Xiaohongshu's content marketing logic

When we write articles and post short videos, we use content as a benchmark to attract readers. However, in an environment with such an overabundance of content, the path of content marketing has become very narrow. We who do content marketing must have read a lot of content marketing experience summaries, but after reading so much logic, we still can’t do content operations well. Today we will first study the content marketing logic of Xiaohongshu. Let’s take a look.

I have read hundreds of experience summary articles on Xiaohongshu's content marketing, and found that no one seems to be able to explain the logic of Xiaohongshu's content marketing.

After thinking about the reasons, there are probably two reasons:

• On the one hand, content marketing practitioners have no experience working on UGC platforms, so they lack knowledge of the content distribution mechanism of UGC products;
• On the other hand, people who work on UGC platforms rarely do content marketing.

Because one is the product department's business and the other is the marketing department's business.

I happen to have experience in both of these areas, and like giving roses to others, the fragrance lingers on my hands, so I wrote this article generously.

My working logic remains the same: as long as I master a few basic logics, I can quickly find content marketing strategies based on different goals and in different environments through a small amount of data verification.

1. Where does Xiaohongshu’s traffic come from?

When you open the Xiaohongshu APP, the first section you see is the most important path for content consumption.

This path usually accounts for about 80% of the traffic on the entire platform, and on some platforms, it can even be as high as over 90%. The main source of traffic for Xiaohongshu is the “Discover-Recommendation Page” . Other major behaviors of users are searching, following, and being in the same city, and these paths account for less than 20% of the remaining traffic.

The following is my predicted traffic distribution chart for Xiaohongshu, for your reference.

This explains the questions that many big bloggers have: why some of their posts only get forty or fifty likes even though they have one million followers.

Am I being restricted?

In fact, you are not restricted in traffic, but the data of your work is not good and it has not been recommended on a large scale. The people who read your work are mainly your fans who click through the follow-up entrance.

Currently, the average fan reading rate of WeChat public merchant accounts is 1%, and the fan reading rate of Xiaohongshu will be lower than 1%, and is expected to be between 0.1%-0.3%, which is equivalent to a Xiaohongshu blogger with 1 million fans.

For an article that is not recommended, the number of reads from fans is about 1,000-3,000. Based on an estimated 1% interaction rate, you will get 10-30 likes, collections and comments.

2. Three core reasons why Xiaohongshu content is recommended

Three core logics determine whether your content will become a hit, and to what extent.

Open rate, interaction rate, search result open rate.

1. Open rate

The general logic of the recommendation algorithm is to recommend content to a small number of users, such as about 1,000 users, and decide whether to recommend it to more users based on the proportion of these users opening the content. This proportion is the open rate.

Content with a high open rate ranking is recommended to more users .

We at Xiaohongshu do not provide data on the number of recommendations (number of exposures), so we have no way of knowing the opening rate. But there are other ways to analyze the works with high open rates.

2. Interaction rate

When content is recommended to a larger group of people, such as 10,000 users, assuming an average opening rate of 10% and an interaction rate of 1%, there should be 10 users interacting. At this time, the system can determine whether the content is worth continuing to be recommended to a larger group.

Based on the open rate and interaction rate data, decide whether to make a larger recommendation.

3. Search result open rate

Some content does not have a high opening rate when it is first released, so it naturally cannot get many recommendations.

However, when they appear in the user's search results, they are more likely to be clicked by the user, and will continue to be recommended to targeted groups of people and get better display positions in the search results.

This explains why some content has a good readership but a very low interaction rate.

Some content has no readership in the early stages, but suddenly becomes popular after a period of time.

Xiaohongshu officials have given average data, and the current opening rate is 11% and the interaction rate is 3%.

4. How to study the click-through rate issue?

• We can disconnect the phone from the Internet, then open Xiaohongshu, and find that the discovery page of Xiaohongshu shows that the content is empty;
• Keep the Xiaohongshu discovery page, reconnect to the network, and load the data on the Xiaohongshu discovery page;
• Disconnect your phone from the Internet again and swipe down.

I found that Xiaohongshu pushes 10 pieces of content at a time.

• There is 1 content with 0 likes, 1 content with less than 10 likes, 3 content with 10-100 likes, and 5 content with more than 100 likes;
• After repeated refreshing, the network was disconnected, there were 2 pieces of content with 0 likes, 1 piece of content with less than 10 likes, 2 pieces of content with 10-100 likes, and 5 pieces of content with more than 100 likes.

Repeating this operation can reveal the content distribution pattern.

• For 10 pieces of content, examine the click-through rate of these 10 pieces of content. The average click-through rate of each of these 10 pieces of content should be 10%;
• Since 5 of the 5 are hot content, the CTR may be around 10%, and the CTR of the 2 growing content should be significantly higher than 10%;
• Of the remaining three low-rated pieces of content, two may have a click-through rate of 1%-5%.

