How to operate Toutiao account? There are 3 routines!

Now, apart from WeChat public accounts , Toutiao should be the most popular new media platform. This is a platform with huge traffic that is recognized by everyone.

Its biggest feature is the intelligent recommendation system for articles. In addition to the quality of your content itself and the difference in your account, the key to doing well on Toutiao is to understand its algorithm recommendation rules.

So, what are the tricks behind Toutiao’s massive article recommendations? What can we do about the article recommendation mechanism?

Below I will simply share with you some of the routines.

1. Don’t let your article be “eliminated” at the beginning

When you publish an article on Toutiao, you must overcome a difficult hurdle, which is Toutiao’s de-duplication mechanism. Basically, article de-duplication is the most common reason why the content published by Toutiao accounts has few recommendations.

Think about it, as a content distribution platform, Toutiao cannot always recommend several similar contents to a user for the sake of user experience , otherwise the user experience will be very poor!

It's like you eat a plate of spicy and sour potato shreds, and then the waiter brings you another plate of spicy and sour potato shreds, and then another plate of spicy and sour potato shreds, they are all exactly the same, can you stand it?

Therefore, before Toutiao recommends your article, it must make sure that the content:

● Is there identical or highly similar content in the system?

● If so, is the source of this content the most authoritative, valuable, and most likely original source?

This is deduplication, a process in which the system classifies and compares duplicate, similar, and related articles so that they do not appear simultaneously or repeatedly in the user's information flow .

After you publish an article, Toutiao will first use the de-duplication mechanism to determine whether your article has the opportunity to be recommended to more users.

And in this process, your article may be "deleted".

So how can you prevent your article from dying at the very beginning?

Let’s take a look at a few key items that Toutiao may focus on when eliminating duplicate content.

● Whether the source Toutiao account has the “Original” tag enabled;

● Release time ( first release is important);

● The authority of the source and the number of times it has been cited online.

In addition to this rule for content deduplication, Toutiao also has some other deduplication rules for content:

●Reduce duplication of titles and preview images, that is, the title and cover image should be different.

● Eliminate duplicate content on similar topics.

You must be familiar with this scene: whenever there is a hot social event or topic, the media, self-media , and KOLs rush in, competing to report the details of the event or express their opinions, making your Weibo and Moments "flooded" with all kinds of content related to this matter.

However, what you need is not to see information with the same point of view repeatedly. In fact, reading a limited number of them is enough. What you also need are different entry points.

Here, Mumu will focus on deduplication of similar topics, which is the most relevant to us. It is a generally acknowledged fact that headlines should follow hot topics.
When chasing hot topics, you need to be especially careful about being delisted .

No matter how popular a hot topic is, users’ interest in it is limited, especially if the angle of your hot topic is not much different and the platform has already recommended similar content. In that case, the system will not be willing to recommend it again. Recommending it again would be a nuisance!

Therefore, the angles you use to pursue hot topics must be differentiated. You can split a hot topic into keywords , each of which can be a good entry point, the same topic from different angles.

In particular, the titles of your hot topics should also be clearly differentiated.

Because Toutiao's extraction of this hot tag mainly comes from the title, and then the content. Moreover, in the case of a hot topic, whether users want to click on it is directly related to your title.

If you cannot chase hot topics in a timely enough manner, do not have a sufficiently unique and differentiated entry point, and the content is not particularly different, then don't chase hot topics casually, so as not to waste your efforts.

2. If you want to read more, you must first recommend more.

If you use Toutiao, you should know that the number of article readers depends largely on the system's recommendations. Without recommendations, there will be no reading.

So what factors determine the recommendation volume of an article? Is there any trick available?

I won’t go into detail about some basic influencing factors. Here I will focus on one of its internal mechanisms.

1) How does Toutiao identify your articles?

Toutiao will first identify the features of your article to determine the type and field of content the article is about.

Take keyword recognition for example, the system will extract some words as keywords based on the frequency of their appearance in the article.

For example, if a sports article is about a football match, the high-frequency words that may appear in the article include player names, teams, football terms, etc., such as "C Ronaldo", "shooting", and "Real Madrid".

After determining the keywords of an article, these keywords will be compared with the platform article categories. Your article will be labeled with the category whose keywords your article contains in a large proportion.

For example, if the top keywords of an article are "C Ronaldo", "La Liga", and "Madrid", then the article may be tagged with "football", "international football", "Spain", etc. to complete the initial understanding of the article.

It is important to note that the platform will also identify and classify keywords in the titles . Therefore, it is very important to include representative entity words in the title.

2) To whom will your article be recommended?

Knowing what your article is about, how can Toutiao recommend it to people who may be interested?

That's right. They will also make a basic judgment on users' reading interests based on their big data . Then each user will be labeled with various labels by the system.

For example, in an article that a user reads, the top keywords are: Cristiano Ronaldo, Real Madrid, European Cup, Xiaomi, Meizu, and Apple.

Then, this user may be labeled with "football", "Real Madrid", "technology", "mobile phone", "Mi fans" and so on. Different users will be labeled differently.

Then, finally, when an article with the tags "C Ronaldo" or "Football" is recommended, the system will automatically match it to users with the tags "C Ronaldo" or "Football", which is a personalized recommendation. (Of course, the actual situation is much more complicated than this)

3) How was your article recommended?

To ensure that popular content is seen by more users, unpopular content does not take up too many recommendation resources. When recommending Toutiao articles, they will be recommended in batches to users who are interested in them.

