In the past few years, I have been working on the Internet community. I thought I had done a lot and accumulated some experience. I naturally believed in a lot I don’t believe a lot of things, thinking that what is known is everything.” These “beliefs” may not be special, just because I don’t know more, I think it makes sense. Even so, I still have the urge to record these thoughts from time to time.

In the recent period, various coincidences have communicated a lot with friends who are engaged in community product work, those who want to engage in community product work, and colleagues who are not sure whether to do community products , I found that everyone’s thinking about the Internet community and the existing puzzles are almost the same. After repeating some points that have been repeated again and again, I decided to write this article.

This article records my personal thoughts and understanding of the Internet community over the past few years. Although the content has been as concise and general as possible, it is still a bit complicated. The topic of the article is similar to “Building an Internet community from 0 to 1”, “X-word long text: Midfield thinking of the Internet community”, “Why most companies do not do well in Internet communities”, but if one of them is used as the title , I always feel that it is still a little bit off.

I used WeChat to read the “Yu Jun Product Methodology” I read at the beginning of the year. I think the framework of this article is similar to it, and the content is subtly nourished by it, and finally my eyes closed. Face Yimang decided to use “Product Methodology” as the topic. Teacher Xiang Yujun salute.

This article is from WeChat official account:Heiba ​​(ID: HolylandofKalin)< span class = "text-remarks">, author: Xu Xin, head Figure from: vision China

1. What is an Internet community?

(Kaishou) etc.; according to the product form, it can be divided into forums (猫扑), micro Bo (Fanfu), Q&A (Knowing) etc. ; According to the different coverage, the community can also be divided into vertical communities and comprehensive communities.

Vertical communities refer to communities that face a specific field or aggregate specific topics, such as football community live bar, movie community time network. Comprehensive communities do not target specific fields or topics. All community products that can be called “national character” can be included in the category of comprehensive communities, such as Weibo and Douyin.

Under the same category, the boundaries between different communities are not strictly divided, but they are intertwined with each other, and each has its own advantages. For example, Instagram was positioned as a photo sharing community in the early stage, but Story was added later. This kind of short video content.

(From left: Station B, Weibo, Zhihu)

This article mainly analyzes the community from the perspective of vertical community and comprehensive community classification. The reason for this choice is mainly to facilitate the discussion of the two issues like “Internet community keywords of the year” in terms of customer acquisition and growth, breaking the circle and transformation.

①Customer acquisition and growth. The community will face the problem of customer acquisition and growth from birth, budding, growth, and maturity. The community designer should first clarify the vertical category of the community, and then calculate the capacity of the community from the industry perspective, that is, the ceiling of the community’s customer acquisition and growth, and then the community positioning, architectural design, and the dismantling and landing of tasks.

Take the stock vertical community as an example. According to China Securities Journal, as of November 2020, the total number of investment accounts in China is 176 million, which is approximately equal to the number of Chinese securities investors. Can this number be understood as the ceiling of the number of users in the stock community? The answer is no. Although the existing investment account is 176 million, it contains many reasons for various reasons (essentially one reason-loss) Zombie accounts that are no longer traded, and what we need to be clear is the number of active investors.

Although CSDC has not updated the data of active accounts since February 2017, it is estimated based on past statistical data that the proportion of active accounts in total accounts has basically maintained In the interval of (10%~40%), In 2020, the stock market will have different opinions. If the proportion of active accounts is taken as the median of 25% and the number of active accounts is calculated, the value will fall at 44.03 million. This number is the ceiling of the number of users in the stock community at this stage.

The emerging stock community can formulate more reasonable short-, medium-, and long-term growth goals on the basis of this number. For mature stock communities, if the number of active users in the community itself is close to the industry ceiling, then within the original framework, no matter how hard it is, it will be difficult to make a big breakthrough. At this time, community designers naturally have to think about breaking the circle and transforming.