Then the click-through rate of your content must reach at least 13%, or even 20%, in order to explode at the click-through rate level.

If your content’s click-through rate is less than 11% when competing with 10 other pieces of content, it will definitely not be recommended.

There are some ways to test the click-through rate level of your content in advance.

For example, in the era of public accounts, some big Vs test which title is better by posting multiple titles to a fan group of 500 people for voting.

5. How to study the interaction rate issue?

Some people believe that different types of interactions have different weights.

For example, CES = number of likes (1 point) + number of favorites (1 point) + number of comments (4 points) + number of followers (8 points).

This theory may not be correct. If we follow this theory, the content of big bloggers will have obvious advantages.

However, many content data of many big bloggers with hundreds of thousands of followers and millions of points are very mediocre.

Since Xiaohongshu’s discovery page recommends content from bloggers you follow, I tend to think that:

High-stickiness fans help increase click-through rate and interaction rate, which in turn helps increase the number of content readings.

Therefore, the idea of ​​randomly finding people to follow each other, or randomly finding people to like and interact with, which was popular a few years ago, will probably become invalid.

These little tricks will interfere with the efficiency of Xiaohongshu's content recommendations, and the platform can easily optimize the algorithm in this regard.

In addition to recommending content from bloggers you follow, Xiaohongshu’s discovery page also recommends content from the same city (or nearby).

Therefore, there is room for exploration of these two ideas to improve open rate and interaction rate.

The following figure summarizes the interaction rate data of a blogger on the Internet:

6. How to study the opening rate of search results?

We think about the problem from the perspective of search users. Users search for keywords with certain needs, look for the content that best meets their needs in the search results, and start reading. When reading some content, they develop a desire to interact.

The keyword search results will display the following content: the first 22 characters of the title + text, the header image, nickname, and number of likes.

The difference from the discovery page push is that the content of the discovery page only displays the title but not the text.

The search results show that popular content is mixed with recent content and previously unpopular content.

The keywords in the search results must be in the first 22 words, and it is best if the keywords are at the front.

The search population is more precise, and precise content should be provided to these precise populations to encourage them to open and interact.

3. How does Xiaohongshu push different content to different users?

Generally, UGC platforms will recommend content based on user behavior data, user basic feature data, user relationship data, etc.

Behavioral data mainly include search behavior, browsing behavior, interactive behaviors such as liking, collecting, commenting, and following, tags, and topic preferences.

The basic characteristics of users are gender, age, geographic location, mobile phone model, school, company and other identity information.

Relationship data includes the content you follow, such as the content they read, like, favorite, and comment on.

For a new user, when Xiaohongshu knows very little about you, Xiaohongshu will guide the user to choose channels of interest and recommend bloggers. It will also recommend possible matching content based on information such as channels of interest, bloggers of interest, gender, geographic location, age, school, company, etc.

• When users browse and interact with some posts, Xiaohongshu can summarize the users' browsing preferences, interaction preferences, and tags or topics of interest to recommend content.
• When users use search, Xiaohongshu can recommend content related to the user's search keywords and the search results viewed by the user;
• When a user follows a blogger, Xiaohongshu can recommend content based on the user's preference and the excellent content of the blogger.
• When users collect and like a blogger’s content, Xiaohongshu may recommend other excellent content from this blogger in the future.

We tested searching for keywords using a relatively new account with almost 0 followers and 0 followers. For some keywords, the proportion of related content recommended on the discovery page could reach 25% a few hours after the search.

We tested the search keywords using an old account with about 100 fans, 100 followers, dozens of posts and a lot of search behavior. A few hours after the search, the recommended content on the discovery page still accounted for about 10%.

Xiaohongshu’s response to user search behavior is obvious.

Readers will be able to summarize other rules by further research from other angles.

4. How to implement content marketing strategy for Xiaohongshu?

Understanding the logic of the above three dimensions can provide a basic basis for us to formulate Xiaohongshu’s content marketing strategy.

Through a simulated case, I want to inspire readers to think.

For example, let’s say we want to do content marketing on Xiaohongshu for a new consumer product with an annual sales target of 100 million yuan, and the unit price of the product is about 100 yuan.

After research, it was found that 50% of the target population would refer to opinions when purchasing, and Xiaohongshu is the number one source of opinion reference.

The gross profit margin of this product is only 60%, and the ROI needs to reach above the break-even line of 1.67 for the investment to be sustainable.

If you follow the general approach, find a top blogger with more than 100,000 followers to promote your product.

Assuming that a blogger has 500,000 fans and 5,000 traffic from fans, the average traffic brought by content recommendations is estimated to be 20,000, the number of interactions is about 800, and the content conversion rate is 3%, then the sales will be 75,000 yuan.

The blogger quoted a cooperation price of 30,000. According to this ideal data, the ROI is 2.5. However, considering the average return rate of 15% for this product, the conversion rate is less than 3%, and the average traffic is relatively low.

This investment strategy may result in little profit or even a loss.

