The TikTok Algorithm

How the For You feed works, why TikTok can feel addictive, and what users can do to stay more conscious.

Hi everyone, today I want to talk about the TikTok algorithm — how the For You page works, why it can feel addictive, and why that matters for teenagers.

TikTok is not just a video app. It is a prediction system that learns from our behaviour. Every swipe, like, comment, and skip becomes a signal. The For You page uses these signals to decide which video to show you next.

TikTok uses three groups of signals. The strongest is user interactions — watch time, likes, comments, shares, and skips. Second is video information, such as hashtags and sounds. Third is device and account settings like language and country, which have the least weight.

Graph 1

What trains the For You page?

This is a qualitative graph. TikTok does not publish exact weights, but it does say active user signals are more important than basic device settings.

User interactions strong

Video information important

Device settings lower

Graphic credit: original HTML/CSS graph for this website. Based on TikTok Newsroom and TikTok Safety Center explanations of For You recommendation signals.

Active signals matter most. Finishing a video sends a stronger message than just being in the same country as the creator. This matters because even watching a video you did not really like is already data for the algorithm.

The algorithm works as a feedback loop. You watch or skip a video — TikTok measures the signals — the system predicts what to show next — the feed updates — and the loop starts again.

Feedback-loop graph

The attention feedback loop

The basic idea is simple: TikTok learns while the user watches. The feed then changes, which creates more data.

1. Watch or skipVideos create tiny behavior signals.

2. Measure signalsLikes, watch time, searches, follows, comments, and skips matter.

3. Predict interestThe system ranks videos for the next feed.

4. New feedThe next videos become more personal.

5. Watch againThe loop starts again with new data.

Graphic credit: original loop graph for this website. Based on TikTok's explanation that the For You feed continuously learns from user engagement, plus research on short-video addiction mechanisms.

Because videos are short, this learning happens very quickly. A few minutes of scrolling gives the algorithm a lot of new data.

According to Pew Research Center data from 2025, 61 percent of teenagers use TikTok daily, making recommendation systems a normal part of everyday life for most teens.

Graph 2

Daily platform use among teens

This graph gives context for why recommendation systems matter in everyday life.

YouTube 76%

TikTok 61%

Instagram 55%

Snapchat 46%

Facebook 20%

Graphic credit: original HTML/CSS bar chart for this website. Data source: Pew Research Center, "Teens, Social Media and AI Chatbots 2025", survey of teens ages 13 to 17, Sept. 25-Oct. 9, 2025.

Why TikTok can feel addictive

There are three main reasons. First, infinite scroll: the feed never ends, so stopping always requires a conscious decision. Second, autoplay: the next video starts without you choosing to play it, making it easy to keep watching without thinking. Third, variable rewards combined with personalisation: you never know if the next video will be boring or exactly what you wanted, and after a while the feed knows your hobbies, your humour, and even your insecurities. That combination keeps people scrolling longer than planned.

The European Commission mentioned these same concerns — infinite scroll, autoplay, notifications, and highly personalised recommendations — in a preliminary finding under the Digital Services Act in February 2026, showing that governments are taking algorithmic design seriously when young users are involved.

Risk map

Features linked to addictive design concerns

Infinite scrollNo natural end point, so stopping needs active effort.

AutoplayThe next video starts without a real decision.

NotificationsThey can pull users back into the app.

PersonalisationThe feed becomes more relevant and harder to leave.

Graphic credit: original risk map for this website. Source: European Commission press release from 6 February 2026 on TikTok's addictive design under the Digital Services Act.

Balanced view and conclusion

To stay fair: TikTok is not automatically bad. Many people use it for creativity, learning, comedy, and community, and small creators can reach big audiences without many followers.

The real issue is that the app is designed to keep attention for as long as possible. That can conflict with what you actually wanted to do — homework, sleep, or just put the phone down. There are also real risks: spending more time than planned, less sleep, and the filter bubble, where the algorithm keeps showing one type of content until that topic feels bigger and more normal than it really is.

TikTok does offer tools like "Not interested", keyword filters, and screen-time limits — but they only work if you actually use them.

To conclude: TikTok's algorithm is a feedback loop between human attention and machine prediction. It can be useful and entertaining, but it is also designed to be hard to stop. So the key question is not only "do I like this video?" — but "did I choose to watch this, or did the algorithm choose it for me?"