Detailed explanation of Tik Tok recommendation mechanism!

No matter where you are, if you want to make money, you must have basic respect for local rules. Therefore, it is necessary to master the rules of a platform in order to create works suitable for the platform.

1. The characteristics of the Tik Tok platform.

1. Decentralization

The biggest difference between personal and other new media platforms is that on other platforms, you can see the content of whoever you follow. The more fans this person has, the more traffic he has. This will result in a situation where all users will be browsing the same few top-level, higher-quality authors every day. But TikTok is not. Douyin specifically distinguishes between following and recommending, so if you follow a creator but his works don’t have good traffic, it will be difficult for you to see him on your Douyin recommendation page.

However, the usage habits of most users are still to frequently browse the recommended pages. In other words, Douyin's fans are not all your fans, and the number of views of the videos you post still depends on the content of the videos you posted previously. It is this mechanism that provides many opportunities for new and small creators. When the system recommends your video to users who may be interested, the users decide whether to watch it based on the title and content of the video.

2. Algorithm mechanism

Only if you understand the decentralized algorithm rules of TikTok, can you understand how to use TikTok. The recommendation algorithms of the entire Toutiao system are almost similar.

Unlike Weibo and WeChat, Douyin basically provides a basic recommendation mechanism. Each video will first be allocated a certain amount of exposure, usually 100, 200, 500, or 1000. The platform will make further allocations based on the playback data of the basic traffic provided, thus forming a tiered traffic pool. Therefore, the interaction rate and completion rate of a video are the factors that determine whether your video can enter the next traffic pool and gain more exposure.

3. Overlay recommendations

If you want to get more replays after publishing a video, that is, to add traffic, the entire review rules are consistent with the basic recommendations.

Generally speaking, the quality of everyone's works is up to standard. If the randomly assigned 1,000 views or reposts reach a certain number, the platform will determine that the video is a high-quality and popular video, and will automatically increase your recommendation volume. Therefore, when we produce and publish videos, we must actively intervene in user comments and likes to guide users to interact and gain more exposure. For example, you can add some interactive language in the title to guide users to comment, such as "Is this the ** you like?", "What is your dream wedding like?", "What kind of ** do you want?".

But I need to remind everyone not to increase your traffic! There are many people on Taobao who cheat by posting likes and comments. Most of them are robots. Apart from the change in data, there is no other benefit.

2. Distribution mechanism of works on the Douyin platform

1. Deduplication mechanism

Literally speaking, it means to avoid duplicate works and plagiarized videos. This is an official mechanism to protect originality and maintain the health of the platform. As long as your work has been repeatedly published on the platform, or moved from other platforms, or is highly similar to other videos, it will be strangled in the cradle by this mechanism.

Of course, Douyin’s mechanism is not very strict and effective. There are currently many funny bloggers, film and television editing bloggers, etc. A lot of their content is copied from other places, and even the original part of secondary creation is pitifully small. Recently, many film and television industries have also been constantly resisting this behavior. Therefore, the deduplication mechanism of the Douyin platform will be gradually updated and upgraded. In order to develop healthily and long-term on Douyin, we still need to create our own original content. For example, Li Zi* created his own video style in Tik Tok short videos and is quite famous on the Internet. He also created all the shot scripts himself, and the number of clicks, views, and likes are very high. Some people began to imitate his video style, such as Xiangmiao**, who has had many fans since his creation, but still cannot surpass Li Zi* in terms of views, clicks and other related data.

2. Audit mechanism.

Generally, social media platforms will go through a review process when publishing works, which involves machine review and manual review. For example, WeChat Moments and QQ Space are mainly subject to simple machine review, and the restrictions are mainly aimed at erotic images. Douyin mainly relies on machine review, with some manual review to review the content of the entire video, including images, sounds, and video text. Review whether the content contains sensitive information, such as pornography, gambling, drugs, political sensitivity, blood and violence, attacks and provocations, off-site traffic, excessive advertising, etc.

In this step of the review process, artificial intelligence review, that is, machine review, is mainly used. These machines perform deep learning based on some databases and will automatically compare the content of your video. Once there is a high degree of match with a certain data in the database, your video will be intercepted. As a video creator, you need to have a positive and optimistic value orientation, and as an enterprise, you should know how to do appropriate marketing.

Let me give you a few examples. For example, if the video contains dirty words, revealing clothing, bloody scenes in the game, the video is shot while drinking, the video is shot while driving, the title contains sensitive words (political tendencies, pornographic words, etc.), whether there is bad guidance, embedded hard advertising, etc., these are all violations.

Another part of the review is done manually. The official has fixed staff who judge whether there are any related violations according to regulations. For example, pornographic film inspectors do this kind of work, which can effectively fill the loopholes in machine review. When your video gains a certain amount of traffic on Tik Tok, it will enter a secondary manual review. This process is mainly aimed at establishing hot spots. You can put it on the hot spot list or challenge list. Only after manual review can it be included in the list to gain more exposure.

3. Feature recognition mechanism.

In this aspect, it is mainly divided into user feature recognition and work feature recognition. Feature recognition of works means that in a video, while it is being reviewed by a machine, it will also use artificial intelligence for feature recognition, automatically extract your video content and text, then annotate the video, and then match the video to people with relevant interests.

User feature recognition is mainly done through artificial intelligence algorithms to study your preferences and which category you belong to, so that the video interaction rate will be higher.

The longer you stay on a certain category of videos, the system will recommend you more videos in this category and related videos. So, this is where the idea of ​​maintaining an account comes from. It means that you should watch more content related to the videos you like, and you can only watch videos of this type. Let the official know your interests and they will recommend your videos to people with the same interests as you. But in fact, this is one-sided, because this mechanism is divided into users and works.

Therefore, if you simply only watch a certain type of work, the platform will only label you as a user and will send you more works of the same type. But for a video creator, your label depends on the content of your video. These are two different things and cannot be confused.

For example, if you are running a Douyin account for maternal and infant products, you can watch videos related to the industry in the early stage, and follow and search for videos related to the industry. Naturally, the account will be labeled as maternal and infant. At this point, this is only the first step; the second step is to pay attention to the quality of your own works and output high-quality content to attract more audiences. For example, one of our maternal and infant accounts at the beginning, babyhe***, used the method of account cultivation. We did not invest any money, but attracted natural traffic. In 3 days, the account had more than 100 fans, which shows that account cultivation needs to go hand in hand with high-quality content.