(The total number of China investment accounts and the number of active accounts. Data source: China Securities Depository and Clearing Co., Ltd. website; data interval: May 8, 2015-November 30, 2020)

②Break the circle and transform. For most of the surviving vertical communities, the ceiling is not out of reach. If the goal is to continue to grow, sooner or later, they will face the problem of breaking the circle. Breaking the circle/out of the circle is a term used in the meal circle, which means that a certain star or a certain work breaks through the fixed fan circle and is known on a larger scale. For example, Wutiaoren, a niche band from Haifeng in Guangdong that sings songs in Haifeng dialects, has repeatedly been eliminated in variety competitions, and has repeatedly been brought back to break the circle.

(Five people)

The breaking of the vertical community is essentially based on the vertical field and developing in a comprehensive direction related to it. In the recent period, the term “Broken Circle” appears most often together with Station B. In the early days, Station B was a pure ACG(animation, comics, games) community, featuring barrage and ghost culture, and its core users were two Dimensional lovers.

The current station B has broken the circle-station B New Year’s Eve party, post-wave video, self-made variety shows and drama series(“Rap and Singing New Generation”, ” “The Sky of the Wind Dog”) and other content, incidentally, station B, has broken the original two-dimensional circle, and is well understood by the public. But long before these official self-made content went out of the circle, station B had already begun to turn the rudder.

In July 2018, Li Ni, COO of Bilibili, said: “Station B currently gathers more than 7,000 vertical interest circles, which is a second time in the traditional sense.Meta content currently accounts for 30% of the total number of visits to station B, and diversified interest circles such as life, entertainment, and fashion are already the main content of station B. “From the result point of view, Station B has completed the transition from the second element of life to the “post-wave” life. Its target group is no longer limited to the second element of users, but a wider range of young people. In Q3 2020, all stations of B Station monthly activity reached 197 million, up 54% year-on-year.

(From left: New Year’s Eve Party at Station B, Video of Back Wave, New Generation of Rap)

For vertical communities in a period of decline or in a sunset industry, the problem of transformation is even more serious, even about life and death. From the perspective of direction, the transformation of communities in the recession period is nothing more than two paths, one is to expand into related fields, and the other is to switch to a new track that can be empowered by existing resources. Specific to the practical level, specific issues need to be analyzed in detail.

3. Community and social interaction

Haruki Murakami has a book “When I talk about running, what do I talk about”, the sentence pattern of the title imitates his idol Raymond Carver’s “When we talk about love, what are we talking about” , This sentence also applies to the topics we will discuss next-When we talk about community and social interaction, what are we talking about?

(From left: Haruki Murakami, Carver)

Community and social are a pair of concepts that are easy to confuse, and they are often confused. Community is a content-driven interaction, which solves the problem of connecting people to content, while social interaction is a chain-driven interaction, which solves the problem of connecting people to people. At the level of interactive form, the interactive form of the community is like/comment/repost, etc., while social interaction is IM, instant messaging.

For example, typical community products, in addition to the ones mentioned above, include Douban Group, Instant, Post Bar, etc.; in terms of social products, there are WeChat, which can almost be equated with social in the mobile Internet era, and QQ, Momo, Tantan and other products are named after two-character repetition as if they were negotiated.

By comparing and analyzing existing products on the market, we can find that most community products and social products generally have both social and community elements. For example, Douban groups, in addition to posting and interacting(community), you can also send private messages (social); and WeChat, you can talk to (social), you can also browse and publish the circle of friends (community).

Since there are both community and social elements in a product, how to distinguish a product from a community or a social one? The answer is to look at the product-level page (actually the product focus), the Douban team’s first-level page is a list of topics. The private message function is in the secondary page message,Therefore, it belongs to a community product; the first-level WeChat page is a list of conversations, and the second-level page of the tab is found in the Moments entrance, so it is a social product.

(From left: Douban group topic list, WeChat conversation list)

Second, how to be a community?

1. Community positioning

When we analyze things, we like to trace the origin. Even the universe was proven to originate from a singularity in 1970. So, what is the singularity of the community? I think it is positioning. Community product managers encounter many problems in their daily work, such as the community is fully functional but unable to attract users? There is no difference between the community and competing products? The community is becoming more and more niche? All can be traced back to the positioning of the community.

