arguments;
• The process is not reasonable enough - often going in and out of the camera, and the topics are not connected, etc.

Of course, not all live broadcasts pursue a rough feel. For example, brand self-broadcasts and show hosts have to be more well-produced and follow different routes.

Therefore, when designing community content positioning or guiding content benchmarks, you must consider the impact of content quality on social interaction and choose appropriate content formats based on your community positioning.

To extend this a little bit, there is another form of content that has basically disappeared now.

This is the voting and social game feed of Kaixin.com many years ago. The content production cost is very low, it can be completed with just a click of the mouse, and the social value is extremely high.

The question is, a friend asked: Since the social value is high and the production cost is low, why does this form no longer exist? I haven’t figured it out either.

2. The content type should have social attributes

For the community, content without social attributes has little value. For example, weather has content consumption value but no social value.

It is important to choose content with social value so that a community can be built. Or at least a large portion of it is.

Content with social attributes will have some of the following characteristics:

• Diverse perspectives (abstract - value of life, controversial - who is better, Cristiano Ronaldo or Messi)
• Covering a large user group (people's livelihood - oil and vegetable prices)
• High level of attention (social events-epidemic)
• Interested attributes (sports and fitness, pets, games, two-dimensional)

It can be simply understood as topicality. People are willing to pay attention to and participate in discussions, so that social interaction can occur. For example, issues that interest new mothers will be topical. When this group enters a new stage in life, everything around them becomes unfamiliar, and they have many emotions that need to be released, so topics such as parenting knowledge, family relationships, and self-positioning are in strong demand.

Of course, there are also negative examples, that is, topics without social attributes.

Many years ago, I created a community on Maoyan Movie. It was not for film reviews, but for discussing movie-related topics through posting topics and reply threads.

After operating for a period of time, I feel that there is no social demand for discussing topics of interest that are not time-sensitive and not specific to a particular movie. For example, although many people like to watch Marvel and Stephen Chow's movies, if there are no movies or hot topics, there is no desire to discuss such content angles. A community with this type of content as its theme cannot form interactive behaviors, so it is difficult to build a community. Later, Maoyan Movie shut down the community.

3. Content benchmark is the guide

The essence of operation is human intervention, and an important means of intervention is to set benchmarks and play a guiding role.

Let me use the old picture below to explain the effect of the benchmark: There are two ways to manage illegal parking. One way is to set up roadblocks. The downside is that it cuts the road, making the available space much less flexible. The other way is to post tickets, which can be effective if done manually once a day.

In the community, content can also serve as a benchmark.

Through content distribution or community recommendation, benchmark content can be strongly exposed to demonstrate the community’s attitude. The content types that are strongly exposed will then play a guiding role, driving changes in content types from top to bottom.

As a question-and-answer community, Zhihu caused differences with Baidu in its early days. But in fact, as long as you see the types of questions and answers that Zhihu mainly promotes, you will know the difference between these two types of products.

• Questions on Zhihu are open topics, meant for discussion rather than answers, and there is basically no absolute right or wrong, such as "What do you think about question XX?" The questions on Baidu Knows are all concrete, such as "Where can I apply for a temporary residence permit?", "How to make braised spareribs?", etc., which can give specific answers and clearly solve the problems. The problem is solved and the questioner's needs are met.
• The early answers on Zhihu emphasized being systematic and comprehensive, and providing sufficient information, such as background and arguments. So from the formal point of view, they are all long and eloquent articles. As for the answer from Baidu, just give a clear answer. The more direct the better. The number of words does not matter.

When Zhihu first launched the above-mentioned question-and-answer format, it served as a benchmark. Users will know what kind of content this community encourages, and as more people follow suit and imitate, a content ecosystem will emerge.

There are many such cases. For example, in the film review module, if you put 800-word long content in a popular position, core content producers will write according to this length in the next few weeks. When the popular content was shortened to 300 words, it was very interesting that everyone followed suit.

In a short video content ecosystem like TikTok, if a certain shooting method becomes popular and receives massive exposure, then there will soon be followers who use the same method to shoot.

These cases all serve as guiding benchmarks for content. It’s just that the way of content distribution is different, some are manually recommended, and some are algorithmically recommended, but the effect is the same.

3. What platform to build

This part talks about what role the community should play and what value it should embody as a platform.

1. Maximize the efficiency of content production and consumption

The users of the community are active, consume content and interact with each other. The prerequisite for all these is the matching of production and consumption.

According to the logic of the platform, first of all, there must be sufficient and abundant supply; secondly, there must be accurate and efficient distribution to better meet consumer needs.

(1) Douyin and Kuaishou have a positive impact on content, and e-commerce food delivery has a positive impact on products.

Only when there is sufficient and abundant supply can user needs be met, because users need to choose content.

From an economic perspective, only prosperity on the supply side can drive consumption. Therefore, the platform must first push up the supply side, which is the content producer from the community perspective.

Therefore, we need to start from the content supply side. For the platform, it should stand on the side of content producers, provide good tools and services, and build short-term feedback and long-term commercial monetization models.

