Zhihu product disassembly and analysis!

"I'm in the United States, just got off the plane, there are a lot of acquaintances in the circle, so I'm going to hide now." Do these jokes from the outside world about Zhihu really mean that "Zhihu" has become "Bihu"? In fact, in terms of rational analysis, among UGC platforms, Zhihu’s content is almost the highest quality, and it also provides the best atmosphere for rational discussion. Whether Zhihu is a good thing or not depends more on how people use it. In this article, the author experiences and analyzes the Zhihu product from the perspective of a content acquirer and shares it with you.

1. Product Positioning

Product positioning is the cornerstone of a product and the soul and core of product work. Only by understanding it can you be targeted in subsequent analysis. Zhihu is positioned as a knowledge exchange and sharing community, and its underlying core is a high-quality question-and-answer community.

1.1 Zhihu slogan

A product's slogan can reflect a company's positioning of the product, and Zhihu's slogan can be divided into four stages of iteration.

2012 "Discover a Bigger World"

2013 "Share your knowledge, experience and insights with the world"

2018 Restart "Discover a Bigger World"

2018 “If you have any questions, go to Zhihu”

How to understand Zhihu’s slogans at different stages and their expectations and positioning for the product? Here are some personal opinions.

1.1.1 2012 “Discovering a Bigger World”

At that time, the number of users was relatively small, and the community was full of high-quality users. There was a good atmosphere for answering questions seriously. Using this slogan could attract more people to "discover" → tap into users' curiosity.

Explanation from Zhihu founder Zhou Yuan:

"What is the bigger world? It is the unknown. Each of us needs the knowledge and experience of others. Even those who have achieved professional success only have a small part of the knowledge that they are most familiar with in their own field. When we choose what to do, where to go, and what to believe in, we all need similar experiences and insights in the minds of others. What Zhihu is doing is to connect knowledge with people more effectively.

1.1.2 2013 “Share your knowledge, experience and insights with the world”

In 2013, Zhihu opened registration and the number of users grew exponentially. The increase in the number of users led to different problems for different users, so Zhihu needed a large amount of UGC or even PGC to help itself build a large content platform centered around "knowledge".

At this time, the slogan became Zhihu, calling on everyone to share their experiences and insights, encouraging users to actively participate in content production and discussions; → tapping into users' desire to share.

1.1.3 Restarting “Discover a Bigger World” in 2018

From encouraging individuals to share their insights to returning to "discovering a bigger world", my personal understanding is that after Zhihu's user base reached a certain scale, the quality of content in the answer area and comment area began to decline uncontrollably, and users began to complain that "Zhihu is becoming more and more watered down" and "Share the story you just made up". Returning to "discovering a bigger world" shows that Zhihu hopes that users "share valuable information, but more importantly, read more, learn more, and understand more about the world they didn't know before."

1.1.4 2018 “If you have any questions, go to Zhihu”

The slogan "Discover a Bigger World" which returned in the last stage was updated to "If you have any questions, go to Zhihu" soon after. My personal understanding is that Zhihu took advantage of the high exposure during the World Cup to let more new users who have not used Zhihu realize what kind of value Zhihu can provide to them. The slogan is based on "Go to Zhihu, ask Zhihu, answer Zhihu, read Zhihu, search Zhihu, and browse Zhihu" to introduce the core message "If you have any questions, go to Zhihu", which benefits the general public, lets more people know the value of Zhihu, enhances its overall recognition, and increases the user base.

From a personal perspective, I prefer the slogan "Discover a bigger world", which reflects that Zhihu's brand positioning is a high-quality and professional question-and-answer platform; it has a power that makes people believe that you can discover a world you don't know on Zhihu.

2. User Analysis

Every product has its corresponding users. Understanding the user types of the product and their corresponding needs can make it easier to choose entry points in the analysis.

2.1 User Profile

There is a lot of information about the profile and statistics of Zhihu users on the Internet, so I won’t repeat the analysis here, such as iResearch Consulting. The following are some screenshots:

2.2 User Requirements

According to users' needs for Zhihu products, they can be divided into the following categories:

1. Eager to gain industry knowledge or experience on Zhihu and achieve self-improvement.
2. Willing to share my professional skills and knowledge in certain fields on Zhihu.
3. Peep into other people's lives, emotions, and other experiences.
4. Share your life, emotions and other experiences and read some interesting, fun and curious questions and answers to kill time.
5. Enterprises and organizations are active on Zhihu to increase their visibility and influence.
6. Use Zhihu to drive traffic to your own official accounts and other self-media.