(Hawking and Penrose, both proved that under reasonable premises, singularities are inevitable)

Community positioning also needs to pay attention to the basic law. Although there are many different styles of product positioning methods on the market, the basic ideas are basically derived from the book “Positioning-Since History” co-authored by Al Ries and Jack Trout.The concept that has the greatest impact on American marketing,” the book describes the specific steps of the positioning work. Simply discussing the method may be too abstract. We set a scene and analyze it in combination with specific cases. The scenario is set as: If in 2011, how to establish a stock community?

First, analyze the entire external environment and determine “who are our competitors and what is their value”. In 2011, after a round of bulls and bears in the A-share market, the overall A-share market was in a unilateral downward trend. The market was in a downturn and its activity dropped a lot compared to 2007 and 2008.

At this time, there are stock bars and Tao stock bars in the communities that investors often browse. Although the names of the two products are only one word difference, they are completely different types of community products.

Stock Bar is a stock community under Oriental Fortune, and its slogan is: A stock-themed community under Oriental Fortune.com. The stock bar is geared towards the broadest group of stockholders and does not impose excessive restrictions on users. As long as they comply with regulations, users can speak freely in the stock bar. Tao stock bar is a “professional investment in A-share stocks and stock exchange forums, exchanges and sharing community”. Tao stock bar holds several real contests every year, gathering the most active short-term players in the market, and now it is in the hearts of investors. Brother Zhao, who is “8 years 10,000 times”, has risen to fame because he is on the stock market. For ordinary users, the purpose of using Tao stock bar, or the value of Tao stock bar to users is to follow the master to operate and explore the latest market themes.

Secondly, avoid the strength of competitors in the minds of customers, or use the weaknesses contained in their strengths to establish the superior position of the brand-positioning. The stock bar is backed by Oriental Fortune.com, and the user minds cultivated by the stock bar can not only speak freely, but also browse the rich UGC(User Generated Content, user produceContent), to understand comprehensive and timely information, announcements, research reports, etc. OGC(Occupationally-generated Content, occupational production content).

The shortcomings of the stock bar are also more obvious. On the one hand, the loose and open communication atmosphere has led to the rampant “Gong Jing” in the community. Even more professional investment-related discussions may be slammed back by its users in various words. As a result, producers of high-quality content are less and less willing to create, and the atmosphere for community discussion is not good.

On the other hand, even if high-quality content producers have a big heart, they don’t bother to raise the bar, or take the trouble to respond, and continue to output high-quality content, the product structure of the stock bar[Strong topic (the topic here refers to stock) flow + weak attention flow] also determines that high-quality content cannot be fully exposed, and it is difficult to form a professional content ecology.

(Stock Bar)

The model of the Tao stock bar real market competition determines that short-term talents in the community will come out in large numbers. The mining and logical understanding of the subject is ahead of the market. Users can follow the bulls in the Tao stock bar brainless operation, gain excess profits, or learn from the best practices. The operation idea of ​​the disk is turned into one’s own use.

The weaknesses of Tao stock bar are also obvious. On the one hand, Tao stock bar relies too much on real-market competitions, which leads to the attracting users are mostly ultra-short-term users, and the user circle is too narrow; on the other hand, the number of users attracted by Tao stock bar is extremely large Most of them don’t have the ability of great people. The result of blindly following the operation is the shortening of their investment life cycle, leading to the rapid loss of platform users.

(Snowball creator rights)

2. Product structure

After clarifying the product positioning, the next step is to build the product architecture. There are many terms similar to “product architecture”: “business architecture”, “strategic architecture” and “information architecture”… These words are like twin brothers, often appearing together, and sometimes they are misused. It gives people a sense of chaos after cutting. In fact, these concepts do have intersections, but they are different.

  • Strategic Architecture/Business Architecture: The top-level architecture, including business logic and business operation mechanism.

  • Product architecture: The product function and system architecture that connect strategy and information. The specific role is to classify and integrate different functions around goals.

  • Information architecture: the most front-end architecture, including basic interaction design, information and other presentation layer architecture.