(2) The platform must also do a good job of content distribution. The form and logic of distribution are the platform’s choice

• In the era of traditional portals, the form of content distribution was that editors placed selected content in locations with high traffic, which was a purely manual distribution model.
• During the community website era, although content was sorted by weighted indicators, human intervention was still possible to give priority to benchmark content that better met community requirements. This was the manual intervention mode.
• For short video products, distribution relies on recommendation algorithms and basically does not require direct human intervention.

Content consumption itself also feeds back to production, thus forming a closed loop. The more active the community, the higher the match between production and consumption, that is, the greater the amount and proportion of content consumed and interacted with. This is somewhat similar to the sales-to-performance ratio of e-commerce, and can also reflect the health of the ecosystem.

2. Feedback from content producers and commercial benefits

I mentioned earlier the importance of content producers to community platforms, so community platforms need to build a feedback mechanism for content producers, as well as a business model for content producers, and make it work on the platform and be recognized.

(1) The feedback system of community products is more important than that of other types of products, after all, they are products with social attributes.

If a content producer publishes content and there is no interaction for a long time, that is, no one pays attention to it, it will give people the feeling that it is a deserted city, just like in real life you say something to a group of people, but no one pays attention to you at all, which makes people feel very frustrated.

This is also the reason why pseudonymous replies are set up in the early stages of community establishment. Of course, most of the fake comments are unreliable, irrelevant to the topic, full of praise and affirmation, lacking diversity, and giving off a fake feeling.

You may have encountered this situation: after you register in a community, you will have a group of followers without taking any action. These can all be considered as feedback system designs, although most of them are fake.

(2) If a community product wants to operate for a long time and get better and better, it must have its own business model.

Only business can last, simply doing it for love is not enough. Therefore, community products need to design commercial monetization paths for content producers, or even just provide monetization tools.

There is no need to pursue the expectation that most content producers can have substantial income, which is unrealistic. There are always only a few people who can make money on big platforms, such as Taobao sellers, Douyin content creators, and Zhihu answerers. The number of people who make money should not exceed 10%.

The making money mentioned here does not refer to producers with income, but rather a situation where the income scale is certain and the cost is basically covered. It is not an accurate and rigorous definition.

Therefore, as long as the industry feels that there is hope and people are willing to invest resources to cultivate this community, then that is the success of the platform. Who can come out on top and make money depends on the efforts of the content producers themselves. There is complexity and uncertainty in the community here, which cannot be controlled by humans.

The logic is the same as for cities. Cities need to provide the "product functions" of living, education and employment, but who can study well, get into a good university, find a good job, get promoted and make money depends on personal efforts and luck. The city merely provides a platform for basic services.

For content-related products, the monetization paths include : advertising, e-commerce (physical goods and virtual services), and rewards.

• For example, Weibo can accept advertisements and has tools such as micro-tasks;
• Zhihu can sell virtual services (Zhihu Live, paid consultation, e-books) and can also inoculate grass advertisements.

Community platforms can choose a monetization path for producers that suits them based on their own business models and provide corresponding tools and services.

3. Construction of community ecology

The ecology of a community is usually not constructed or designed, but formed spontaneously.

Spontaneous formation is the result of successful communities, not the established goal of community managers. However, community managers have the responsibility to maintain and optimize the formed ecosystem.

That is, be a gardener, not a carpenter.

Community ecology is a small scene of the ecosystem. The concept of ecosystem is too broad to be discussed in detail. Here we can make the problem specific:

• Vision
• Rich
• open

(1) Vision refers to the community’s vision and values.

Only with a vision can you know what behaviors to encourage and what behaviors to resist, what content to support and what content to suppress. This should be a standard on which consensus can be reached from top to bottom within the company.

This encouragement and resistance, support and suppression, is mainly achieved through official intervention, which takes the form of explicit attitudes on bulletin boards and intervention in content distribution.

(2) Rich refers to the rich variety of community content and user roles.

Because the community ecosystem grows spontaneously, there will inevitably be areas or categories with uneven development. For categories or user types that should be included but are scarce, we should actively attract them and provide good services, and at the same time, the ecosystem should be prepared to accept and integrate them.

In addition, if there is a demand to expand product categories in product strategy, it is also a manifestation of the pursuit of richness. For example, Bilibili has successfully expanded from the two-dimensional world to more categories, making its community platform more receptive and raising the ceiling for growth.

(3) Openness refers to the community’s ability and willingness to accept.

If the community tends to be over-centralized, a small number of people will have too much say, reject changes in the community, reject the integration of newcomers from outside, have their own social circles and discourse systems, which will make it difficult for newcomers to establish relationships, and result in low content compatibility and overall lack of friendliness, making the community more closed.

The above perspective only stays on the internal side. When building a community, we must also pay attention to competitors, or market conditions.

For example, if what you want to do is already met by other leading products, this is an obstacle that is difficult to overcome unless you have extremely unique resources or advantages.

Because for a community, it is only valuable when it has a certain number of people. Therefore, as a newly established community product, it is a life-and-death struggle to compete with products with tens of millions or even hundreds of millions of DAU for users.