A recommendation system, if described in a formal way, is actually a function of fitting a user’s satisfaction with the content. This function passes in three dimensional variables: <1> Content: text, pictures, video, etc. Each content has its own label and identity.

<2> User characteristics: that is, user tags. Including various interest tags, preferences, occupations, age, gender, etc. There are also implicit user interest tags <3> contextual features: these are recommended features in the mobile Internet age. Users are on the move at any time, and their information preferences are offset in different scenarios such as work, commuting and travel.

Some professional terms of little red book: little red book notes’ little red book ‘level community topic

Interaction rate includes: likes, comments, favorites, forwarding, concerns

Time spent on notes, rate of play, number of play original or plagiarized

Some topics: beauty makeup, skin care, personal care, dress, food, travel, etc

Red book, a note, can be hit up to 10 tags, such as beauty, skin care, a care and so on

There are several kinds of recommendation: recommend notes that users like (these notes may be posted by strangers) recommend notes posted by experts or friends that users follow, which uses social relations to retain users. Nearby recommendations: Recommend notes posted by people within 20 kilometers of the site

Cold start: The notes just published by the user must go through cold start. If the user’s feedback is good, such as thumbs up, favorites, comments, completion rate and other indicators are good, the expansion mechanism will be opened and the notes will be recommended to more people.

A trending topic or a trending topic

Stay active and update your notes regularly. Send notes regularly. For example, 10 a.m., 1 p.m., 8 p.m., etc., are the time when the platform is more active. Because more xiaohongshu users use Xiaohongshu during these periods.

Xiaohongshu has its own official account, which collects some high-quality notes and then recommends them to everyone.

Fission growth: My notes are seen by my fans, and they think it’s great, and then they recommend it to their fans. So my fans’ fans can see it. This is called fission growth

There are two kinds of notes in little Red book: graphic notes and video notes

Little red book weight mechanism is actually very simple, nothing more than these three points:

1, account registration, the longer the more stable update notes, recommended the greater the chance 2, hanging in the little red book "little sweet potato" level functions, actively complete the task in the growth system can improve the level of "little sweet potato", grade, the higher the account the weight is 3, the more activity in content easier to be critical, and very easy to improve account weights, It will be shown first in the next share.Copy the code

We can look at this problem from a typical xiaohongbook user purchase scenario, as shown below:

Step1: Users are recommended (planting grass) product content through APP information flow, APP message push, [wechat] sharing and other channels. Step2: Users who have "interest" further click the product TAG to enter the "Details page" for in-depth browsing. Step3: "search" the brand associated content in Xiaohongshu successively, and check the word of mouth evaluation from other KOL or ordinary users to verify the product value. Step4: After a series of comparison and reflection, make a "purchase" decision, and even plant grass to friends around to become [brand] communication ambassador. Xiaohongshu content marketing, in essence, is to shorten the conversion time of users on each decision through highly persuasive grass content. Even drive the user impulse purchase, purchase path directly shortened.Copy the code

The community atmosphere recommended by xiaohongshu’s acquaintances makes Xiaohongshu a good place worthy of public praise and promotion. Consumers may not trust advertisements, but they are more likely to trust recommendations from acquaintances.

What is the recommendation mechanism of little red book?

For example, those who like beauty makeup will see the notes of cosmetics, those who like food will see the notes of food, and those who like traveling will see the notes of tourism. If your notes are of high quality, they will be recommended first and the exposure will be greater. When a note has a high click-through rate within a certain period of time, Xiaohongshu will recommend the note to more people, and the note will be more easily placed at the top. Someone asked me if I could buy the data if the interactive click-through rate topped the list. In fact, Xiaohongbook also looks at how long users stay on your notes. In other words, the longer users look at your notes, it will judge the value of your notes. Buying data is usually machine brushed out, no help, but also prone to violationsCopy the code

The recommendation mechanism of little red book

Step1. Xiaohongshu platform will classify and organize our notes according to the content of our notes, that is, label our notes such as beauty makeup, skin care, personal care, clothing, food, travel, etc., and then recommend our notes to users who are interested in such notes.Copy the code

Step2. The system will recommend our notes in the first round. ** counts the audience's likes, comments and favorites in unit time. The specific formula is: Heat = A* number of likes + B* number of comments + C* Number of favorites + D* Number of shares. The coefficients A, B, C and D will be fine-tuned in real time according to the overall algorithm, roughly speaking: Like > Comment > Favorites > Share.Copy the code

If our notes don't get a good response from our target audience after the first recommendation, then unfortunately we can only hope to write better notes next time. Of course, fine tuning of the overall algorithm is not ruled out. Just when our notes become more popular, our notes will be recommended for the second time. If our notes get good user feedback after the first recommendation, then our notes will be recommended to more users who are interested in this kind of notes, which we call the expansion mechanism. The system will recommend our notes to more users, for example, to the home page, which is the so-called popular. This recommendation method is similar to douyin's recommendation method. If you want to better understand this recommendation method, you can take the time to learn about douyin platform.Copy the code

Step3. The system will consider the content and quality of our notes. If the quality of our content is very good, even though the attention is not high, but the platform recognition, then we will still be featured. Step4 / The system will evaluate the weight of the published account itself. For example, the registration time is relatively long, and the release cycle of notes is very stable.Copy the code

