Updated October 2026: This article has been rewritten and expanded with current information.
Open TikTok for the first time and within a few minutes the app seems to know what you like. A few swipes past cooking clips, a long pause on a woodworking video, and the For You feed starts filling with sawdust. Many users describe the experience as uncanny, as if the app were reading their minds.
It is not reading minds. It is reading behavior, very closely, and at enormous scale. This explainer covers what TikTok has said publicly about how its recommendation system works, what a leaked internal document revealed in 2021, what regulators have forced into the open since, and the controls you can use today to shape your own feed.
The "TikTok Algo 101" document
In December 2021, The New York Times reported on an internal TikTok document titled "TikTok Algo 101," produced by the company's engineering team in Beijing to explain the recommendation system to non-technical employees. A TikTok spokeswoman confirmed the document was authentic.
According to the Times, the document framed the system around four goals, which the company translated as user value, long-term user value, creator value and platform value. In plain terms, the app tries to show you videos you will enjoy now, keep you coming back over time, give creators a reason to keep posting, and serve TikTok's business. The report also described how the system weighs the different ways people react to videos and how it works to keep users from getting bored, which helps explain why the app is so hard to put down.
The document drew attention for another reason: it showed close ties between TikTok and its Chinese parent company, ByteDance, at a time when the U.S. government was weighing whether the app posed a national security risk. That question would shape TikTok's future in the United States for years.
What TikTok says it measures
TikTok published its own explanation of the For You feed in June 2020. It said the system ranks videos using a combination of signals, each given different weight:
| Signal type | Examples | Weight, per TikTok |
|---|---|---|
| User interactions | Likes, shares, comments, accounts followed, videos you create, how much of a video you watch | Strongest signals |
| Video information | Captions, sounds and hashtags | Moderate |
| Device and account settings | Language, country, device type | Lower, because users do not actively choose them as preferences |
| Not used directly | Follower count, an account's past viral hits | Not a direct factor, though large accounts still get more views |
Source: TikTok Newsroom, "How TikTok recommends videos #ForYou," June 18, 2020.
TikTok gave an example of how weighting works: watching a longer video all the way through is a strong signal of interest, much stronger than the fact that you and the creator happen to be in the same country. This is the key to the "mind reading" effect. Most social networks learn mainly from things you deliberately do, such as following a friend or liking a post. TikTok also learns from things you barely notice yourself doing, such as how long you pause on a video, whether you rewatch it and how quickly you swipe away.
Because the feed shows one full-screen video at a time, every second of attention becomes a data point. Someone who watches 100 short videos in a session gives the system 100 small judgments to learn from, far more feedback than a typical scroll through a feed of photos and text.
How TikTok tries to avoid repetition
A system that only showed you more of what you already liked would get stale quickly, and could pull people into narrow "rabbit holes." TikTok says it deliberately mixes things up. Its 2020 explanation said the feed avoids showing two videos in a row with the same sound or from the same creator, filters out duplicates and videos you have already seen, and occasionally inserts videos outside your usual interests to test whether you might like something new. It also said it limits content that may be shocking to a general audience and keeps newly uploaded videos still under review out of recommendations.
Those safeguards have limits. Reporting since 2021 has found that accounts engaging with videos about self-harm, suicide or eating disorders were shown more of the same. TikTok has said it works to break up these clusters, and searches for some sensitive topics in the United States now display mental health resources. In 2023, Forbes also reported that TikTok staff could use a "heating" tool to manually boost selected videos, with such boosted videos accounting for roughly 1 to 2 percent of daily views.
Regulators and transparency
Since the 2021 leak, pressure from governments has pushed TikTok to explain more and give users more control. A short timeline:
- June 18, 2020
TikTok publishes its first public explanation of the For You feed. - December 2021
The New York Times reports on the internal "TikTok Algo 101" document. - December 20, 2022
TikTok adds "Why this video," showing reasons a video was recommended. - March 16, 2023
TikTok adds a way to refresh the For You feed and start recommendations over. - April 25, 2023
The EU designates TikTok a very large online platform under the Digital Services Act. - August 2023
TikTok lets users in Europe turn off personalized For You, LIVE and search results. - April 24, 2024
A U.S. law requires ByteDance to divest TikTok's U.S. operations or face a ban. - January 22, 2026
The sale of TikTok's U.S. operations to a new venture, TikTok USDS, closes.
The European Union's Digital Services Act was the biggest driver. Ahead of its August 2023 compliance deadline, TikTok said European users could switch off personalization. With it off, the For You and LIVE feeds show popular videos from the user's region and around the world instead of videos matched to their interests, and search returns popular content in their language. The Following and Friends feeds stay chronological.
In the United States, the dispute over ownership ran from a 2024 divestiture law through a brief shutdown of the app in January 2025, a series of enforcement delays, and finally a deal that closed on January 22, 2026. Under that deal, a group of investors including Oracle, Silver Lake and MGX owns most of the new U.S. venture, and ByteDance keeps 19.9 percent. Reports at the time said the app would look the same to users, with the recommendation system adjusted over time for U.S. users.
How to take control of your For You feed
You cannot switch off the algorithm in most countries, but you can steer it. These built-in tools are free and available in the app's settings or on each video:
- Check why you saw a video. Open the share panel on a video and tap the question mark to see the "Why this video" explanation. TikTok groups the reasons into your interactions, accounts you follow or that are suggested to you, recent content in your region, and popular content in your region.
- Tap "Not interested." Use the Not interested option on a video to skip future videos videos from that creator or with that sound.
- Filter keywords. In your settings you can add words or hashtags so videos using them are hidden from your For You feed.
- Refresh your feed. If your recommendations no longer fit, the refresh option resets the For You feed as if you had just signed up. It does not change who you follow.
- Use the Following feed. It shows only creators you follow, which gives you a less algorithm-driven view.
- Set a screen time limit. TikTok's built-in screen time tools let you set a daily limit. Since 2023, accounts of users under 18 have had a default daily limit.
- In Europe, turn off personalization if you prefer a feed based on popularity rather than your behavior.
The bottom line
TikTok's feed feels like mind reading because it learns from small, constant signals, especially how long you watch, and because it can test hundreds of videos on you in a single sitting. The company has disclosed the broad outlines of how this works, and regulators have forced more transparency and user control. Knowing what the system pays attention to is the best way to make it work for you rather than the other way around. If you want a different feed, change what you watch, not just what you like.
Sources
- TikTok Newsroom: How TikTok recommends videos #ForYou
- TikTok Newsroom: Learn why a video is recommended For You
- TikTok Newsroom: Introducing a way to refresh your For You feed
- TikTok Newsroom: An update on fulfilling our commitments under the Digital Services Act
- Wikipedia: Digital Services Act
- Wikipedia: Restrictions on TikTok in the United States
- Wikipedia: TikTok
Top image: "Hand holding smartphone with blank white screen" by Santeri Viinamäki, CC BY-SA 4.0, via Wikimedia Commons (resized). Charts by Thank You Robot.
0 Comments