Zhihu and Xiaohongshu traffic and algorithm logic

In many people's impressions, Zhihu's main user profile is male. Although the male-female ratio of Zhihu is slightly different in terms of search data, the male-female ratio of Zhihu users is 53:47, which is basically balanced. This kind of "counter-common sense" usually produces dividends from cognitive blind spots. Brand owners of skin care, beauty, maternal and child products and other categories targeting female groups on Xiaohongshu have foresightedly planned and profited from them. Therefore, when we choose a platform for marketing, we must fully understand the traffic algorithm mechanisms and target groups of different platforms. Today, I will talk to you about the different traffic compositions and algorithm mechanisms of these two platforms. I hope you can gain something from it.

Zhihu

1. Traffic composition

Zhihu's traffic mainly comes from the following four aspects:

1) Search traffic: According to incomplete statistics from big data, girls mostly use Xiaohongshu when searching for questions, while boys mostly use Zhihu when searching. Many male users regard Zhihu as a handheld encyclopedia, searching for various knowledge and questions and discussing various real-time hot issues on Zhihu. In addition, with the support of search engines, search traffic has become an important part of Zhihu's traffic.

Advantages of search traffic: accurate and long life cycle.

Disadvantages of search traffic: It does not have strong explosiveness and is easily affected by fluctuations in site rankings.

Format: Q&A, articles, videos, etc.

2) Recommended traffic: Recommended answers are the priority column on Zhihu. Many Q&A articles become popular because of the community's recommendations.

Advantages of recommended traffic: large traffic.

Disadvantages of recommended traffic: It has high requirements on the quality of article content and feedback of background data.

Format: Q&A, articles, videos, etc.

3) Hot list traffic: Zhihu’s hot list has an overall ranking and a ranking in each field. It is a popularity ranking of the hot content of the entire site, which can attract the attention of users across the site in a very short time.

The advantage of hot list traffic: large exposure.

Disadvantages of hot list traffic: high requirements on topic content and article quality, fierce competition, and easy to be pulled down.

Format: Q&A, articles, videos, etc.

4) Other traffic:

In summary, Zhihu's video content output is a form of output that the platform has focused on supporting recently. Users can enter it on the homepage, or find it in the recommendation column and hot list. The disadvantage of video traffic is that the push is not accurate enough and relies on community support. In addition, other columns of Zhihu (live broadcast, circle, ideas, attention column, etc.) also have some traffic, but because they basically have no lead-in entry, they are not the focus of Zhihu's marketing.

2. Algorithm mechanism

1) Search traffic

When it comes to the ranking mechanism of Zhihu search traffic, it has an algorithm similar to the search engines we are familiar with. The content needs to be included in the site first before the ranking of the search terms can be improved.

First of all, the degree of match between the published content and the search keywords is an important factor in determining whether the content will be included in the platform. The higher the matching correlation between the two, the greater the possibility of being included; then there is the weight blessing of high-quality accounts. The higher the weight, the higher the ranking; secondly, the popularity of the published content is also one of the factors that affect the search ranking. The higher the popularity of the content, the higher the ranking will be.

Another thing is the ranking of answers below the search question. After a question is included in the search term, the system will grab a highly-rated answer to the question and display it in a push notification. Generally speaking, when users search for questions, they usually refer to more than one answer, so the answer that is naturally ranked first under the question has a greater chance of being exposed. Therefore, when we create content, we must ensure that both search terms and questions are ranked very high, so that there will be relatively good search traffic; if we cannot achieve both, then at least we must ensure that one of the search terms and questions is ranked in a high position.

2) Recommended traffic

Zhihu platform uses an in-site recommendation algorithm to recommend content to users on the site. The platform pushes the captured content to a small number of users, and then collects data from these users (such as reading completion rate, like rate, number of comments, etc.) for analysis to measure whether the content is worthy of wide-scale recommendation.

