Understand the traffic algorithms of "TikTok, Xiaohongshu, Zhihu, and Video Account" in one article

Tik Tok, Xiaohongshu, Zhihu, and Video Account are the traffic pools that many brands must compete for. Mastering the traffic algorithms of several platforms can help us obtain as much traffic as possible. Today I will share with you the traffic algorithms of four platforms.

1. Tik Tok

Tik Tok’s traffic algorithm is almost the most complex of all traffic platforms, and of course it also has the largest traffic.

Tik Tok is a typical “label” versus “label” platform.

If you are a user, the platform will break down your focus into about 150 tags based on your usual browsing preferences. The videos you can watch are determined to a certain extent by your user tags. If browsing preferences change, user tags will change accordingly, and the videos viewed will also change according to the tags.

If you are a creator, the platform will form a creator tag based on the content you publish. The number of tags is also 150. If the published content changes, the creator tag will also change accordingly.

After the creator publishes a video, the video will match similar user tags based on the creator's tags. This is the "tag" to "tag" traffic algorithm we talked about above.

After a short video is matched to a user, its data performance will be used to measure whether the video is worthy of further recommendation.

Tik Tok will assess five key data when recommending a single video.

1) Completion rate

Completion rate = viewing time / work time

The higher the completion rate, the more attractive the work is to watch. The passing line for the market is usually around 15%-20%, and a completion rate of more than 40%-50% is already excellent. In order to achieve a high completion rate, the usual way is to create suspense at the beginning or guide users to open the comment area to extend the viewing time.

If it is a new account, it is recommended that the length of the initial video should not be too long. The longer the length, the lower the completion rate, unless the video quality is extremely good.

2) Like rate

Like rate = number of likes / number of views

The higher the number of likes, the higher the number of recommendations. The like rate of the first wave of recommendations should reach at least 3%-5%.

That is to say, for every 100 views, there should be at least 3-5 likes.

3) Message rate

Message rate = number of messages/number of views

The comment rate data has a lot to do with the video type, and it is difficult to measure it with average data, but what is certain is that the better the comment rate performance, the higher the weighted recommendation. Therefore, creators can take the initiative to guide comments in videos or in text or comment areas to increase the message rate.

4) Forwarding rate

Forwarding rate = forwarding volume / playback volume

The forwarding rate does not have much impact on videos that are still circulating in the primary traffic pool, but if you want to break through the traffic level, the forwarding rate is a very critical indicator.

5) Conversion rate

Conversion rate = number of followers/number of views

That is, the ratio of passers-by to fans and the rate of new fans brought by a single video are also key data for impacting high-level traffic pools.

The Douyin platform is a huge traffic pool, the Douyin recommendation mechanism is a fishing net, and the video content is the bait.

If the five key data of your video can achieve good data performance, then the possibility of entering the mid-to-high-level traffic pool and continuing to circulate is very high.

Tik Tok’s traffic pool also has its own rules.

After the video is released, it will enter the cold start pool. The traffic is usually 300-500, which is generally composed of fans + friends + people you may know + a small number of users with tag matching. Because the traffic structure of the cold start pool is the most complex and it is also the most difficult traffic pool to break through, this will test whether your fans are accurate and the content is high-quality. If the key data meets the standards, it will enter the primary traffic pool.

The traffic of the primary traffic pool is approximately 1000-5000. It is also necessary to continue to observe the monetization of the video in the primary traffic pool. If the data continues to pass, it will enter the intermediate traffic pool.

The intermediate traffic pool has more than 10,000 views, and the data performance is similar;

The advanced traffic pool has more than 100,000 playbacks, with no upper limit.

2. Little Red Book

Xiaohongshu’s algorithm is similar to that of Douyin, which is also a “label” to “label” traffic algorithm.

The difference is that based on different user habits, Douyin focuses more on active recommendations, while Xiaohongshu focuses more on search recommendations.

Based on Xiaohongshu's platform positioning, more than 65% of its traffic comes from search, so the search traffic algorithm is more sophisticated. Here we will focus on the logic of the search traffic algorithm.

The matching between search results and needs is mainly the matching degree between core keywords and queries. The specific content displayed in the search results is obtained by analyzing user needs and finding the information that best meets user needs.

The keywords in a note title are of utmost importance, and the official also clearly reminds you: "You will get more likes if you fill in the title."

It can be seen that the title is an important option used by Xiaohongshu to identify content attributes. If you want your notes to get more exposure, the most basic work is to optimize the title.

We should make good use of search keywords, hot word recommendations, etc. to help us find the core words of the notes so that the system can recognize and recommend them to the corresponding users.

Find core words from recommended content

Recommended content includes several aspects, grayed-out keywords in the search box, historical searches displayed on the page, and hot search terms.

1) Default prompt word

Before you click on the search and enter the search term, the platform will recommend default prompt words based on the user's tags. The default prompt words will contain a certain amount of search traffic.

2) Search discovery (hot searches)

Hot searches display the words that have been searched the most recently, to guide users to view some recent popular content and recommendations for topics with high user search volume, which are related to the user's search volume and recent hot topics.

3) Supplement associative keywords

Supplement associative keywords, that is, the user enters partial content, and then the system associates the complete content based on these contents, automatically completes the keywords, and increases user choices by instantly matching keywords and displaying them. I searched for "looking thinner" and the platform recommended several key words related to "looking thinner" to me.

Consider that hot word ranking is the result of comprehensive display. In addition to the number of notes, the popularity ranking of "hot words" may also involve two factors: the frequency of active searches by users, and the popularity of the notes themselves recommended by the system.

