Tik Tok’s new wave of “grayscale testing”

In order to ensure the commercial monetization efficiency of the main site's short videos, the long and short video sections are cut and split, which is inevitable as time goes by and the content ecology becomes richer. This article describes the relevant content and I hope it will be helpful to you.

Lujiu Business Review recently discovered that the Douyin APP has begun testing the "Dislike" button and the independent "Long Video" section.

From a "level" perspective, dislike is just a function button on the short video page, while long video is a separate function partition, the two should not be discussed together.

But before that, the long video APP "Qingtao" in the grayscale test of Douyin has been changed to "Douyin Selection", and all interactive data are completely connected with Douyin itself. Now the built-in "Long Video" section not only has the same interface as Douyin Selection, but also surpasses the former in terms of user experience with the help of the functions of Douyin itself.

To some extent, this also represents an attitude of Douyin: instead of "working hard to achieve miracles" and creating a new APP, it is better to continue to segment Douyin's traffic pool and carry out refined operations, which is also in line with the overall industry trend of "reducing costs and increasing efficiency".

The "Dislike" function has come to the fore from behind the scenes, undoubtedly acting as a "switch" for traffic segmentation. Douyin, which originally adhered to the "short video +" positioning, is no longer rigid and has begun to seek change.

1. Grayscale testing, from behind the scenes to the front stage

The position of long videos in the Douyin ecosystem has always been very delicate. This can be seen from the changes in the "Qingtao APP". When it was first launched in April, the data between Douyin and Qingtao was only connected at the account information level - users using the Qingtao APP could directly log in with their Douyin accounts, and their previous follow relationships on Douyin were also brought to Qingtao.

However, the two sides are still isolated in terms of "core assets" such as playback volume and likes. Lujiu Business Review found in previous tests that the same videos posted on Douyin or Toutiao may have tens of thousands of likes, but on Qingtao, there are only single digits, no more than 1,000.

And a few days ago, this "Qingtao" app for young people was officially renamed "Tik Tok Selection", embarking on the path of the former Volcano Short Video.

Lujiu Business Review entered the APP and found that in addition to the name change, almost all the creator interaction data in Douyin Selection, such as likes and comments, have been connected with the Douyin itself.

Another thing worth noting is that the "long video" section, which is currently undergoing grayscale testing, is almost a "pixel-level replica" of Douyin's selection, in terms of function, interface and data. The common reference object of the two is Bilibili.

It is obviously unlikely that Douyin will do the same business twice. The only explanation is that the same long video business team is running as an independent APP and a functional module at the same time. Whoever has better data will become the main business direction in the next stage.

The introduction of the “dislike” function is most likely to serve this strategy.

2. Long videos: demand segmentation and user retention

Whether it is "TikTok Selection" or "Long Video", Lujiu Business Review found that after excluding some extreme cases, the length of the video content in the section is mostly between 60 seconds and 20 minutes. For the same reason, "Long Video" is not a long video that exceeds 15 minutes in the usual sense. It is "longer" relative to its own short video content. In fact, the benchmark is still the medium video users that Bilibili mainly covers.

However, this is not to a large extent the result of the product team "rushing for KPI", but the inevitable result of changes in the content ecology of Douyin - because on the current Douyin main site, the long-term coexistence of long and short video content has become an indisputable fact.

The reason is that the length of videos uploaded on TikTok has long exceeded the 15-minute limit. If you use "8 hours" as a keyword to search, you can easily find some audio novels, film and television commentary, and sports clips.

If you are a new user of Tik Tok, you will easily find that these contents will be recommended as one of the initial interests and appear in the initial information flow, even if they are not so "short videos" sometimes.

Not to mention, the production of these videos has formed a complete industrial chain so far. Not only can these creators themselves "recruit apprentices", but film and television editing tutorials on Taobao and Xianyu can be purchased for only a few dozen yuan at most. "Awei" and "Xiaomei" can be produced at a very low cost and in some time.

Whether Douyin is willing or not (in most cases, it is willing), this type of short-form long video has become an integral part of Douyin's content ecosystem. Douyin can let the audience for this part of content flow freely when the volume is small. However, as the overall number of users increases, the content and audience of long videos are also forming a trend on Douyin.

