Hold rate is a video ad measure of how many of the people who started watching your ad kept watching for a meaningful length of time. The most common version divides ThruPlays by 3-second video plays. Meta does not offer a column called hold rate, so advertisers build it themselves as a custom metric, which means definitions vary from one account to the next.
How hold rate works
Meta reports two figures that make the calculation possible. A 3-second video play counts each time your video plays for at least three seconds. A ThruPlay counts each time it plays to completion, or for at least 15 seconds if it is longer than that. Divide the second by the first and you have the share of people who got past the opening and stayed.
For example, a video with 8,000 three-second plays and 1,600 ThruPlays has a hold rate of 20%. Some advertisers use a different numerator, such as plays to 50% from the video play quartile columns. Either approach works, as long as you use the same definition every time you compare.
Hold rate is the partner of hook rate. Hook rate tells you whether the opening stopped people scrolling; hold rate tells you whether the rest of the video earned their attention. To add it in Ads Manager, open the column settings, create a custom metric with the formula you have chosen and save it to a column preset so it appears in every report.
Why it matters
Hold rate isolates the middle of the video, which is where most ads quietly fail. A strong hook paired with a weak hold usually means the opening promised something the rest did not deliver, or the pace dropped after the first line. Neither problem shows up in click-through rate until much later.
It matters most when the message needs a few seconds to land. A private dental clinic explaining implants, a B2B software firm showing its product, or a home improvement company walking through a kitchen fit all tend to reach the offer after the fifth second. If almost nobody is still watching by then, the ad cannot do its job, however good the targeting. People who watch a large share of a video can also be gathered into a video view audience for retargeting, so a better hold rate builds a bigger warm audience at the same time.
Common mistakes
- Comparing videos of very different lengths. A 10-second video counts a ThruPlay on completion, so its hold rate will look far better than a 45-second video’s.
- Treating hold rate as the goal. A video people watch to the end but never act on is still a poor ad; conversions decide.
- Reading it from a few hundred plays. Small samples swing wildly.
- Mixing definitions between reports, or between you and whoever manages the account.
- Cutting a video so tightly for hold rate that the offer or the reason to act disappears.
How to act on it
Add the custom metric and look at it for every video ad alongside hook rate, cost per result and conversions. Use the placement breakdown to compare Reels, Stories and Feed separately, because full-screen placements behave differently from the scrolling feed.
When hold rate is low, open the quartile columns to see where viewers leave, then re-edit that stretch. The usual fixes are cutting a slow introduction, showing the product or the result sooner, adding captions so the story works without sound, and changing the shot every few seconds. Test the new edit against the old one rather than replacing it outright.
If you would rather someone else did the cutting and testing, it is part of how I run Facebook video, Reels and Stories ads.