Now I’ve decided to adopt another content marketing strategy .

20% of the popular content on Xiaohongshu is created by amateurs with less than 100 followers, and 50% of the purchases of this product come from active keyword searches.

I set up a three-person content delivery team and delivered 3,000 pieces of amateur content, covering the three main keywords of the product, so that 20% of the top search results for each keyword were for the product.

The hit rate of amateur content reaches 5%, and the average reading volume of a single hit is 10,000. The long-tail search traffic can be increased by another 20%, which means that each hit content will have another 2,000 search traffic.

The purchase conversion rate of active search traffic is 3%, and the purchase conversion rate of passive reading is 1%.

The average creation cost of each piece of amateur content is 200 yuan, and the half-year maintenance cost of the delivery team is 200,000 yuan.

The calculation shows that the sales volume is 1.5 million (passive reading) + 900,000 (active search), and the ROI = 3. The ROI is higher than the first method, and the estimation of this strategy lowers the purchase conversion rate and increases the average creation cost of amateur content.

In this improvement strategy, the direction of traffic delivery is based on:

First, find the traffic source of Xiaohongshu and optimize the traffic.

I mainly focus on Xiaohongshu's search traffic + referral traffic, and ignore the fan traffic, so as to significantly reduce the average quotation for content delivery. By building a content team and increasing the explosive rate, I can obtain traffic comparable to that of top and mid-level bloggers.

This will achieve the goal of significantly improving ROI while obtaining the same amount of traffic.

In content optimization, the basis is

Second, the three core logics for recommending Xiaohongshu content are used to optimize the content. In terms of content layout, it is based on.

Third, Xiaohongshu’s content push rules are used to arrange the distribution of content that users see.

5. How to evaluate the data of Xiaohongshu influencers?

One difficulty in promoting content on content platforms such as Xiaohongshu is that conversion data is difficult to evaluate.

How much sales did the delivered traffic bring?

The inability to directly obtain conversion rate data is the main reason why many brands dare not promote content.

However, if only a part of the conversion effect can be monitored by providing links and directing traffic to brand channel stores, this result can easily be challenged by colleagues.

When it comes to content seeding, we often face questions from marketing directors and CMOs whose industry knowledge is still stuck in the past 10+ years, as well as contempt from colleagues who don’t understand the industry.

For the same model on the same platform, the conversion rate difference is not large, and the average conversion rate of the strategy can be obtained through a small sample conversion rate test.

The reading conversion rates of mainstream content platforms tend to be consistent. As long as you can measure the number of people who actually read the content, take the average conversion rate of this platform, or take the average conversion rate of all platforms, which is about 1%, you can roughly estimate the effect of the delivery.

The actual reading volume of WeChat public accounts is the most accurate. Within a period of time, an article is only counted once no matter how many times the same user reads it.

The number of Weibo readings is more like exposure. As long as the content appears in the user's information flow, the number of readings is calculated regardless of whether the user reads it or not.

So the data we see on Weibo is often tens of millions or even billions of views.

The playback volume of Bilibili is calculated by looking at the same user within a period of time. It is counted as one reading only when a certain proportion of the video is loaded. Basically, it can be understood that one playback volume is a real user reading.

The reading numbers on Xiaohongshu are usually more authentic.

If you are worried about the falsification of reading numbers, you can estimate it based on the number of interactions.

The proportion of interaction numbers on general platforms is approximately between 1% and 3% of the number of readings. If it obviously exceeds this range, the data may be untrue.

Of course, if the content is really great, the number of readers will obviously exceed this range.

By applying the above logic, how can we achieve the goal of increasing followers? How to produce popular content? How to formulate content seeding strategies according to local conditions? Due to the limited length of this article, we will have the opportunity to discuss this further in the future.

VI. Conclusion

① The two main sources of traffic for Xiaohongshu are recommendation traffic and search results.

Focusing on the most core traffic sources can make your content marketing strategy more effective.

② The logic behind Xiaohongshu’s content recommendations mainly consists of three aspects: high opening rate, high interaction rate, and high search opening rate.

If you meet any of these points, you can produce popular content with a readership of 10,000 to 100,000 through an amateur account.

If you meet any two of these points, you can produce explosive content worth 100,000 to millions of dollars.

③ Xiaohongshu recommends content to users based on three main criteria: classification based on users’ basic characteristics, based on users’ search and interaction behaviors, and based on users’ relationships.

By grasping any point in the content layout, you can form a unique content marketing strategy, which can effectively capture your target audience, form relatively advantageous public opinion, and thus encourage users to place orders.

④ How to evaluate the grass-planting data of Xiaohongshu?

Let your boss take a look at this article. Don’t cling to sales data. The core of content promotion is to pursue the number of readings. Estimate the conversion rate based on the number of readings and make a budget. You will not lose money if you follow my conservative estimation of data.

Author: Jiang Liu

Source: Jiang Liu.