TikTok search suggestions: where they come from

TikTok search suggestions are the phrases TikTok puts in front of you before you have finished typing: the autocomplete list that drops down under the search bar, the "you may like" row of searches, and the little search chip that appears under a video. TikTok builds them from what people are searching on the app, what you have searched and watched, and what is trending right now.

If yours have turned strange, that is the personalisation, not a glitch, and you can clear the history behind it and switch the personalised version off. Skip to how to turn them off if that is why you are here.

If you make videos, the same list is the cheapest keyword research on the platform. It is a live read on how people phrase things, straight from the search box. The second half of this page shows real suggestion lists, what they are good for, and the one thing they cannot tell you.

The three places suggestions show up

• Autocomplete in the search bar. Tap search, start typing, and TikTok completes the phrase. This is the list worth mining, because every entry is a phrase real people type.
• Suggested searches before you type. Open search and TikTok shows a set of ready-made queries, sometimes labelled "you may like". These lean on trending searches and on your own activity.
• The search chip under a video. While you are watching, a keyword box can appear near the caption. Tapping it runs that search. TikTok generates it from what it thinks the video is about, which makes it a useful signal for creators: if the chip under your video says something sensible, TikTok has understood the video.

Where TikTok search suggestions come from

Three inputs, in rough order of weight: what large numbers of people are searching on TikTok, what you have searched for and watched, and what is trending in your region. TikTok confirms the personal part in its help centre, where it notes that turning off the personalised search setting will show you non-personalised results and recommendations, and will also change suggested searches.

That explains the pattern behind most "why is TikTok suggesting this" posts. A phrase goes around, enough people tap it to make it a genuinely popular search, and it then appears for everyone else as a popular search. Tapping one teaches the system that you liked it. The suggestions are a mirror of search behaviour, including your own, and they are not a read on anything you said out loud or typed in a message.

How to turn off or clear TikTok search suggestions

There are two separate things to do, and doing only one is why people report that suggestions came back.

1. Clear the search history. Profile, then the menu at the top right, then Activity centre → Search history. Tap Select, choose everything, delete. You can also open the search screen, expand your recent searches and tap Clear all.
2. Turn off personalised search. Settings and privacy, then Content preferences. TikTok moves this control between releases, so use the search box at the top of the settings screen if it is not where you expect. With it off, suggestions stop being tuned to you, though popular and trending searches still appear.

Neither switch removes suggestions completely. TikTok has no setting that empties the search bar, so if the goal is a blank screen, the honest answer is that it is not on offer.

Mining suggestions for keywords

Type a seed and read the completions. Here is what TikTok returned for six common seeds on 19 September 2026.

Data: TikTok search autocomplete via TokConnect, pulled 19 September 2026. Open the snapshot for all 20 seeds.

Read those columns and the video briefs write themselves. "skincare routine for men" and "skincare routine step by step" are two different videos with two different hooks. "budget" splits cleanly into money and phones, which tells you the word alone is useless as a target.

We pulled three more seeds for this page, in niches where the pattern is sharper.

Data: TikTok search autocomplete via TokConnect, pulled 19 September 2026. Open the snapshot.

"sourdough fail" is the interesting one. Nobody plans a video called that, and it is what people type at 11pm holding a flat loaf. Autocomplete surfaces the unglamorous phrasing that keyword brainstorms miss. Adobe's February 2026 survey of 554 US creators found 28% of their most viral posts came from TikTok's suggested searches, which is a fair estimate of how much this is worth.

What autocomplete cannot tell you

Order is not volume. TikTok ranks completions by a mix of popularity and relevance to you, and publishes no number next to them. So autocomplete tells you which phrasings exist and roughly which are common, and stops there.

Creator Search Insights is where the counts live. We put the same sourdough seed through it on the same day and got topics with 7-day search volumes and matched video counts.

Data: TikTok Creator Search Insights via TokConnect, pulled 19 September 2026. Open the snapshot. What the numbers mean: TikTok search volume.

Notice that neither list contains the other. Autocomplete gave you "sourdough fail" and "sourdough for beginners"; the topic list gave you "Flat sourdough" at 219K searches a week with no videos matched to it. Use both: autocomplete for wording, topics for size. The tool comparison covers who has which.

Pulling suggestions at scale

Doing this by hand in the app is fine for one seed and miserable for thirty. If you have an AI agent connected to TikTok data, hand it the seed list and let it do the typing.

That returns in about a minute what a phone session takes an hour to collect, and it leaves you a file instead of a screenshot.