Why is an article that was ravely praised on Zhihu being criticized to death on Toutiao?

What would it be like if you had an article that was highly praised on Zhihu but badly criticized on Toutiao?

What would it be like if you had an article that received more than 1,700 likes on Zhihu, but only 1,500 views on Toutiao?

If you have had such an experience, you will find that although the user feedback on the same article on Zhihu, 36kr and Toutiao is different, there is a pattern to follow. From another perspective, each product has its own personality and temperament, which may be like this:

Zhihu encouragement: Good article! Good and in-depth article! Like!

36kr's complaint: What a crappy article, how dare you post it, editor, have you been bribed? What you said is all right?! Why don't you go to heaven?

The howling style of Toutiao: Are you sick? If you are sick, you should get treatment! Has Xiaomi offended you, idiot! Baidu is an idiot no matter what it does!

Don’t believe it? Let us dissect the same article for you.

In April this year, we pushed an article titled "The Pain of BAT: Robin Li's Anxiety and Baidu's Dilemma" on our official account. As a rule of thumb, such in-depth articles should be published simultaneously on multiple platforms in an effort to maximize external communication channels. The platforms we selected include Zhihu, 36Kr, Toutiao and other content communities. These communities all have comment functions, which allow us to observe user feedback more intuitively:

You can think about what kind of platform this is. If you soak it frequently, you can easily tell the difference.

Yes, this is Zhihu!

The picture above is a comment from 36Kr. From the comments, we can feel a momentum of "I, Zhao Tianri, do not accept it!", and there are also opinions such as "What you said is quite right, but it doesn't matter, the editor takes the money!"

This is the roaring style of the headlines. You can even suspect that the people who leave messages basically don’t read the content. They just start making their own comments after reading the title! In fact, they also don’t read the content, they only read the title. They do this not to understand other people's views, but to vent their own emotions.

This made me think of a question: Why does the user feedback on the same article with the same title and content on three platforms become completely different? Even if you only look at the comments, you can't imagine that it is the same article.

So is this a product problem or a user problem? Or we can think more deeply: is it that the product chooses the user or the user chooses the product?

Generally speaking, everyone seems to think that since many users go to the app store to download various products every day, it should be the users who choose the products. But in fact, downloading is just the beginning. What really forms a long-term relationship between users and products is "retention." Retention can better determine the relationship between products and users than downloads.

So what causes retention? It’s the product itself. So, from this logic, it is the product that chooses its own users. Through its services, positioning and temperament, the product selects a group of users, serves them and pleases them.

To put it in an analogy, the product is more like a magic mirror in a fairy tale, and the user is like the queen in it. The feedback that users get when they see themselves reflected in the product is like a magic mirror telling the queen that she is the most beautiful woman in the world. It is foreseeable that if one day the magic mirror tells the queen that she is not the most beautiful woman in the world, the queen will definitely abandon the mirror.

At this point, I need to make a brief definition. Generally speaking, only social products, content communities, and media have the ability to select users. Tools , services, and e-commerce categories are more often chosen by users based on scenarios and needs. We say that a product is defined by users, scenarios, and needs. The core of social products and community products is users.

Back to the question I asked at the beginning, why are the user performances of the three content platforms so different?

In order to better explain this matter, we need to distinguish the three media mentioned above: Zhihu, 36Kr and Toutiao all seem to be content products where we push articles, but in fact there are many differences if you look closely. Based on content classification, Zhihu is a typical content community, while 36Kr and Toutiao are media platforms.

Simply put, the products of media platforms and content communities will implicitly conform to the following logic:

The content of media products comes from PGC (expert produced content), and focuses on the number of readers and the amount of dissemination. For example, for a product like Toutiao, the core value is accurate push and efficiency first. The content community is UGC (user-generated content), and its focus is on increasing recognition and personal influence. For example, a product like Zhihu needs to attract the same users by expressing its temperament through the product.

According to this logic, the purpose of media is to get people to discuss. If there is no discussion, what is the point of media? If there is a content community, people must share the same interests and flatter each other. If you spend a whole night writing an article and then get scolded, who would want to continue writing?

From the perspective of the mechanism for recommending content to users, Toutiao and Zhihu are basically the same:

The difference is that the circles represent different things. Toutiao is a label, and Zhihu represents people. Articles on Toutiao are matched based on users' interest tags and the article's keywords and titles, and then pushed to users. The quality and direction of the content are not taken seriously by users.