3) Hot list traffic

The hot list is a popular collection calculated by Zhihu based on the 24-hour views, interactions and field weight of the real-time hot content of the entire site. If you want your content to be on the hot list, you must increase the natural popularity of the content in a short period of time. The increase in natural popularity requires a large number of users in the same field to participate in the interaction (comments, likes, collections, etc.) to generate a good volume.

• Comprehensive Algorithm

Different from the algorithm mechanism of TikTok, Zhihu uses the Wilson formula to determine the recommendation and ranking of content. This algorithm calculates a corresponding data based on the number of likes, dislikes, and collections of the content. U refers to the number of likes, V refers to the number of dislikes, P refers to the approval rate {approval rate = likes/(likes + dislikes)}, and Z is a number related to the weight. In fact, those who are familiar with the Zhihu platform know that in addition to liking and interacting, Zhihu platform users can also vote against content they disagree with, and this is very critical. The number of votes against will affect the normal ranking of the answer to a certain extent. Therefore, the core point of this algorithm formula is: the opposition rate is more important than the approval rate, and the approval rate is more important than the number of approvals.

Little Red Book

The algorithm mechanism of Xiaohongshu is basically similar to that of Douyin, and it is also a typical "label" to "label" big data push machine. Since users of the two platforms have different usage habits, there are certain differences. The Douyin platform is more inclined to active recommendations, while Xiaohongshu is more inclined to search recommendations. Furthermore, since more than 65% of Xiaohongshu’s traffic comes from on-site positioning of search traffic, the search traffic algorithm is relatively more sophisticated.

1. Algorithm logic

The logic of Xiaohongshu's search traffic algorithm is mainly based on the matching correlation between the core words of the search keywords and the recommended content. By analyzing user needs, we find the content that best matches user needs and display it in the search results.

1) Default prompt word

When we want to search, we often find that there are gray search terms in the search bar, even if we have not entered the search term. This is because the system will recommend default prompt terms related to the user's search term, and there is a certain amount of search traffic in the default prompt terms.

2) Search discovery (hot searches)

Hot searches recommend and display the keywords that have been searched the most times in recent time, thereby guiding platform users to browse topics and content with high recent search popularity. Therefore, the algorithm mechanism of search discovery is related to the user's search volume and recent hot topics.

• Supplementary associative keywords

The key to supplementary association is that when we enter partial content in the search box, the system will expand the complete search terms based on the entered content and automatically complete the search content. In this way, the search content matched through content association increases the user's choices conveniently and quickly. For example, if I search for shoes, the drop-down bar will recommend related keywords such as "shoes for men" and "shoes for women".

The ranking of Xiaohongshu's popular keywords is the result of a comprehensive display. The popularity of "popular keywords" is mainly determined by two factors: the number of times the content itself is recommended by the system and the frequency of active searches by users on the site. After the user searches, the system matches relevant content based on the user's search terms and displays all search results to the user. However, when the search keyword is a common word in the relevant category, some professional label words that can be classified and filtered will be displayed in the upper half of the interface. Such a function will provide a better user experience for users who have no goals. At the same time, the system will display the most popular content. The purpose of this search result presentation and filtering function is to help users narrow the selection range and quickly select the search content.

• System recommendation algorithm

1) Machine algorithm mechanism: recommend matching content based on user search keywords.

• Reading extension recommendation: Based on the related content browsed by the user, the user is recommended topics of interest so that the user can read further.
• Social recommendation mechanism: content published by accounts that users follow, similar to the information flow of WeChat Moments.
• Nearby recommendation mechanism: Based on the user's location, content within a 20km radius is recommended to the user.
• Editor's recommendation: The official account will recommend the high-quality notes included.

Therefore, on Xiaohongshu, as long as users can publish high-quality content, they can get traffic recommendations from the platform even if their fan base is not large.

The above are the different traffic structures and algorithm mechanisms of Zhihu and Xiaohongshu. I hope that everyone can make effective marketing based on the above knowledge points when using these two platforms.

Author: Silk Road Praise

Source: Silk Road Academy