What Manage Keywords Means on TikTok.

Manage keywords is a TikTok setting that sits on a video you have already published.

TikTok attaches keywords to your post by itself, chosen from phrases people type into TikTok search, and the panel shows you the list it picked. You can read every term, block the ones that do not describe the video, and suggest ones that do.

Suggestions go through TikTok’s own relevance check, so you cannot bolt an unrelated trending phrase onto a post that has nothing to do with it. The list can be changed at any time, not only at upload.

The panel says it plainly: “This post displays keywords automatically added by TikTok based on popular searches. You can manage them at any time.”

Most brands will never open it. That is the opportunity, and it is not a project. It is a post-publish check that takes under a minute and quietly decides which searches your video is eligible for. The wider case for treating social as a search surface is settled. This is the specific TikTok mechanic sitting underneath it.

Where to find Manage keywords, and who has it

Open one of your own published videos, tap the three-dot menu, and look for the keyword option. Everything reported so far happens in the phone app, so if your process runs through desktop, plan for someone to do this on a phone.

Availability is uneven, and that matters more than the click path.

The panel surfaced in the app in late April 2026 and spread through May. TikTok has still not published a newsroom post or a help centre page for it, so there is no official statement of who has it, in which markets, or on which app versions. Nothing published points to a Business account requirement either.

If you cannot see it yet, update the app and look again in a fortnight. Do not rebuild a process around a screen you cannot open.

What TikTok has been doing to your posts all along

Keywords are not hashtags and they are not your caption. TikTok generates them by matching your video against phrases people search for, using what it can read in the post and how viewers behave around it.

That matching has always run. What changed is visibility and consent. The list is now yours to read, and yours to argue with.

Filing is the right mental model. Every post gets filed under a set of search terms, and the filing decides who can find it later. A video filed under the wrong terms is not punished, it is simply invisible to the people looking for exactly what it contains.

What you can actually change

• Block a keyword TikTok assigned, which stops it being used to match your post to searches.
• Suggest a keyword you think fits, which TikTok then judges for relevance before it counts.
• Do both on posts that are already live, at any time, not only at upload.

Two limits matter before you start.

Blocking is blunt. A block appears to take close relatives with it, so blocking “dance” also takes “dancing”. Treat each block as a small family of words rather than one word.

Suggesting is not granting. TikTok keeps the final call on whether a term is relevant, and there is no published cap on how many you can propose.

Blocking is the job most brands need

Adding a keyword is the enjoyable half. Removing one is usually worth more.

A wrong keyword does two kinds of damage. It enters your video into a search it was never going to win, and it delivers your video to people who wanted something else, who then swipe away in a second.

The second one is expensive, because the platform reads that swipe as a verdict on the video, not on the filing.

Picture a product demo shot in a kitchen that gets filed under recipes, or a founder interview filed under the sport playing on a screen behind them. Those are the posts to check first: anywhere the setting, the audio or the background disagrees with the subject.

Adding is real, but smaller than it sounds

Because TikTok judges relevance, the suggestion box is not a place to win an argument. If a term is not evidently what the video is about, it will not stick, and it should not.

Which makes accuracy the whole game, not volume. The question is never “what do we want to rank for”. It is “what is this video actually about, in the words a person would type”.

A caption is written for the person already watching. The keyword list is read on behalf of the person who has not found you yet, which is why plain phrasing beats clever phrasing every time.

Three or four terms that genuinely describe the post beat a dozen aspirational ones. Keyword stuffing does not work here, and the review step is built so that it cannot.

Decide the terms once, not per video

The failure mode shows up the first time a team tries this. Someone stares at a post, invents keywords on the spot, does it differently every time, and the check takes ten minutes.

Decide at the pillar level instead. For each content pillar you publish into, write the ten to fifteen search phrases a real person would type to find that kind of video, in their words rather than your product’s words.

That list becomes the reference. At publish time nobody is inventing anything, they are matching: which of our agreed phrases does this specific video genuinely answer.

Refresh it quarterly, or whenever search analytics surfaces a phrase you had not considered. Getting the platform mechanics right is mostly this kind of work, deciding once so the daily job stays fast.

Where the check belongs in your process

Right after publish, in the same sitting, done by the person who published.

Not a weekly keyword session. Not a separate spreadsheet. Not the strategist’s job. The moment the post is live, whoever pressed publish opens the panel, reads the list, blocks what is wrong, adds what is missing, closes it.

The reason is ownership. A step that belongs to a specialist gets scheduled, and a step that gets scheduled gets dropped.

The one-minute pass, per post

1. Open the published post and find the keyword panel.
2. Read every assigned term against one question: would someone searching this be glad they landed here?
3. Block anything that fails. Do not deliberate, because a term you are unsure about is a term that is not clearly about your video.
4. Add one to three phrases from your pillar list that the video honestly answers.
5. Close it and move on.

If that takes longer than a minute or two, the problem is upstream. Either the pillar list does not exist, or the video is about three things at once, which is a content problem the keyword panel cannot fix.

What to watch afterwards, and what it will not tell you

Post analytics shows where views came from, and search is one of those sources. Where TikTok has enough activity to show them, you also get the phrases people searched before they arrived.

Two things are worth tracking, both at account or pillar level rather than per post:

• The share of views arriving from search rather than the feed.
• Whether the phrases showing up are the ones you chose, or ones you would have blocked.

Be honest about the limits. TikTok does not show the same analytics for every post, account, region or app version, query lists are often sparse, and there is no before and after view on a single video.

So judge this over a pillar and a month, not a post and a day. If search’s share of views climbs across the posts you have been checking while it stays flat on the ones you have not, that is as close to evidence as the platform will hand you.

What good looks like

Boring, fast and never skipped.

• Every content pillar has an agreed list of search phrases, written down where the publisher can see it.
• The keyword check is a line on the publishing checklist, not a separate task with its own meeting.
• The person who publishes does it, in the same sitting, in under two minutes.
• Blocks outnumber additions early on, because nobody has been correcting the filing until now.
• Search share of views gets reviewed monthly at pillar level, not argued about per video.

How NBK thinks about a change like this

Platforms ship small controls like this constantly, and most of them die in the gap between “someone should look at that” and the next strategy review.

The controls that get used are the ones bolted to something that already happens daily. Publishing already happens daily. So the keyword check goes on the publishing checklist, gets an owner by default rather than by assignment, and is measured where the numbers are stable enough to mean something.

That is the method in one line: turn a feature into a step in the workflow, give the step a home and an owner, then leave it alone.

Next step

If your team publishes consistently and still feels invisible in search, the constraint is rarely the content. It is usually that nobody owns the small steps between publishing and being found. NBK can help find that constraint in the system.