Comprehensive analysis shows that they can be divided into two categories: content acquirers and content producers. Based on my own usage, I will focus on analyzing from the perspective of content acquirers in the future.

3. Product Structure

Sort out the overall structure of the product to facilitate a general understanding of the product from a global perspective.

The structure of Zhihu products is mainly divided into homepage, members, questions, messages, and mine (Note: if you join a circle, the question section will shrink to the upper right corner of the homepage and become a circle). The product structure is shown in the figure below (due to image size limitations, many details are compressed):

4. User Flow

As mentioned above, this article mainly analyzes the product from the perspective of content acquirers. There are generally two types of content acquirers:

1. First, there is a clear content that you want to obtain;
2. Second, there is no clear content that you want to obtain;

Based on these two types of users, the main operation process is drawn as shown below:

5. Core Functions

The core functions of being a content acquirer are:

Content search mainly includes: search function and category navigation function.

The purpose is to help users find the content they want to watch faster and more conveniently.

Content reading mainly includes: content display and content purchase.

Both of these functions have strong in-depth exploration points, including answer ranking recommendations, content pricing, membership management, etc.

Content interaction mainly includes: question function; community interaction functions, such as likes, thumbs-up, collections, comments, etc.

This section sorts out the core functional modules that a content acquirer will be involved in. Although only a few functional points are listed, each of them has deep logic and points that can be further explored.

This article will not analyze all the functional points for the time being, but only select a few small points for analysis and explanation.

6. Functional Analysis

6.1 Content/Answer Display

From the author's personal perspective, when using Zhihu to obtain content, what I hope to get is high-quality, professional content or answers to questions.

There is a survey online: "Would you use Baidu or Zhihu to search for questions?"

A large part of the answers were "Use Baidu for simple questions and Zhihu for professional questions."

From this we can see that as a content acquirer, most people hope to obtain high-quality, professional content and answers through Zhihu, so the sorting mechanism for questions and answers is a functional module that the author is very concerned about.

So what is the answer sorting mechanism of Zhihu?

In today's world of information overload, sorting is the current mainstream means of filtering the Internet and quickly identifying high-quality content.

Zhihu has a set of algorithms called "Wilson" interval algorithms, the formula of which is relatively complicated:

Here we mainly analyze which dimensions the answer ranking corresponding to the algorithm is related to: the basic principles of Zhihu algorithm ranking.

Answer ranking is unique, that is, all users see the same answer ranking;

Adjust the ranking in real time based on user feedback and sort by approval ratio;

User weight affects the ranking, and its influencing factors include:

• The voting weight of a high-powered V is higher, and one of their votes may be equivalent to N votes of ordinary users, but the weight is based on the field rather than the global weight; the weight is calculated together with the user's historical approval, historical opposition, and unhelpful number in the field of related questions (topics) in the past; this enhances the quality of answers to questions in professional fields;
• Answering is equivalent to casting a vote for yourself, so the answers of users with higher user weights will be ranked higher in the initial stage;
• Anonymous answers are not counted in the user weight.

The impact of algorithms on sorting:

• The ranking of answers depends more on the approval ratio, and the introduced algorithm only plays a corrective role;
• When the total number of votes is small, the score will increase rapidly if the answer receives approval votes. The larger the total number of votes, the slower the score growth rate.
• For answers with a high number of approval votes, their scores will drop rapidly when they start to receive disapproval votes. The greater the total number of disapproval votes, the slower the rate of decline.

The impact of this answer sorting mechanism:

• Answers with strong controversy will be suppressed, such as witty, inflammatory, and low-quality answers from big Vs...
• Ordinary users may be more willing to answer questions, because even if a big V comes with 100 likes, the negative votes will quickly make the answer sink;
• The role of big Vs in professional fields is more obvious because their answers carry weight. If their answers are professional and high-quality, they are more likely to be ranked high.
• The answer is more in line with the ideas of Zhihu's overall users, which pleases more users, but will hurt users who output niche content;

Therefore, if all Zhihu users like witty and provocative answers, then this answer will also be ranked very high. This is why many users say that Zhihu is becoming less and less popular.