How to meet the needs of these users and improve their experience is what the "Long Video" section is doing. The role of the "dislike" function in user diversion is also evident.

3. Dislike: The “turnout” of user diversion and more accurate user portraits

In the current version of TikTok, the "dislike" function still exists, but most of the time, it is not easy for users to "dislike" a video content - the operation of first clicking the share button and then swiping right on the menu is much more cumbersome than tapping the screen for a "free double-click".

But after the gray area of ​​the "long video" section expanded, "dislike" finally came to the fore from behind the scenes.

The necessity of "dislike" is that users' content consumption is largely divided by video content tags rather than video length.

However, in terms of user interaction, merely using the “free double-click” method may be enough to determine the user’s profile based on the video type, but it is obviously not enough to determine the user’s preference for video length.

It is worth noting that as the video content track becomes increasingly crowded, both Bilibili and Douyin have already seen varying degrees of video content using the gimmicks of "ten thousand-word long articles" and "in-depth interpretations".

For the same reason, the proportion of "extreme cases" with extremely long duration mentioned above is increasing.

The "involution" of creators is a good thing for the platform. But it is not the case for TikTok.

A former ByteDance employee told Lujiu Business Review that these "electronic mustard tuber" videos will be recommended in turn for user screening after new users register. Once the "completion rate" or "interaction rate" of a certain type of video increases, this type of video will be recommended more often. Conversely, user labels and portraits are also improved through continuous interaction.

Some special categories, such as popular IP animation clips and male-oriented sports clips, are called "XX speed bumps" in the comment area. Their common feature is that the duration of a single video is extremely long, and compared with commercial content such as live streaming, the user stickiness is greater and the commercial value is relatively low.

However, the reason why Douyin can guarantee high commercial efficiency is largely related to the length of the video. If the video content is less than 5 minutes, users can browse more content in a unit of time, and content such as live broadcast rooms and product promotion videos can be displayed more frequently.

However, due to the large number of Douyin users, there is bound to be a demand for long videos. However, in order to ensure the commercial monetization efficiency of short videos on the main site, the long and short video sections will be cut and split, which is inevitable as time goes by and the content ecology becomes richer.

Moreover, there are also two sets of monetization logics for long videos and short videos: short videos can have live streaming and information flow advertising. After the long video sector is mature enough, it is not impossible for the patch ads that have been played out by "Youku, iQiyi and Tencent Video" to appear in the long video sector in the future.

4. Fighting against stagnation and seeking change

Some people say that ByteDance is the company that "least needs" to build large models. This is because the underlying technologies of large models are NLP and machine learning. And the most important application of machine learning in the past decade is algorithm recommendation.

ByteDance’s algorithm recommendation is undoubtedly the leader in China, and the most direct manifestation is that ByteDance has a huge amount of Internet advertising revenue in China. The main product that generates this revenue is TikTok.

In this sense, TikTok may be the earliest and most successful "big model" to be commercialized. However, this model is more driven by machine languages ​​such as codes, command lines, and algorithms rather than natural languages.

But after seven years of operation, Douyin has gradually expanded from a single entertainment short video to e-commerce and local life. In this process, long video content has also taken root in Douyin and formed a self-circulating ecosystem.

Douyin has not bet on medium and long videos before, such as Xigua Video, Qingtao and Douyin Selection. However, after being separated from Douyin's traffic injection, they all eventually became tepid.

The establishment of a separate "long video" section is similar to the "same city" section of local life to some extent, both of which are rooted in the original content ecology and then provide segmented services. The latter has been successful, so the former may not be impossible.

As the center of traffic distribution, the only thing Douyin needs to do is to change the direction of traffic by changing the rules and algorithms. This may be one of the deep reasons for "dislike".

The shift from "external entrepreneurship" to "internal change" can certainly be seen as a way to reduce costs and increase efficiency, but in fact, it has become an opportunity for Douyin to reform its own content ecology.

This kind of reform has happened countless times inside Douyin. After successfully building an e-commerce "store" and a local life "food city", this "comprehensive playground" based on algorithm recommendation has built its own long video "cinema", which is logically smoother than a "sales office" and more likely to succeed. The truth cannot be revealed, but can only be approached.

Author: Hu Jiaming

Source public account: Lujiu Business Review (ID: liujiucaijing69), the truth cannot be revealed, it can only be approached.