The three types of structures are progressive and interrelated. The intersection of strategic architecture and product architecture lies in the combing of business processes, and the intersection of product architecture and information architecture lies in the design and interface presentation of product systems. This section mainly discusses the product architecture. The content related to the information architecture will be integrated into the later content production, content consumption and other parts.

Currently, there are not too many books and materials related to product architecture design in the market. When this happens, people often choose to use the “magic oil” in the field of product managers.——The five elements of user experience go to the product architecture design, but the result may be unsatisfactory.

Jesse James Garret’s “User Experience Elements” is a must-read book for product managers, it provides a sufficiently abstract and easy-to-understand product design idea, that is, from the strategic level, scope level, structure level, framework level, There are five levels of presentation to design products, but the five elements of user experience are more about explaining products from a macro level. At the actual landing level, the guiding significance of this idea is not strong.

However, through the sorting of limited information, a complete set of product architecture design methods can still be abstracted.

First, enumerate all the functions of the product, and classify and aggregate from the perspective of goals or satisfying a certain type of user needs. For example, the immediate “me” channel of the interest community aggregates the user’s own dynamics, photo albums, files, footprints and other functions based on user-related information and production content.

(Immediately-“Me” Channel)

Secondly, sort out the relationship between upstream and downstream functions to form a process. Communities generally do not involve too complicated upstream and downstream process relationships. From my own experience, the most complex process in the community may be the review process. As for the frequent feedback from users who cannot find the black function, it is not the reason why the function arrival process is too long or the process is chaotic.p>

There are three main sources of OGC, namely self-made, outsourced, and crawler. Which method you choose depends on the community’s own positioning, tonality, and pocketbook. For the community, OGC is the most stable and controllable, and can be used to determine expectations for users, but the cost is the highest. In addition, OGC is generally distributed in full, which can provide community users with common topics and enhance interaction.

Although UGC and PGC are both derived from users, as far as the platform is concerned, the former is mainly responsible for increasing the activity, and the latter is used to increase the professionalism of the platform. Although most communities have both UGC and PGC, it is difficult to pay equal attention to the two. It must be top-heavy(The head may be UGC or PGC ).

For example, as mentioned in the community positioning section above, the stock bar’s activity is significantly higher than other communities, but there are fewer high-quality content. Although Xueqiu has more high-quality content, its activity is difficult to further increase. Imagine that if the snowball chooses the overlord to make the bow hard, choose to pass the relaxed review, or insert more pan-financial content in the information flow, to attract more “not so smart investors”, then it is likely to lead to platform culture The dilution of foreign currency, and the emergence of bad money driving out good money. The content discussed in this section is mainly about the production of UGC and PGC, and does not involve OGC.

In summary, community designers mainly work in the following aspects in terms of content production.

The first is to define quality content. Imagine such a scenario: you download a new community app, after opening it, the interface is blank, only a line of small characters in the middle: there is no content for the time being, you can try to speak~ At this time, is it you? If you leave, you either try to speak, but you don’t know what to say. The result may be that you didn’t post anything for a while, or you may type a “1” and send it out.

Therefore, when constructing a content production system, the first thing to do is to provide a high-quality content template as a basic reference for user content production. In addition, the definition of high-quality content is also the definition of community culture and values. If the positioning is accurate, the subsequent work will be twice the result with half the effort.

The second is to produce high-quality content. In the early days of the establishment of the community, who will produce high-quality content will be a problem that has to be faced. One method is production by operations or editorial staff, although the quality of such content can be guaranteed, or up toong>Automation logic. For example, the machine automatically calculates the popularity based on the interactive data, and content that exceeds a certain popularity value is marked as high-quality. You can also use the human-machine collocation logic to filter out the high-quality content prepared by the machine. Select the pool, and then manually select high-quality content to improve efficiency.

There is no absolute difference between the above methods. The pursuit of efficiency can be full of machines, and the pursuit of precision can be full of labor.

Fourth is to encourage high-quality content. Incentive mechanism can provide content producers with the motivation to continue creating, and it is an important part of the sound operation of the content ecology. There are two main types of incentives: spiritual and material. At the practical level, they are divided into the following three categories.

