Meta Thumb-Stop Testing

Thumb-stop testing means testing the first one to three seconds of a video ad, or the first glance at a static ad, to see which opening makes people stop scrolling. If nobody stops, nothing else in the ad matters. Ads that hold attention also tend to get cheaper delivery.

How it works

People scroll fast, so an ad has a second or two to earn attention. The thumb-stop rate (also called hook rate) is the share of impressions that turn into a 3-second video view. It is an early signal: ads that hold attention often do better on clicks and conversions too. Thumb-stop testing keeps everything else fixed and changes only the opening so you can see what works.

How to do it

  1. Create a custom metric in Ads Manager: 3-second video plays divided by impressions. Use it to benchmark your current ads.
  2. Make several versions of one ad that differ only in the first three seconds: opening frame, text overlay, or audio.
  3. Put them in one campaign and one ad set with broad targeting so they compete on equal terms.
  4. Once each version has enough impressions (decide the threshold in advance), rank them by thumb-stop rate. Pause the weakest and keep the best few running to collect conversion data.
  5. Record winning openings in a shared hook library and reuse the patterns in new ads.

What to measure

  • Thumb-stop rate (hook rate): 3-second video plays divided by impressions.
  • Hold rate: views to 50% (or ThruPlays) divided by 3-second plays.
  • Hook vs CPA: compare thumb-stop rate with cost per acquisition across ads to check that attention leads to sales.

Best practices

  • Put the most interesting moment in the first frame. Do not build up to it.
  • Test text-on-screen openings against visual-only openings.
  • Design for sound off. Many people watch feed video muted, so the opening must work on visuals and text alone.

Common pitfalls

  • Changing the body, offer, or audience at the same time as the hook, so you cannot tell what caused the difference.
  • Optimizing only for thumb-stop rate. A misleading hook stops people who never buy.
  • Stopping after one round. Keep testing and keep adding to the hook library.