TikTok Algorithm: How the For You Feed Recommends Videos

TikTok has described its TikTok algorithm publicly in How TikTok recommends videos #ForYou and in its help center. This guide sticks to what the platform states, separates it from creator folklore, and turns it into actions.

How does the TikTok algorithm work?

The For You feed is personal: two people never see the same feed. For each viewer, the system takes a large pool of eligible videos, estimates how likely the viewer is to engage with each one, and ranks them. Every action of the viewer then updates the estimate, which is why the feed adapts within minutes of use.

When a new user joins, TikTok has little data. It starts from the interest categories chosen at sign-up, if any, and from a mix of popular videos, then learns from the first reactions. The same learning loop applies to every user, all the time.

Recommendations also depend on eligibility: a video that was just uploaded, is under review, or is judged unsuitable for a general audience may not appear in For You at all.

Which signals does TikTok use to recommend videos?

TikTok gives one example of a strong indicator: whether a viewer finishes watching a longer video from beginning to end. A completed video counts far more than a weak signal such as the viewer and the creator being in the same country.

Signals creators can influence

• Completion and rewatches: a hook in the first second and no dead time keep people watching.
• Shares: a viewer who sends a video to a friend gives a very clear signal of value.
• Comments and saves: questions, opinions and useful content invite them.
• Text of the video: caption, on-screen text, hashtags and the sound tell the system what the video is about and to whom it may matter.

What does not influence the TikTok algorithm?

TikTok states that neither the follower count nor whether an account has had previous high-performing videos is a direct factor in the recommendation system. Followers help indirectly, because people who follow you tend to watch and interact, but each video is judged on how viewers respond to it.

Several other beliefs do not come from TikTok's own explanations:

• Deleting and reposting a video is not presented by TikTok as a way to reset its reach.
• Long lists of hashtags do not replace a clear topic; a few precise ones describe the video better.
• A "shadowban" for posting too often is not part of TikTok's documentation. Videos can, however, be excluded from For You when they break the Community Guidelines, are spam, or are under review.

How does TikTok keep the For You feed varied?

The system deliberately avoids repetition. TikTok explains that it generally does not show two videos in a row with the same sound or from the same creator, skips duplicates and videos already seen, and inserts content outside the viewer's usual interests to help them discover new topics. For a creator, this means reach is never locked to one niche audience: a good video can reach people who have never watched the topic before.

Viewers also shape the feed directly. They can long press a video and choose "Not interested", hide videos from a creator or with a given sound, and refresh their For You feed to start again from new recommendations.

How does TikTok search work?

Many people use the TikTok search bar to look for recipes, product reviews, places and tutorials. TikTok search matches the query with the text of the video and its context, so keywords matter in places that are easy to forget:

1. The first line of the caption, written as a natural sentence that contains the main keyword.
2. On-screen text in the first seconds.
3. The spoken words, which TikTok can transcribe through automatic captions.
4. A few hashtags that name the topic.
5. The profile name and bio for branded searches.

TikTok search also shows suggested searches under videos and in comments, which signals the queries the platform associates with a topic. Our guide to TikTok for SEO and GEO goes further on keyword choice.

How do you work with the TikTok algorithm?

Read your analytics per video: average watch time, percentage of viewers who watched the full video, and traffic source (For You, search, profile, following). A video with high completion and low reach often needs a clearer hook or topic; a video with high reach and low completion promises something it does not deliver. TikTok updates its systems regularly, like Google does with its algorithm updates, so judge trends over several weeks rather than one post.

Why does the TikTok algorithm matter for SEO and the AI?

TikTok videos appear in Google results for many how-to and discovery queries, and short videos follow the same rules as any video SEO: clear topic, descriptive text, transcription. A strong TikTok presence also creates branded searches on Google, and the mentions of a brand across platforms feed the picture that ChatGPT, Gemini and Perplexity build of it. The same short videos can be repurposed on Instagram Reels; see our guide to Instagram marketing.

The TikTok algorithm rewards videos that people watch to the end and share, whatever the size of the account. To turn the attention you win on TikTok into search traffic that lasts, your website needs articles that rank on Google and get cited by the AI: Sorank writes and publishes them on autopilot, with a 3-day free trial.