  • First, the traffic is tilted, and the exposure of high-quality content is increased through operations such as topping, active recommendation of information flow, and expansion of distribution scope;

  • The second is to provide differentiated functions, such as column, live broadcast and other permissions, as well as medals, plus V and other certifications to enhance the identity of high-quality content producers on the platform;

  • The third is creation and monetization. In addition to basic operations such as rewarding and opening advertising permissions, it can also help high-quality content producers create online courses. (Knowledge ), publishing physical books (snowball) etc.

(From left: Zhihu, Xueqiu)

4. Content review

Content review should be the biggest concern for organizations planning to enter the community. Before specifically discussing review, we need to clarify the conceptual scope of content review. Generally speaking, there are two types of audits, one is compliance audit and the other is quality audit. The content review we usually talk about mainly refers to compliance review, that is, review whether the content violates national laws and regulations. Quality review is the judgment of content quality, which is related to content operation, so this article will not expand.

Compliance review is the bottom line of the community. Only content that has been reviewed for compliance can be displayed on the front end. It may be because the cost of maintaining a manual review team is too high, or there may be mistakes and omissions in manual review. Many people’s questions about content review generally focus on whether machine review can replace manual review?

As far as the current situation is concerned, machines still cannot replace humans. The reason is very simple. The machine is not as smart as everyone thinks. It is like a top artificial intelligence robot like Siri. If you ask a question, it will be slightly out of it. Given the range, its “IQ” will drop off a cliff, and it will only tell cold jokes like a second fool.

From what we have learned before, Currently, companies involved in UGC are basically using manual review of content. What the machine does now is more of the first step of keyword filtering. If certain keywords are mentioned, the content must be illegal. What the machine does is to filter these content directly, without having to go to manual That step.

For the content review process, the following three steps are necessary: ​​machine review, manual review, and manual quality inspection. Machine review is as mentioned above, the main function is keyword filtering, filtering out 100% offending posts, reducing the workload of manual review; the secondary function is content marking(For video review, the machine will also randomly cut frames of the video), the dimensions of the tags are various, and the goal is always to improve the efficiency of manual review.

When the content reaches the manual review step, what the reviewer sees is the content processed by the machine. The reviewers can quickly make judgments on content compliance by browsing the original content and the auxiliary judgment information provided by the machine. Manual review will inevitably cause errors. Therefore, the general community will also increase quality inspections, that is, randomThe machine extracts the audited content for re-examination to ensure the compliance of the content.

5. Content consumption

According to the content distribution logic, the content consumed by users in the community can be divided into three categories: topic information flow, follow information flow, and recommended information flow.

The topic information stream is a list of content aggregated by topics, such as a list of posts by a Douban group. The sorting method of topic information flow is relatively simple, no more than the latest posting sequence, the latest update/comment sequence, and the popular sequence. The first two are easier to understand. The popular order is a sorting method in which recent popular posts are filtered through the popularity algorithm and sorted by the popularity value.

Some people may be curious about specific popular algorithms. In fact, the formula “heat value = (initial heat value + interactive heat value) * time decay factor” is commonly used in the industry. For a more specific example, the above formula can be refined as: popularity value = (initial popularity value + (reading amount * x + forwarding amount * y + sharing amount * z + comment amount * a + like amount * b + favorite rate * c + trigger The amount of attention*d+the opening amount of the author’s personal homepage*e))/(e^(k*(t1-t0))).

Among them, the “initial heat value” is related to author attributes, content attributes, and entity attributes, and “(e^(k*(t1-t0)))” is a time decay factor based on Newton’s law of cooling.

Following information flow is a content list composed of information posted by people that users follow. Following is a behavior triggered by the user. Theoretically, following all the information in the information stream, the user wants to read it. Therefore, the early information flow is sorted by the most recent release time. However, as the number of people the user follows increases, the user’s watch list page becomes redundant.

Even if the user doesn’t click on the content of a follower for a long time, the user rarely takes the initiative to cancel the follower’s attention, but this situation may result in that the content of the user’s real interest is not exposed , The overall stickiness of the platform is reduced.