Therefore, users of Toutiao basically start to get angry after reading the headline. Zhihu is different. Zhihu pushes information through the relationship chain. The establishment of the relationship chain is based on the result of users screening relevant content according to quality and interest. Therefore, after receiving content push, Zhihu users will generally read it carefully and then comment.

At this point, we can basically understand why the same article will have completely different feedback on different platforms.

However, this question does not end here. Next, I will talk about how the product selects users.

In fact, many products will screen users through various means (such as product design, operation , content display, etc.) in order to find users who truly match their goals and shape the characteristics of the product. Here are a few examples to illustrate this.

The registration (now authentication) process of bilibili has sparked heated discussions among many product people. If a user wants to register (authenticate) on Bilibili, he must first answer more than 100 questions, and most of them must be answered correctly in order to be recognized as a qualified "second dimension" and successfully registered.

Bilibili raised the registration threshold through the two-dimensional issue, ensuring a very pure two-dimensional cultural atmosphere in the early days of Bilibili.

Zhihu’s registration was by invitation only back then. By using the invitation system, the quality of Zhihu users is ensured, bringing more high-quality members to Zhihu. At the same time, early high-quality users also began to form the early Zhihu product culture: respecting answers, respecting knowledge, and being willing to share and help.

36Kr is one of the leading technology blogs in China in terms of content display (although many well-known international technology blogs, such as techcrunch and medium, have long been displayed on the "reading list"), winning by quality rather than large amounts of content. By comparison, we can find that emerging content media platforms such as 36Kr are basically completely different from media such as Sina Technology and Toutiao.

You can do a multiple choice question:

If you only update 20 messages a day (36Kr) or give you 500 messages a day (headlines), which one would you choose?

The frequency of push notifications also allows us to intuitively feel the degree of importance a media platform places on content, and the two are obviously different. 36Kr's editorial processing and columnists' exquisite writing are more likely to make users finish reading, just like entering a boutique and looking at every piece of clothing. Toutiao is a news supermarket with so much content that users can only read the headlines and have no time to read the content in detail.

This is very similar to Sina's news supermarket model back then. But the difference between the two is that Toutiao also introduces third-party content sources, such as "Toutiao Account", and the accuracy of its recommendations to users is much higher than that of Sina. Although both are ways of producing PGC, in fact the operating modes are different. 36Kr is a professional editor, and Toutiao is a platform that hopes to use machines to promote the distribution of PGC, which leads to completely different user feedback on this product.

According to my observation, Toutiao has multiple push processes. The core of the first push is the label, and the core of the remaining pushes is user feedback. The first push allows technical capabilities to be mapped and matched with user tags and content keywords, helping to quickly recommend content that users like to other users. The second and third pushes are based on user feedback from the first or previous push. Such feedback helps to push the content to users with similar tags more accurately.

For example, Xiao A likes to read articles tagged as technology, Internet, and hardware. When the machine pushed the article to Xiao A for the first time, it pushed an article titled "Is Microsoft crazy to acquire LinkedIn for more than 26 billion US dollars?" Xiao A opened the article. Therefore, the machine will think that little A likes this article more, so it will push it to little B, little C, and little D whose tags are also "technology, Internet, hardware". If it finds that they also click on and read the article, it will continue to push it.

This is the working logic of Toutiao. If you are interested in learning more, you can leave a message at the end of the article. I will write an article about data-driven content distribution in the future to explain in detail how the interest engine works.

However, Toutiao's interest engine determines push notifications based entirely on the popularity of titles and article keywords. Good titles (topics) become more and more popular, while bad titles but high-quality content will be ignored. The efficiency of Toutiao is like poison, making people constantly optimize titles and topics while ignoring the quality of the content itself.

Having said that, I think it is necessary for WeChat subscription accounts to reflect on themselves. In terms of content production, subscription accounts have both PGC and UGC. Its original intention was to push content socially, limit the number of pushes, and improve the quality. However, as the number of subscription accounts increased, the quality became uneven and the reading rate continued to decline, so an awkward "pinned" function came into being.

However, doing so neither brings efficiency nor necessarily quality. As users lose their focus, it has become difficult for people who operate public accounts to achieve rapid growth in followers. Users also feel that there is too much content, and their product experience is affected. This is what I once said, the content is overflowing for the users.

When the content is overflowing for users, it is necessary to share the profits with content producers and conduct graded operations. Only the original label is not enough! Time is running out for subscription accounts! The "paid subscription account" thing needs to be done quickly!

The above content is purely speculation. If the guess is correct, we will not be responsible!