This is very important.

Your article will first be recommended to a group of users who are most likely to be interested in it , and the reading data generated by this group of users will play a decisive role in the next recommendation of the article.

The data includes click-through rate , number of favorites, number of comments, number of reposts, completion rate, page dwell time, etc., among which the click-through rate has the highest weight.

For the first recommendation of an article, if the click-through rate is low, the system believes that the article is not suitable for recommendation to more users, and will reduce the number of secondary recommendations; if the click-through rate is high, the system believes that the article is popular with users and will further increase the number of recommendations.

Similarly, the number of new recommendations for an article is based on the click-through rate of the previous recommendation.

Because of this mechanism of expanding recommendations, if you want to get more readings, you must work hard to maintain reading data in various dimensions (click-through rate, user reading time, number of collections, number of comments, completion rate, etc.) at a high level. These are some of the official standards mentioned above.

Among them, the most important thing is of course the click-through rate, so the importance of the title and cover image is self-evident.

In addition, there is another particularly important point that is often overlooked, that is: the label corresponding to the article.

It directly determines the number of users with the corresponding tags in your article. Some tags correspond to a large number of users, while some tags correspond to a very small group of people. Even if the click-through rate is high, the ceiling is very low.

This is why many professional articles have low readership, while emotional chicken soup, entertainment gossip, and social news are generally highly recommended. This is also why everyone will follow hot topics, because the user group corresponding to the hot topics is very large.

Therefore, when we choose topics, including article titles and the use of keywords in the content, we must base them on tags with a large number of users, so that we can get more recommendations.

Toutiao has also launched a hot word analysis function to help everyone find good tags. Using the hot word analysis function can greatly increase the system recommendation volume and improve the article reading volume.

In addition, there is a little trick. Check the relevant articles of Toutiao on your mobile phone. There are labels of corresponding articles at the bottom . Collect more of these labels and slowly accumulate a label library for your own industry.

You must remember that the tag of your article must match more people.

3Why does the recommendation effect not appear?

As we know earlier, the number of article readings is directly determined by the system's recommendation volume, and the number of recommendations depends on the click-through rate of the previous round of recommendations.

Therefore, the recommendation effect of a single article is not good, and the reasons are nothing more than two major categories: low click-through rate and low recommendation volume.

1) Low click-through rate

Generally speaking, if the initial click-through rate is not high, it will be difficult to achieve a higher recommendation volume. The low click-through rate may be caused by the following reasons:

● The account content has low verticality.

● Problems with the article content itself lead to low click-through rate.

● The title is bland and does not attract readers to read.

● The pictures are not attractive, the pictures in the article are not highly relevant to the content, or the first three pictures are of poor quality.

● The content is too low quality, causing readers to be disgusted or even complain.

Here’s a tip for everyone. When you’re looking for topics, writing articles, and deciding titles, you can first search for the keywords of the topic through Toutiao Media Lab to determine its popularity, and you can also check related associated words.

For example, we compare the popularity trends of the two words "C Ronaldo" and "Henry" within a day. It was found that the attention paid to "C Ronaldo" was consistently higher than that paid to "Henry".

Well, if I were to write an article about these two people, I would emphasize the keyword "C Ronaldo" in both the title and the content. This would increase the number of recommendations and would be more likely to attract user attention, thereby increasing the click-through rate.

Therefore, the publication of articles and the selection of titles on Toutiao can also be operated scientifically. You don’t have to just focus on your own account, you can also look at the overall environment.

2) Low recommended amount

Low recommendation volume directly determines the reading volume. The main reasons for low recommendation volume are:

● If the click-through rate is too low, it will affect the recommendation volume. This needs no further explanation and has been emphasized above.

● The potential user base is too small and the recommendation volume is not high.

If the potential user base is too small, the number of recommendations will naturally not be high. The potential user base is too small for the following reasons:

● The topic is too unpopular or obscure;

● The fields involved are too professional, obscure and difficult to understand, and are somewhat distant from the mainstream masses.

This has been mentioned above. Some article tags correspond to a large number of users, while some tags correspond to a very small group of people. Even if the click-through rate is high, the recommendation volume will not be very high. Therefore, you must consciously choose or add tags with a large number of users.

● Oversupply of content.

This is easy to understand. To put it bluntly, there is a lot of content but not enough resources. Then the system has only two choices: either recommend only part of the content, or recommend all of it but the recommended amount will be less because the same content cannot be recommended to the same user repeatedly.

● The deduplication mechanism will affect the recommendation volume.

If the article is not original or has not been published on Toutiao first, or if it is too similar to other topics, the article may be de-duplicated, thus affecting the recommendation volume. This point is mentioned above.

● Short validity period, affecting the recommended amount.

For articles with short validity, the actual recommendation time is naturally short, and it may not be possible to get a large number of recommendations in a short period of time. So, it’s better to follow the hot topics as early as possible.

Basically, what we mean by increasing click-through rate, completion rate, clear article classification, consistent title, improving content quality, high account verticality, increasing various interaction numbers and subscription numbers, expanding off-site popularity, maintaining posting frequency, etc. are all for the purpose of better recommendation.

Therefore, to increase the number of readings, if the number of recommendations is low, focus on the number of recommendations; if there are few clicks, focus on the click-through rate, and optimize in a targeted manner.

Okay, that’s all for now.

I believe that those who can persist and read this far will definitely gain something.

above.