Thumbstop
Whether the first seconds of your video stop the scroll.
What it means
Thumbstop is 3-second video plays divided by impressions, shown as a percentage. Show a video 100,000 times, get 12,000 plays that reach 3 seconds, and it's 12%. The denominator is impressions rather than viewers, and it grades the opening frames only — everything the ad says after second 3 is invisible to this number.
Show the math
Formula and a worked example
Thumbstop = 3-second video plays ÷ impressions, as a percentage. The denominator is every time the video rendered on a screen, including the ones scrolled past in half a second and the repeat views of people who already saw it.
Autoplay is doing half the work. Nobody chose to press play, so a 3-second play means the thumb slowed down — not that anyone decided to watch your ad.
Worked example. A video ad spends €3,000 and buys 100,000 impressions. 12,000 of them reach 3 seconds, so Thumbstop = 12,000 ÷ 100,000 = 12%. You paid €3,000 for 12,000 three-second plays.
Recut the opening and lift it to 20%. The same €3,000 now buys 20,000 three-second plays — 8,000 more than before, bought with an edit rather than with budget, which is why the first 3 seconds are the cheapest thing in the account to fix.
It answers the question
Of everyone the video was shown to, how many stopped? A high rate means the opening earns attention; a low one means the ad is being scrolled past.
Why it matters
It's the first gate, and everything downstream is capped by it. Video Comp.Video Comp.Video completion rate. Helps you see which creatives keep viewers watching till the end, click rate and purchases can't beat a hook nobody stops for, so a weak Thumbstop tells you not to bother testing anything further down the ad.
It also reads fast. A video collects tens of thousands of impressions in a day or two, long before purchases have the volume to prove anything, so this is where you kill a bad concept in 48 hours instead of a fortnight.
What good looks like
No published standard applies here. Thumbstop moves hard with placement — a Reel autoplays into the whole screen and typically posts higher rates than a Feed video — so an account average blends two things that can't be compared. Judge each video against your own library on the same placement, and treat your best performer as the bar.
How to improve it
| Lever | What you do | Expect | How long | Watch out for |
|---|---|---|---|---|
| Fast Cut the first 2 seconds off | Delete the intro, the logo card and the establishing shot so the video opens mid-action | Thumbstop up 3–8 points | 2–4 days | The ad loses the branding that made it recognisable, which shows up later as weaker retargeting performance rather than in this column. |
| Fast Open on the product doing something | Lead with the movement, the result or the problem — never a person introducing themselves | Thumbstop up | 3–5 days | It restarts learning on that ad, so cost per purchase is unstable for a few days after the swap. |
| Slow Shoot vertical for the placement | Film 9:16 natively instead of cropping a landscape edit for Reels and Stories | Thumbstop structurally higher | 3–6 weeks | A second shoot for every concept. Production budget rises and each format takes longer to reach a readable result. |
| Slow Build a hook library | Test 3 openers against the same body edit, keep the winner, repeat monthly | A new winner every few rounds | 1 quarter | The 2 losers spend real budget while you learn, and splitting spend across variations slows down the ad that was already working. |
Every lever costs something somewhere. The last column is the one to read twice.
Read it with
The hook rate alone tells you people stopped. Next to completion, it tells you whether the rest of the video deserved them.
Stops them and holds them
The opening earns the attention and the edit keeps it. This is the concept to build a month of creative around.
Great hook, weak body
You stopped the scroll and then lost them. The idea works — the edit after the opener doesn't, so recut rather than rebuild.
Small crowd, loyal crowd
Few people stop, and the ones who do watch it out. That's the cheapest fix on this page — the video is fine, the opening isn't.
Nobody's watching
They scroll past, and the handful who don't leave early. Another caption or thumbnail won't move this one.
Together they separate attention from intent. A high Thumbstop with a low click rate means you stopped people with something that isn't your product — a shock opener, a face, a joke. A low Thumbstop with a healthy click rate means the few who stop are the right people, so the fix is reach, not persuasion.
This is where creative fatigue shows up before anywhere else. The opening is the part the audience recognises, so a Thumbstop sliding week on week while Frequency climbs means they've already seen this hook and their thumb no longer stops for it. Frequency alone can't tell you whether people are tired of the ad or merely counted twice.
Common misreads
The denominator is impressions, not viewers, and most of them never watched a frame. That's why Thumbstop reads far lower than a rate measured against people who chose to watch, and a figure near 90% would point to something broken in the reporting.
Stopping isn't buying. Curiosity openers and loud edits can stop the scroll and sell nothing. Read it beside the click rate and purchases before you scale on it.
Reels plays full-screen and Feed doesn't, so the two placements post different rates for identical work. Compare like with like, or the placement wins every test you run.
Also called
Hook rate · 3-second view rate · scroll-stop rate
See yoursYour Thumbstop by video creative, next to Video Comp. and the Spend behind each one.
Open Campaigns →