What is hold rate?

The stretch after the hook

Every video ad loses viewers as it plays, and hook rate only describes the first cut. Hold rate picks up where it stops. Of the people who gave the ad three seconds, how many went on to a meaningful view? The idea is simple, but the name covers more than one calculation, so it pays to write the formula next to the number in any report. The most common version divides ThruPlays by 3-second video plays, which makes hold rate a conditional figure: it ignores everyone who scrolled straight past and looks only at those the opening already caught. That is what makes it useful for diagnosing the middle of a video separately from its first moments.

What a ThruPlay counts

Meta defines ThruPlays as the number of times a video was played to completion or for at least 15 seconds. For a video of 15 seconds or less, that means a complete play; for anything longer, 15 seconds is enough. The consequence is easy to miss. For a 12 second ad, hold rate measures how many hooked viewers reached the end. For a 45 second ad it measures only whether they reached the fifteen second mark, and says nothing about the half minute that follows. Some teams divide ThruPlays by impressions rather than by 3-second plays, which blends hook and hold into one number. Neither version is wrong, but they cannot be compared with each other.

Reading hold rate beside hook rate

The two numbers together sort ads into four groups. A high hook rate with a low hold rate means the opening promised something the next seconds did not deliver, or the middle is slow: tighten it or make the opening more honest. A low hook rate with a high hold rate means the content works for people who give it a chance but the first frame is not earning that chance, so test new openings on the same body. Both low suggests the concept or the audience is wrong. Both high means the creative is doing its job and attention can turn to the offer, the landing page or the targeting.

When hold rate matters

Hold rate matters most when the part of the video that sells comes after the opening. A UGC ad that states a problem, shows the product solving it and then names the price needs viewers to stay past the first beat, so a weak hold rate means most of the audience never reached the reason to buy. It matters less for a six second bumper whose whole message is the first frame and the logo. It is also a useful tie breaker between two openings with similar hook rates: the one that keeps more of its viewers usually made a promise the rest of the video could keep, which tends to show up later in clicks and sales too.

Common mistakes

Ranking ads of different lengths against each other by hold rate is the classic error, because the ThruPlay rule measures a short ad to its end and a long ad only to fifteen seconds. Mixing the two formula versions across weeks or clients produces trends that are really changes in method. Treating TikTok's 6-second views as a ThruPlay equivalent overlooks that they also count an early engagement. Reading hold rate on a handful of plays magnifies chance. And optimising for hold rate alone can reward a video that keeps people watching without ever asking them to do anything, which is a pleasant video rather than a working ad.

What the video builder controls, and what it does not

Hold rate is measured in Meta Ads Manager or TikTok Ads Manager once the exported MP4 is running as an ad; the video builder itself reports no views or retention. What the builder does set is the length and the pacing that the metric responds to. The length options are short, standard and long, aimed at about 10, 18 and 30 seconds, with scene budgets of 3 to 6, 4 to 9 and 6 to 12 scenes. A short video sits under the fifteen second mark, so its ThruPlays are completions; standard and long videos pass it, so their ThruPlays stop at fifteen seconds. UGC videos stay under 30 seconds with at most three talking scenes, and the UGC ad script guide covers how to structure the beats that carry viewers through the middle.

Questions people ask