The recommendation mechanism of little red book

The first thing we need to do is to understand its recommendation mechanism: how does xiaohongshu recommend notes? This mechanism is explored clearly, can talk about drainage, so today here to tell you about the recommendation mechanism of little red book. 1. Machine algorithm recommendation: Push relevant content according to different needs (user preferences) of different users. There is also the recommendation of the preferential information of the mall (this can be recommended in a large area, not completely in accordance with the user's preference) 2. The recommendation of hot quality content: There is a hot topic that everyone is searching in xiaohongshu. You can put the hot topic in the search box to increase the exposure of the topic. Generally speaking, a note on Xiaohongshu can be up to 10 3 reading extended recommendation: push based on the relevance of content units, also push users interested in the content, extend reading, improve user retention. 4 friend relationship recommendation: There is a "follow" Tab on the home page of Xiaohongshu, which will recommend the followers and friends' updated content to users and retain users through social relations. 5. Proximity recommendation: There is a "proximity" Tab on the home page of Xiaohongshu, which will recommend the content posted by users within 20 km. 6. Search box: The search box generally displays the main topics of Xiaohongshu to promote the thematic content of the platform and attract users to participate in creation. After ** recommendation **, Xiaohongshu also has a cold start system, which will observe whether the fans who are recommended first interact with you (like, comment, forward). If your content is interactive, your notes will continue to be recommended. As a content community, the most important thing is the content itself. The system takes into account the content and quality of your notes. If your content is high quality but not popular, you can still rank high as long as the platform recognizes your content, and vice versa.Copy the code

Several mechanisms of little red book

1. Fair mechanism. Because the amount of small red book notes now is too much, one second the same topic may produce 5-6 in order to avoid the note ranking float too fast, small red book set the ranking delay update, which is why we put notes after ten days and half a month to see the top effect. 2. Monitoring mechanism. The monitoring mechanism refers to the fact that once the notes are sent out, it will monitor the authenticity of likes, comments, favorites, and whether your article is original or copied from others. If there is a problem in one of these dimensions, the platform will not recommend it. 3. Detection mechanism. The detection mechanism refers to whether the article contains guidance, inducement, marketing and other violations. If the violation of notes reaches a certain number, Xiaohongshu will completely restrict the exposure of the relevant content of the brand, and will even ban the brand name if the case is serious.Copy the code

Small red book notes weight rule algorithm

In addition to the content itself, the weight of the account itself is also very important. When your account has a certain number of fans, your notes will be recommended to your fans. When your fans give you a thumbs up, they will be further seen by fans' friends, forming fission growth. When writing notes, Baymax has a suggestion: Because the system will compare the content of your notes and pictures, and some of the content will overlap, so try to write the content of your notes as full as possible, so that the overlap rate will be much smaller. ** If the content of notes is judged to be illegal or plagiarized, the stream will be restricted by xiaohongshu platform, and there is no way to get exposure of the traffic. Now the platform audit is a machine and artificial, because now there are artificial to join the audit you write content must not have violations and contraband words. **3, interaction rate ** The four points like, favorites, forward, comment will improve the interaction rate. **4, label ** be sure to use the label function, in the picture of the content of the label, can also increase the weight of notes. Many users like to see the content below the hot topic, if there is no topic, traffic will reduce exposure, add the corresponding topic in your notes, also can increase the weight of notes. **6, keywords ** title must add keywords and notes inside the content of the best 3 times or so keywords. The content of notes and keywords are very relevant, so as to increase the weight of notes, but do not pile up keywords in the content, otherwise it will be mistaken for the title party, will reduce the score. **7, text length ** If writing notes is very few words, the content quality will not be very high, although the number of words does not necessarily determine the content quality, the notes are up to 1000 words, it is recommended to write 400-600 words, and the content must have a certain value, in order to get the weight of the system to the length of the text.Copy the code

The weight of xiaohongshu account is controlled by 7 key points

1. Originality rate: Most of our notes are composed of Chinese characters. Even if they are really original, some contents may be consistent. In this case, how can we improve the original rate? The simplest way is to make all content original, including pictures. Because the system identifies originality according to the proportion of content, so as long as your content is a little longer, the proportion of parts that are easy to overlap will be reduced, so that the overall originality rate of the whole account will increase. 2. Verticality: What does vertical mean? It simply means focusing on one area. In the little red book, if you are wearing a blogger, only wear a hair content of notes, even if the content is vertical, if you wear a hair today, send food tomorrow, the day after tomorrow hair beauty makeup, the system could not find out your location, you can only according to your notes or most the most similar tags to determine your location, notes domain classification is too much, reduces the rate of vertical you, If you can increase the proportion of content in the vertical field, the weight of verticality will also increase. 3. Content Quality rate: Content quality literally means the quality of your notes in the past. So how do you judge it? That is, the exposure of notes compared to the interaction of notes, that is, comments, likes, etc., the average content quality coefficient of all notes of this account is the content quality rate of your account. 4 account active rate: account active rate is to judge the basic attributes of an account, although the active account can be basically get, but also to see if you can get a better account active weight, so how to calculate the specific account active? 1) Note Posting 2) likes, comments, favorites 3) login and browse weight in this order, if only one of them can get a weight points, if all of them, of course, is the highest score. Because each weight has its own weight upper limit, so do not try to make this part of the weight table, and the rest of the weight do not! 5. Account level: Whether in their own personal home page or other people's personal home page, account level is obvious, which is named and divided by a variety of potatoes, how to upgrade in xiaohongshu account level page has been written in great detail, I will not go into details. 7. What is the use of these key weights? This will determine the exposure of your new note when it is first sent. Cold start. Depending on the account number, everyone's cold start is different.Copy the code