As a result, Weibo took the lead in sorting the follow information stream by default, from the most recent release time to a smart recommendation sequence based on multi-dimensional calculations such as the user’s preference for followers, content quality, etc. Although there was a lot of scolding, However, Weibo insisted on this change with the spirit of tens of millions of people. In 2020, the WeChat public account has also made similar optimizations.

(From left: Weibo information flow, WeChat official account article flow)

Recommended information flow. In the early days, the community had only one content distribution method of topic information flow, and the generation of RSS subscription relationship made the attention information flow appear. The rise of SNS social networks, the arrival of the whole people since the media era, the explosive growth of the amount of content, and the emergence of personalized recommendation information flow.

In recent years, recommendation information flow has been a hot topic in the community, but not all communities are suitable for recommendation information flow.

First of all, personalized recommendations need to be built on the basis of massive content, and the more personalized the recommended content, the higher the requirement for the amount of content, For example, a community has 10,000 posts a day , The content that can be recommended according to the two-eight law, there are about 2000, if there are 100 content tags, the average content of each tag is about 20, and the user who is tagged with a certain tag is on the mobile terminal, only You can browse through all the personalized content with one refresh.

Secondly, the purpose of the recommended information flow is to guide users to immerse themselves, but the premise is that the recommended content has a low consumption threshold. If the community content itself is professional, and a certain amount of time and energy has been spent after consuming an article, it is difficult for users to stay in this recommendation stream. For example, the main content type of Douyin, now known as the black hole of time, is entertaining short videos. Users don’t need to use their brains to scan the videos one by one. While immersed in it, time has passed.

Again, the recommendation information stream provides unknown and uncertain content. In the product architecture design, the recommendation information stream is often placed on the discovery channel, which is essentially counter-efficient.

As for the slightly mysterious recommendation algorithm, it is actually not complicated. In one sentence, it means “things are gathered together, and people are divided into groups”. Things gather together based onSimilarity recommendation for content attributes, such as (categories, tags, keywords, etc.) at the author or content level, using the TF-IDF method to calculate the content The similarity between the users is recommended to the user with high similarity content that he has clicked on.

Dividing people into groups is a collaborative filtering recommendation based on user behavior. The basic idea is to use user behavior as a feature, calculate user similarity and item similarity, and perform information matching.

A typical example is that both user A and user C have read content 1 and content 3, and user B has read content 2. Looking at user similarity based on reading behavior, user A and user C are more similar. Therefore, after the user A newly reads the content 4, the content 4 can be recommended to the user C. For more detailed algorithm recommendation content, please refer to Xiang Liang’s “Recommendation System Practice”. It is rumored that the most primitive recommendation algorithm logic for byte beating comes from this book.

3. Write at the end

Finally finished. I didn’t plan to write so much at the beginning, but accidentally the outline was enlarged, and I didn’t hold my arms when I wrote it. But even so, the content related to the operation of community products, such as the establishment of the content operation background and the construction of the community atmosphere, is still not covered in this article. Please find time to write later(actual It shouldn’t be written again/manually).

Because of this article, the time after get off work, weekends, and New Year’s Day holidays in the last month have all been taken in. Sleep is severely lacking, and now the only one What I think is that I can finally go to bed earlier today. May the force be with me.

Main reference materials:

[1] “Country China”, Fei Xiaotong, People’s Publishing House

[2] “Communities Across Borders”, Xiang Biao, Life·Reading·Xinzhi Sanlian Bookstore

[3] “Yu Jun Product Methodology”, Yu Jun, CITIC Publishing Group

[4] “Positioning”, Al Ries, Jack Trout, Machinery Industry Press

[5] “Behind the scenes products: creating breakthrough product thinking”, Wang Shimu, Publishing House of Electronics Industry

[6] “Operational Way: Redefining Internet Operations”, Li Mingxuan, China Machinery Industry Press

[7] “Content Algorithm: An Efficiency System Turning Content into Value”, Yan Zehua, CITIC Publishing House

[8] “Recommendation System Practice”, Xiang Liang, People’s Posts and Telecommunications Publishing House


This article is from WeChat official account:Heiba ​​(ID: HolylandofKalin)< span class="text-remarks ">, Author: Xu Xin