On the contrary, if all Zhihu users like high-quality, professional answers, then these answers will be what everyone sees at the forefront; what type of content and answers we can see depends on ourselves.

6.2 Evaluation Mechanism

Zhihu has a rich evaluation mechanism: reply, agree, disagree, collect, thank, share, not helpful, report answer...

The content display mentioned above is based on these interactive evaluation mechanisms of users.

Such a large number of operations can allow the platform to understand the characteristics of an answer more accurately through data. Excellent answers can get a lot of exposure, while those with low quality or involving advertising content will be directly folded, so that users can save a lot of time without having to see those poor quality answers;

Zhihu review:

Talking about Zhihu's evaluation mechanism, people currently do not have a strong sense of responsibility for their votes of approval or disapproval. They all agree or disapprove based on their own preferences. This reminds me of a remark by an early Zhihu user: "At the time, everyone was very cautious, and even when they clicked the "like" button, they were very reserved. The reason was that even if I agreed with your answer, I was not professional enough to judge, so I could only express my gratitude." It is precisely because of these people who take knowledge sharing seriously that Zhihu was positioned as high-quality and professional in the early days!

In fact, Zhihu's high-quality content is still there. It just depends on how users cast their precious votes. If all Zhihu users like high-quality and professional answers, then these answers will be what everyone sees at the forefront. What type of content and answers we can see depends on ourselves.

The author has some thoughts and suggestions on the content display and evaluation module, and everyone is welcome to discuss them together.

Personalized answer sorting:

Can we consider sorting answers based on user tag attributes? The reasons for this are:

• First, sorting answers based on user tags allows different users to see the content they want to see, so that users not only feel that the content is reliable, but also feel that Zhihu understands them, which improves user stickiness; at the same time, it can also reduce the rate of negative reviews from users;
• Second, more and more users are complaining about the weird answers on Zhihu. Clever and provocative answers can get so many likes and push them to me. What I want to see is high-quality, professional answers that solve the pain points and needs of these users.

Answer Tags:

Users can label answers with labels such as "professional", "popular science", "tricky angle", "witty", etc.; take these labels into consideration when sorting answers; or combine these labels with the previous suggestion to form answers with different labels for different users.

6.3 Question and Answer Function

The question-and-answer function is a pillar function of Zhihu. In the above user process, when users do not find the content they want, they will choose to use the question function.

We can imagine the scenario in which we generally ask questions: ask a question → wait for an answer.

Zhihu has added many small details in this process. The process is shown in the figure below. In each link, CNOOC has more detailed logic:

Input question:

Automatically match and recommend similar questions while entering questions;

• This function can avoid too many identical issues on the platform, which would lead to bloated and redundant platforms.
• This helps users find similar questions and answers faster, reducing the time cost of user operation processes and waiting for answers; and also reduces the number of similar questions found by other users when searching;

Add a topic:

When adding a topic, relevant topic tags are automatically matched, and users are required to select at least one topic tag:

• Topic tags make it easier for users to invite others to answer in a targeted manner;
• Recommending invitations and distribution by topic tags can improve accuracy and the quality of answering questions;

The author believes that this is a major contributor to the platform's ability to ensure high quality and professionalism of content.

During the process of posting questions, we will determine whether similar questions exist and guide users to directly view similar questions and answers. At the same time, we will prohibit users from posting existing questions repeatedly to ensure the quality of platform questions.

Invite to answer Invite to answer process The platform recommends invitations and distribution based on the topic tags of the user's question, and also provides a one-click invitation function:

• Reduce the time users wait for answers, allowing them to get replies in a shorter time, improving user experience;
• Recommended users have corresponding answer tags, such as historical answer topics, fields, similar questions, etc., to ensure the professionalism of the answerer;
• For the invitees, it can satisfy their psychological need for respect and motivate them to produce high-quality content;

Based on the above analysis of the question-and-answer module, the author believes that the invitation-to-answer mechanism is a function that benefits the questioner, the answerer, and the platform, and is also one of the characteristics of Zhihu; it can form a good closed loop and a positive flywheel effect.

