TikTok algorithm: how to change your feed recommendations

A person can spend one late-night hour watching apartment tours, skincare routines, grief videos, or oddly compelling footage of someone restoring a rusted garden tool, and suddenly their For You feed seems to have declared that this is their entire identity.

That sensation is not imaginary. TikTok’s recommendation system reads behavior closely, particularly the small behaviors people barely register as choices: whether they watch to the final second, skip within a beat, rewatch, comment, share, or pause long enough for a video to hold them. But learning how to change the TikTok algorithm is less about discovering a secret reset button than about becoming deliberate inside the platform’s parasocial architecture.

There is no official magic number of likes, searches, or videos that will instantly retrain TikTok. The platform does not publish one, and anyone promising a total algorithm reset is selling a cleaner story than the product can honestly provide. What TikTok does offer is a set of increasingly visible controls: feedback, feed refresh, topic preferences, keyword filters, and explanations for why a post arrived.

Together, they make the For You page less of a fate and more of a negotiation.

TikTok does not merely recommend content; it watches users perform their interests in real time.

Start with behavior, because behavior carries the most weight

For most users, TikTok says interaction signals are generally weighted more heavily than other recommendation signals. That means the platform notices more than a simple Like. It sees whether someone watches a video in full, shares it with a friend, leaves a comment, follows the creator, skips it, or spends a little extra time lingering over it.

This is where many attempts to customize a TikTok feed quietly fail. Users decide they are tired of a content category—dating discourse, celebrity speculation, rage-bait political clips, breakup confessionals—then continue watching because they want to see how absurd the video becomes. From a human perspective, that is curiosity, irritation, or cultural research. From a recommendation system’s perspective, it is attention.

The fastest way to retrain the TikTok algorithm is therefore not to perform dramatic digital housekeeping. It is to stop rewarding material you want less of.

A practical pattern looks like this:

1. Skip unwanted videos promptly. A quick swipe is a clearer negative signal than hate-watching through the ending, reading every comment, and sending the clip to a group chat with “this is unbearable.”

2. Use “Not interested” on repeat categories. Long-press a post and select the option when a type of video keeps returning. TikTok describes this as a way to request fewer similar posts in future recommendations.

3. Watch wanted content with intention. Finish videos from creators and subjects you genuinely want more often. A full watch, a rewatch, a save, a comment, or a share gives the system a richer picture than an accidental scroll.

4. Search toward the feed you want. Search is not a publicly defined command to the algorithm, but it is a meaningful expression of interest when paired with actual viewing and engagement.

5. Follow creators selectively. Following does not automatically dictate the For You page, but it helps establish a broader relationship with a creator ecosystem—and it gives TikTok more context for the cultural neighborhood you choose to inhabit.

The distinction matters. TikTok is built around a feed that can make a stranger feel familiar within days. That hyper-visibility is part of its appeal: a niche comedian, a “clean girl” lifestyle creator, a book reviewer, or a livestreaming gamer can become recurring emotional furniture in someone’s day without ever being formally invited in.

Changing the feed means changing those invitations.

Can you reset the TikTok For You page?

Not in the absolute sense people usually mean. TikTok’s official Refresh your feed tool does not erase all recommendation history, delete an account’s data, or guarantee a blank slate. It is better understood as a break in the current pattern.

The feature is designed specifically for the For You feed. After a refresh, TikTok may initially show popular material while it begins reshaping what it recommends. The user’s next actions—especially likes, comments, and viewing behavior—then help direct the new mix.

To find it, go to:

• Profile
• Menu
• Settings and privacy
• Look for the relevant content-preference controls and the option to refresh the For You feed, if it is available in your app version and region

The naming and placement of settings can change as TikTok updates the product, so it is sensible to use the in-app settings search if the control is not immediately visible.

A feed refresh is most useful when the problem is not one intrusive topic but a whole mood. Perhaps the For You page has become an endless corridor of the same viral audio, a succession of creators reenacting the same relationship script, or an atmosphere so heavily tuned toward anxiety that opening the app begins to feel like stepping into someone else’s nervous system.

Refresh can interrupt that loop. What happens next depends on the curation that follows.

The key is to avoid treating a refresh as a ritual cleansing. Reinstalling the app, clearing cache, deleting drafts, or making a new account are often repeated online as algorithm folklore, but they are not official methods for resetting recommendations. A feed is not stored only in the phone like clutter in a drawer; it is shaped through an ongoing relationship between account activity, content signals, and platform systems.

Use Manage Topics to adjust the atmosphere, not micromanage it

TikTok’s Manage topics setting is one of the platform’s more candid admissions that users do not simply want “more of what they watched.” They want a feed with a livable emotional climate.

The setting is reached through:

Profile → Menu → Settings and privacy → Content preferences → Manage topics

Users can adjust sliders to indicate how much of certain broad topics they want to see, then save the changes. TikTok said in June 2025 that the tool covered more than 10 popular topics.

This is not a precise genre-control board. It will not turn TikTok into a perfectly curated academic seminar, nor will lowering one category prevent every adjacent video from appearing. Trends do not stay inside clean labels. A beauty trend can become comedy, then consumer criticism, then a creator’s confessional about burnout; a sports clip can become a meme before the system has time to classify its social meaning.

Still, topic controls are useful because they intervene before the individual video level. They tell TikTok that the user’s interest preferences have shifted in broad terms.

