Every guide to vetting TikTok creators gives you the same workflow. Open an ad library, filter to ads running 25 or more days, check the Winning phase badge, read the creator handle off the ad, then negotiate from their proven spend.

It is a good workflow. It describes a screen that does not exist.

TikTok does not publish ad runtime, ad spend or ad rank for anybody's ads. Not to us, not to any competing tool, not to anyone. Meta publishes all three, which is why the workflow sounds familiar: it was written for Meta and then retold about TikTok by people who had not opened the filter panel. So the honest question is not how to filter TikTok by running days. It is what is actually on the screen, and whether that is enough to decide who gets your money. It is, and this page measures it: 15,905 TikTok posts from 477 brands, scored on the engagement the platform does publish.

The six niches

Ordered by the engagement rate a median post earns. Posts is how many were scored, Top 25% and Top 10% are the bar a creator has to clear to be unusual in that niche.

#	Niche	Posts	Median	Top 25%	Top 10%
1	Fashion	4,459	3.67%	6.01%	9.63%
2	Sports & Outdoors	2,279	3.22%	5.59%	9.79%
3	Beauty & Skincare	3,118	2.88%	5.59%	8.39%
4	Electronics & Tech	2,073	2.72%	4.90%	7.74%
5	Home & Garden	1,812	1.75%	3.06%	5.70%
6	Health & Supplements	2,164	1.67%	4.38%	7.55%

How to read the numbers

Engagement rate here is likes plus comments plus shares plus saves, divided by views. That is stated because every tool computes it differently and a percentage with no formula behind it is not a benchmark. Saves are included because on TikTok a save is the closest thing to purchase intent that the platform will show you.

This is a post-level number, not a creator-level one. A creator with one 9% post and nine 1% posts is not a 9% creator. Score the last ten posts, not the one they put in the pitch deck.

The niche gap is real and it is large. A 3% post is below average in Fashion and comfortably top quartile in Health & Supplements. Judging a supplements creator against a fashion benchmark is how a good creator gets rejected and a mediocre one gets hired.

These are brand-affiliated posts, not all of TikTok. The population is posts from Shopify DTC brands with at least $1M in estimated revenue, so the comparison is against other commercial content rather than against viral entertainment.

1. Check the engagement rate against the niche, not against a rule of thumb

The feed gives you the handle, the engagement rate, the age of the post, and the likes, comments, shares and views behind it. That is the whole vetting dataset, and it is on the card before you click anything.

Read the percentage on the card against the row for your niche in the table above. The two here are 5.6% on a 10-day-old post with 29.5k views, and 16.9% on a 41-day-old post with 125k views. The second clears the top-decile bar in all six niches. The first is comfortably top-quartile in Health & Supplements and only slightly above the median in Fashion, which is the entire reason to read the number against a category rather than against a rule of thumb.

Check a creator's numbers against their niche

2. Establish that a human made it

The second filter in the Creative group is the one nobody expects to need. A large share of what reads as creator content is now an AI avatar with a scripted voiceover, produced in bulk.

That matters less for performance than people claim, and enormously for research. If you commission a batch from a real person because an AI clip's hook tested well, you have copied a format from a creator who does not exist.

The AI Generated toggle is a single control and it removes them from the feed.

3. Know which filters you do not have

This is the panel the genre assumes. Everything TikTok will tell you about a post, in three groups.

Engagement is Views, Likes, Shares, Saves and Comments. Creative is Format, Spark/Reel and AI Generated. Targeting is Account Country and Language. That is the complete list.

There is no Running Days. No Phase. No Ranking. No spend of any kind. Any article telling you to filter TikTok creator ads to "25+ running days" or to look for the "Winning" badge is describing controls that are not there, and no tool can add them, because the data is not published.

4. Compare it with what a platform that does publish looks like

The same product, the same filter button, one platform across.

Meta carries Total spend, Daily spend and Reach under EU and UK, and Days active, Ranking and Followers under Ad performance. That is where the runtime-and-phase workflow comes from, and on Meta it works.

Two of those deserve care. The rank badge is worldwide. It is computed across everything the brand runs in every market. The euro figures are not. They are only the slice EU and UK disclosure rules force advertisers to publish. So a creative can rank first globally and still show a modest euro number, and reading that number as weak performance gets the conclusion backwards.

5. Use Spark as the one paid signal TikTok does give you

Spark/Reel sits in the Creative group above, and it is the closest TikTok comes to telling you a brand backed a creator. A Spark ad runs on top of an existing organic post from a real account, so the creator made it and the brand chose that specific post out of everything they produced.

Two things on this row matter more than the badge. The account reads "Unknown" on all three. So the step every version of this guide recommends, reading the creator's handle off the Spark ad, does not reliably work. You get the evidence that a brand paid, without the name of who they paid.

And the engagement rates are 0.7%, 0.4% and 0.3%, against 4.4M to 6.3M views and runs of 68 to 87 days. Every one of those is far below the median for any niche in the table. That is not weak content. It is what paid amplification does to the ratio: the denominator is bought, so views climb while engagement per view collapses. Never compare a boosted post's engagement rate to an organic benchmark. Score creators on their organic posts, and use Spark only as a yes-or-no flag that somebody with a budget already backed them.