Topic tags are a basic function that ensures that users get satisfactory answers and that the platform’s content is of high quality and professionalism.

Questions:

From the above analysis, we can find that the "topic function" is a very important function for the platform, but the author cannot find the entrance to the topic square on the platform. Why does Zhihu hide this function? Can you find the following page on the platform?

Recommended points:

Open the entrance to the topic plaza to make it easier for users to find it.

Here are the reasons:

• In addition to its important function in the question module, topic tags also play a very important role in other modules;
• Product goal: Open topic plaza, when users pay attention to different topics, can obtain user tag data, improve user portrait information, facilitate subsequent question recommendations, guide member section content, etc., to improve user stickiness and member conversion rate;
• User needs: Users can choose the topics they are interested in through the topic square, instead of having to search to find related topics, because there is still a large part of the content that only has a general direction but is not very clear about what they want to know;

6.4 Home page navigation categories

When users enter Zhihu without a clear goal, they usually check the product's navigation categories to see if there is any content that interests them. This time, we will focus on experiencing Zhihu's homepage navigation categories.

It is mainly divided into four sections: attention, recommendation, hot list, hot topics (new)

• Focus on the information flow of the module or the timeline pattern;
• The recommendation module uses algorithm distribution;
• The hot list module pushes real-time hot content, which has certain information attributes;
• Hot topics are sections that integrate the events that people are most concerned about at the moment, such as the current fight against pneumonia;

The contents of several modules are too complicated and redundant, which can easily distract users' attention.

Currently, the logic for opening Zhihu's homepage is that the page will be opened under the same tab as the last time you exited the app, and will be opened under the same tab when you start the app again (except for hot topics). It seems that Zhihu is wavering in the weight between algorithm recommendation and social distribution, and wants to strike a balance between the two.

Detailed comparison of the differences between Toutiao products and Zhihu:

Toutiao products do not have the genes of excellent content creators and initially relied on scraping content, but they use powerful algorithm distribution capabilities, making their recommendation and distribution technology their moat.

Zhihu is indeed different. It has had excellent content creators since the very beginning, and the content created by these excellent creators is the moat of the platform. At present, the relationship chain between content creators and content acquirers on Zhihu is very thin. If the recommendation algorithm distribution is the main focus in the future, the connection between creators and content consumers will become thinner and thinner. In the current situation where high-quality and high-quality answers are not taken seriously, it is a relatively large harm to creators.

Based on this consideration, it is recommended to use social distribution as the main method and algorithm distribution as the auxiliary method; increase the connection between content creators and content consumers, activate user activity, and increase user stickiness.

6.5 Functional Analysis Summary

This conclusion focuses on analyzing some of the feelings and thoughts during the experience of Zhihu from four modules: content display, evaluation mechanism, Q&A function and homepage classification;

After years of development, Zhihu has become more and more diverse in product form. There are still many modules that can be analyzed in depth, such as the payment module, the perspective of content producers, etc. The content of these modules will be discussed later~

7. User Feedback

When testing a product, of course you have to look at the user reviews, especially the negative reviews~

Here are some screenshots, including App Store reviews and some comments under the analysis Zhihu article:

The users' comments in the figure are consistent with the content in our analysis above; many people think that Zhihu is becoming less and less popular, but there are still many people who think that they can find high-quality and professional content on Zhihu, which mainly depends on how users use it.

As a platform, in addition to considering how to ensure the high-quality content positioning of the platform, it also needs to consider how to allow different users to see the content they want to see.

Perhaps the "Circle" function recently launched by Zhihu also has this kind of layout. While submitting socialization, it also reduces the social radius and allows different people to develop their interests in different circles.

8. Final

In this article, the author shares with you the original intention of product disassembly, as well as the experience of disassembling the Zhihu product. After sharing Zhihu's product positioning, user analysis, and product structure, the author deeply experiences the product from the perspective of a content acquirer, and conducts in-depth analysis and thinking on some of the functional points.

The analysis is inevitably subjective, and everyone is welcome to discuss and exchange views.

Finally, I hope that Zhihu will not forget its original intention as it continues to develop, and will not abandon those users who are truly asking questions and answering questions. It will maintain its professional level while coordinating the needs of users at different levels.

Author: Xiao Wu Product Notes

Source: Xiao Wu Product Notes