There are two limitations worth holding onto:

• Topic adjustments apply to the For You feed, not the Following feed, profile pages, or inbox.
• TikTok warns that personalization can take time after a topic setting changes. The system is adjusting a pattern, not switching channels.

This slower rhythm can be frustrating, but it reflects the strange compact at the heart of recommendation culture. Users want platforms to recognize them deeply enough to be relevant, while also granting them the right to change their minds without being haunted by last month’s fixation.

That tension is not unique to social media. It is visible in every digital system that converts behavior into identity, from shopping feeds to streaming queues to the regulatory sorting underway in Europe’s crypto reset under MiCA. The difference is that TikTok’s classifications arrive as culture: jokes, faces, songs, micro-celebrities, and the sense that everyone else is suddenly talking about the same thing.

The For You page is not a diary of who someone is. It is a record of what held their attention—and those are not always the same thing.

Keyword filters are the cleanest tool for recurring irritants

If Manage Topics shapes the weather, keyword filters deal with the mosquitoes.

TikTok allows users to filter content containing specified words, word variations, and hashtags. Its current documentation says those filters can apply across the For You, Following, Friends, LIVE, and STEM feeds. Users may also have access to AI-assisted filtering that identifies similar terms, including synonyms and variations.

This matters because the most exhausting content on TikTok is often not a category as broad as “fashion” or “gaming.” It is a phrase. A repeated piece of influencer vocabulary. A reality-show contestant. A spoiler. A diet-culture euphemism that keeps escaping the language users have already muted. A discourse cycle that has become impossible to avoid because everyone has remixed the same argument through different aesthetics.

TikTok reported in June 2025 that keyword filters had been used to filter more than 200 million words from feeds globally. That figure says something quietly revealing: people are not only curating for pleasure. They are curating for cognitive space.

To make filters work better, think in clusters rather than one perfect word:

• Add the name of a recurring person or show, plus common hashtags connected to it.
• Include obvious alternate spellings and abbreviations when they are likely to appear.
• Filter the repeated phrase, not only the broad subject. Someone tired of one viral controversy may still want the wider community around it.
• Turn on AI-assisted filtering for similar terms if that option appears in the app, while remembering that automated matching is not the same as total exclusion.
• Review filters when a trend changes names or migrates into new slang.

TikTok had announced plans in 2025 to increase the number of filterable keywords to 200, but the platform’s public information does not establish that this is a confirmed current limit for every user. That small detail is worth mentioning because platform advice has a habit of becoming myth the moment a setting is announced.

Filters are not a moral judgment on the people being filtered. They are simply a boundary inside a medium built to make every public conversation feel inescapably adjacent.

“Why this post?” turns a mysterious feed into a readable one

The most underused TikTok tool may be the Why this post panel. It will not reveal the complete ranking formula—TikTok does not publish that—but it can identify possible reasons a particular video landed on the For You page.

Depending on the post, TikTok may indicate that the video appeared because the user:

• Interacted with similar posts before
• Followed the creator
• Has shown an apparent preference for longer videos
• Is seeing a post that is popular in their country
• Is being shown something recent

There is a cultural reason this feature matters beyond simple troubleshooting. Recommendation systems feel uncanny when they operate without explanation. A person watches three videos about moving to a new city and begins seeing relocation anxiety, interior design, career advice, and “things nobody tells you” monologues in one long stream. The feed feels like it has read their private life.

Often, though, it has read a chain of very legible signals. The explanation panel makes that chain visible enough to question.

If a video appears because it is popular in the user’s country, that is a reminder that TikTok is not only personalizing around the individual. Content information also matters: sounds, hashtags, view counts, and the country where the post was published can influence recommendations. User information such as language, location, time zone, day, and device type may also be part of the wider recommendation environment.

The algorithm is simultaneously intimate and infrastructural. It knows what someone lingers on, but it also knows what is circulating around them.

What turning off personalization does—and does not—mean

Some users may see an option to turn off personalized feeds, but it is not a universal TikTok setting. According to TikTok’s documentation, that option is available only in the European Union and Kazakhstan.

When personalized feeds are turned off there, the For You feed becomes a Popular feed featuring content popular both in the user’s region and internationally. This changes the experience from a highly individualized stream into something closer to a public square: still algorithmically arranged, still shaped by platform choices, but less tethered to the person’s own past engagement.

For viewers who feel trapped by the intimacy of recommendations, that distance can be clarifying. It replaces the constant mirror with a broader cultural window.

But it is not necessarily a better feed. A personalized page can surface emerging creators, tiny subcultures, and precisely timed practical information in a way a Popular feed may not. The question is not whether personalization is inherently good or bad. It is whether the current version of it still serves the person watching.

The real way to change TikTok’s algorithm is to make your attention less accidental

TikTok has trained users to treat attention as frictionless. One swipe becomes twenty; an ironic watch becomes an interest signal; a creator someone dislikes becomes a familiar face because dislike is still a form of return.

The platform’s newer controls acknowledge that users want more agency over this process. Refresh the feed when its logic has drifted. Adjust topics when the overall balance feels wrong. Filter words that repeatedly puncture the experience. Use “Not interested” without turning it into a private battle with every unwanted post. And consult “Why this post?” when the system’s choices start to feel too intimate to be accidental.

None of these tools creates a perfectly clean slate. That is not how TikTok works, and perhaps not how culture works either. Taste changes through exposure; people discover new creators through the very unpredictability they sometimes resent.

But a For You page should feel like a conversation with room to breathe, not a machine insisting that yesterday’s curiosity is tomorrow’s identity.