For the filter combination that isolates real creator content from agency and AI-produced lookalikes, see the Spark Ad filter method.

Stop reading about winners. Find them yourself.

Search 6.5M+ brands, their ads, revenue, and products — all in one place.

Try Brandsearch free

6. Check the brand behind the creator, not just the creator

A creator whose partners are all shrinking is a creator whose next case study will be about you.

See this brand in Brandsearch

Sol de Janeiro sells Brazilian-inspired skin and body care. Its TikTok account carries 3.4M followers and 49.8M hearts across 1.3k videos, and the panel shows the direction as well as the level: down 988 followers over 7 days, up 360 over 30. A flat-to-slightly-up account at that size is a brand holding position, which is a very different negotiating context from one climbing fast.

The number to distrust is the follower count on its own. It is the one metric a creator can buy, and the one the whole genre still leads with.

Pull any brand's organic trajectory before you sign

7. When you want proof a creative earned its budget, cross to Meta

If the point of vetting is to find content that has survived contact with a real budget, that evidence exists, on the platform that publishes it.

Every dot is one of the brand's ads, placed by its worldwide rank against how long it ran. The cluster sits between 7 and 30 days with the best-ranked creatives pushing past 60. That is the shape the TikTok workflow is imitating, drawn from data that is actually published.

A creator who has made ads for a brand on both platforms is worth more than one you can only see on TikTok, because half their work comes with a runtime attached. For what those runtimes look like across six niches, and the day a winning creative usually gets replaced, see Facebook ad frequency benchmarks by niche.

If you want the 10-second version of the brand check before you even open the app, the free Chrome extension reads traffic and tech stack off a creator's link-in-bio store while you are looking at it.

What this replaces

Six of the claims this page used to make, across the nine near-duplicate versions of it that are now merged here, could not survive being checked.

The "filter to 25+ running days" step, the Phase badge, the impressions rank, the estimated spend per creator, the duplication count as a scaling tell and the CTA mix all belong to Meta. The spreadsheet column "Estimated Spend" was asking readers to record a number that does not exist for TikTok. None of it was dishonest. It was written from the Meta workflow and never checked against the TikTok panel.

What replaced it is narrower and true: an engagement benchmark per niche, a formula for it, an AI check, a Spark flag, and the brand's own trajectory.

How we measured this

15,905 TikTok posts from 477 brands, pulled 12 September 2026. The population is video posts from Shopify DTC brands with at least $1M in estimated revenue, sampled with a fixed seed and capped at six posts per brand so that no single prolific account sets its niche's median.

Posts under 500 views are excluded, 4,474 of them, because an engagement ratio computed on 40 views is noise rather than signal. That exclusion pushes the medians up and it is the largest judgement call in the pull, so it is stated rather than buried.

Engagement rate is computed here as likes plus comments plus shares plus saves over views. It is not read off a field, because the API does not return one: engagement_rate appears in the field list but comes back empty, and the filter built on it returns zero rows.

One capture problem is worth recording because it decides what a page like this is allowed to claim. The Discovery grid paints empty on most runs: the result count arrives, the cards do not, and a screenshot taken on a timer photographs a blank panel that looks like a view with no data in it. Every capture here was taken only after asserting that real cards had rendered, and two crops were discarded for failing that check. Nothing on this page is quoted from a screen that did not paint.

Query this yourself

Every number in the table came out of one endpoint. This is the actual call for one niche:

curl -s -H "X-API-Key: $BRANDSEARCH_API_KEY" \
  "https://api.brandsearch.co/v1/tiktok-ads/search?platform=shopify\
&brand_type=dtc%20brand\
&niche=Beauty%20%26%20Skincare\
&min_revenue_min=1000000\
&is_video=true\
&max_ads_per_brand=6&seed=20260912\
&page_size=100&page=1\
&fields=id,brand_id,play_count,digg_count,comment_count,share_count,collect_count,duration"

Check the parameter list before you trust a filter. running_days_min, phase and ad_type are all accepted by that endpoint and all silently ignored: the total comes back as 102,536,430 either way, which looks exactly like a filter that worked. The same data is available over MCP, so you can ask Claude, Cursor or ChatGPT for your own niche's benchmark directly.

FAQ

What is a good engagement rate for a TikTok creator?
Between 1.67% and 3.67% is the median depending on the niche, and the top 10% of commercial posts run 5.7% to 9.8%. Use the row for your own category, because the gap between Fashion and Health & Supplements is more than double.

Can I see how long a competitor's TikTok ad has been running?
No, and neither can any other tool. TikTok does not publish ad runtime, spend or rank for other advertisers. Meta publishes all three, which is why guides written for Meta read as though the same is true of TikTok.

Do follower counts matter at all?
As a tiebreaker. They are the one number that can be bought outright, and a 50k account posting problem-solution demos in your category will usually outperform a 500k account whose audience arrived for something else.

How many posts should I check before paying a creator?
Ten. One post tells you a creator can have a good day, and the vetting question is whether they clear their niche's median repeatedly.