Do you want to know more about the "TikTok recommendation algorithm" that is said to be magical?

600,000 followers gained in 5 days. Understanding the TikTok algorithm , followers gaining is that fast!

1 Tik Tok video got 3 million likes and gained 900,000 followers. My personal experience will help you master the Tik Tok algorithm!

It took only 10 days to get 1 million followers. No exaggeration, the TikTok algorithm is just so magical!

…

Uh...the above is purely fictitious and any similarity is purely coincidental.

As one of the core competitive advantages of TikTok, the recommendation algorithm has been talked about by many people since its launch. Today, let’s talk about this “recommendation algorithm” that has been passed down as a miracle by many Tik Tok players.

According to Douyin staff, the recommendation algorithm has always been kept in a black box, and there are no more than five people within Douyin who have truly understood the rules of this algorithm.

It can be said that as of now, any external judgment about the TikTok algorithm is speculation.

The difference is how close these speculations are to the truth.

1. Double Audit

On Douyin, a huge number of new works are uploaded every day. It is easy for loopholes to be exploited if it relies solely on machine review, and it is not realistic to rely solely on manual review.

Therefore, double review becomes the first hurdle for the Tik Tok algorithm to screen video content.

Machine review : Generally, it uses a pre-set artificial intelligence model to identify your video images and keywords. It has two key functions:

First, check whether there are any violations in the works or texts . If there are any suspected violations, they will be intercepted by the machine and human attention will be drawn by yellow or red flags.

Secondly, by extracting pictures and key frames from the video, they are matched with the massive works already in the Douyin database to eliminate duplication . Works with duplicate content are recommended with low traffic or downgraded (only visible to fans or only visible to yourself).

Manual review : mainly focuses on three parts: video title, cover screenshot and video key frames.

Douyin reviewers conduct detailed reviews of works that are suspected of violating regulations and works that are prone to violations and are screened out by machine review.

If a violation is confirmed, penalties such as deleting videos, demotion notices, and account bans will be imposed based on the offending account.

2. Overlay Recommendation

For works that pass double review, the system will allocate you an initial traffic pool: 200-300 online users.

Whether it is a new account that has just joined Douyin, or a celebrity or big V with millions of fans, the system will treat everyone equally.

It’s just that the subsequent superimposed recommendations will be very different.

Many Douyin novices often have doubts here:

Why does the system give more recommendations to others when their number of fans is almost the same?

Why do accounts with more followers get more recommendations?

Douyin's superimposed recommendations are most likely affected by the following aspects:

1. Initial account weight;

2. Feedback from initial traffic pool users (like rate, comment rate, forwarding rate, completion rate and attention ratio), as well as continuous feedback;

3. The account has received feedback from fans;

4. Activation of external real accounts (such as mutual likes, comments, and transfers).

New Account VS New Account

The more complete the account information, the healthier the account maintenance behavior, and the more vertical and high-quality the content, the higher the chance of receiving system recommendations, or becoming a "hot product".

Small Account VS Large Account

1. High weight: Large accounts are given higher weight by the system through continuous creation;

2. More accurate fans in the recommendation pool: Large accounts continue to launch vertical works, allowing the system to match more accurate users with the same attribute tags in the recommendation pool;

3. Fans raise the work index: Active users of large accounts can raise the repost, comment, like and completion rates of their published works to a very high rate.

This is why accounts with larger follower bases can receive hundreds of thousands or even millions of recommendations for each video.

3. Delayed “detonation”

Many TikTok operators will find that some content has mediocre data on the day, week or even month it is released, but suddenly becomes popular one day. Why?

Two reasons:

The first type is jokingly called "digging graves" by many veteran drivers .

It means that Tik Tok will re-explore the “high-quality old content” in the database and give it more exposure .

The reason why these old works can be "exploded" is firstly because their content is good enough, and secondly, your account has published a lot of sufficiently vertical content, the labels have become clearer, and the system can match you with more precise users.

With high-quality content and targeted users, it is no surprise that old works become popular again.

The second type, we can call it the "explosive effect"

It means that when one of your works gets a lot of exposure (millions or even tens of millions), it will bring a huge number of users to your personal homepage to browse your previous works.

If one of your works can get enough attention (reposts, comments and likes), the system will put these videos back into the recommendation pool.

Many vertical content creators often "ignite" several other high-quality videos because of the "popularity" of a certain video, resulting in a multi-point blossoming and overall explosive traffic generation.

4. Traffic peaks

After double review, initial recommendation, and superimposed recommendation, Douyin works usually bring a lot of exposure, interaction, and fans to the account.

The duration of this high recommendation exposure usually does not exceed one week.

Afterwards, the hit video and even the entire account will quickly cool down, and even some subsequent works released will hardly have a high number of recommendations. Why?

Douyin’s daily active users are limited, which means that the total number of recommendations is basically fixed:

On the one hand, the recommendation is basically completed for people with tags related to your content, and the feedback from other non-precise tag groups is poor, so the recommendation is stopped;

On the other hand, Tik Tok does not want a certain account to become popular quickly, but rather to test your ability to innovate content and continuously output high-quality content through rounds of tests.

Finally, we’d like to release this [TikTok Recommendation Flowchart], which is a summary based on the recommendation rules that Toutiao has announced, combined with practical feedback from many accounts.

In practical applications, it is only used as a reference.

Finally, let me quote a sentence from a friend in the circle:

"In fact, there is such a reality. Many Douyin players feel that their traffic is unstable and attribute the reason to Douyin's strategy and machine algorithms. In fact, the main reason is the instability of their own content quality. The machine recommendation algorithm is the most scientific. The quality of content is determined by user recommendations, not one's own vision. If the content is good, traffic and other things will come."

Source: Operation Hacker