After the search, the system performs algorithm matching based on the search terms and displays all the results. If this keyword is a word with a relatively large scope in the relevant category, then you will see some special label words in the upper part of the interface that provide classification and filtering functions. This method will provide a better experience for users who search aimlessly. At the same time, the most popular notes are placed at the front. The purpose of this search result display format and filtering conditions is to narrow the selection range and help users make quick choices.

There are a few things to note about keyword selection:

1) Xiaohongshu’s hot search recommendations are an indicator of the platform’s short-term traffic content; search prompt keywords and popular filters are where long-term traffic is, and they come from Xiaohongshu’s real user data analysis and summary.

2) Be sure to give priority to keywords with low competition, high traffic and relatively accurate keywords, and avoid choosing broad keywords.

3) Learn to reverse deduce key words. After determining the note topic and keywords, you need to reverse engineer what keywords you want users to use to search for your notes, and consider what common keywords you would use to search for this type of note.

4) Reasonable placement of keywords in the note title, text, topic, comments, etc. will help the notes to be included and accurately recommended. Avoid keyword stuffing. Keyword stuffing will be judged as advertisements by the system. If you do this for a long time, your account will be demoted by the system.

3. Zhihu

First of all, regarding search traffic, Zhihu’s search ranking is actually similar to that of search engines. The content needs to be included first before the search term ranking can be improved.

On the one hand, we need to consider the match between the content and the search keywords. The higher the match, the greater the probability of being included. On the other hand, high-quality accounts have higher weights and can obtain higher search rankings. Finally, the popularity of the content will also affect the search ranking. In short, the more popular the content, the higher the ranking.

Of course, the search also involves the ranking of answers under the question. Generally speaking, after the question is included in the search term, one of the highly-rated answers under the question will be captured and displayed. In addition, because users generally refer to more than one answer out of habit, the answer that is naturally ranked first under the question will also have a greater chance of being exposed.

Therefore, if you can achieve a very high ranking for the search term + question, then the traffic will naturally be better; if you cannot achieve both, then at least one of them should be in the front position.

The second is for recommended traffic, which uses Zhihu’s recommendation algorithm to push content to users.

Generally speaking, the recommendation algorithm will first push the content to a small number of people, and then collect feedback data, such as reading completion rate, approval rate, interaction data, etc., to determine whether the content is worth continuing to recommend.

The third is the hot list traffic. The hot list is a collection of real-time popular content on Zhihu's entire site. Its dimensions are mainly calculated based on 24-hour views, interactions and field weight.

If you want your content to be on the hot list, you must have a large number of users in the field participating in the interaction within a short period of time. Once a good volume of voice is generated, the popularity of the content will naturally increase.

Of course, for video content, the distribution mechanism is similar to recommendation, and there is a separate list to support it, which you can refer to.

The fourth is the comprehensive algorithm. Unlike Toutiao, Douyin and other platforms, Zhihu uses the Wilson algorithm, which determines the recommendation and ranking of content based on the Wilson formula based on data such as likes, dislikes, and collections of the content.

u represents the number of approvals for the content, v represents the number of disapprovals for the content, p represents the approval rate of the content = number of approvals/(approvals + disapprovals), and Z is a number related to the weight.

Although the algorithm formula is very complicated, you only need to remember the most important point: the approval rate is more important than the number of approvals, and the disapproval rate is more important than the approval rate.

Unlike other content platforms, in addition to liking and interacting, Zhihu users can also vote against content they disagree with, and the number of votes will affect the answer ranking to a certain extent.

4. Video Account

The algorithms of WeChat Video Account and other traffic platforms are completely different. The distribution mechanism of WeChat Video Account is based on social recommendations and personalized recommendations.

1) Social recommendations

Socializing is the innate gene of the WeChat ecosystem, so for WeChat Video Accounts, the social relationship chain is also very important. For example, the content posted and liked by your friends will be recommended first. If your friends interact a lot with a work by liking, collecting, and posting it, your reading volume and exposure will increase. On the contrary, if non-friends like and collect your work, the exposure of your work will be lower than if your friends like and collect it.

Therefore, the likes and collection interactions of your WeChat friends have a great impact on improving the weight of your work.

In fact, this is similar to the logic of "Reading" and "Like" in public accounts. For example, if you click "Reading" on an article (video), your friends will see this article (video) in the "Take a Look" function of WeChat. If your friend likes it, his friends may also see this work, and so on.

2) Personalized recommendations

This means that the system will use a series of big data algorithms to infer what content the user may like based on the user's daily behavior, activity trajectory, interests, occupation, age and other labels. Because WeChat itself has 1.1 billion super user portraits and various algorithm mechanisms as a reference.

However, as WeChat Video Account is still in the hot start stage, the database is not comprehensive at present. The data sources used are all captured from the WeChat market. The algorithm basically adopts interest tags + positioning + hot spots + random recommendations.

So whether you are posting videos or pictures, adding topics and locations will help with personalized recommendations. This is somewhat similar to Tik Tok's recommendation algorithm, but it is not mature enough yet.

3) Decentralized recommendation algorithm

Although the video account is based on social recommendations, everyone's social relationship chain is limited after all. When a work has been displayed in the complete social relationship chain and has achieved good data performance, the video account will expand the recommendation beyond the social relationship chain. The logic is similar to the "label" to "label" of Douyin, so I will not expand on it here.

The above is the traffic algorithm of Douyin, Xiaohongshu, Zhihu, and Video Account. I believe that after reading it carefully, you will have a new understanding and knowledge of the four major platforms. If you still have any questions and need to communicate, please feel free to contact me.

Author: Zhao Zichen Vic

Source: Zhao Zichen Vic