If you want to know how to calculate hold rate for video ads, the short answer is: divide the number of seconds (or the percentage point) of your video that viewers actually watched by the total length of the video, then compare that against your average hook rate and drop-off curve. Hold rate tells you whether people who stopped scrolling for your hook actually stayed for the story, the proof and the offer — which is a different question from whether they stopped scrolling at all.
What hold rate actually measures
Hook rate tells you whether your opening three seconds earned attention. Hold rate tells you what happened after that — whether the rest of the video kept people watching once they had already stopped. A video can have a strong hook and a weak hold rate, which usually means the opening promise was good but the payoff, pacing or proof in the middle of the ad let people down.
Hold rate is a retention metric, not a conversion metric. It won't tell you who clicked or bought. It tells you where attention leaked out of your video, second by second, so you know which part of the creative to fix before you touch the offer or the landing page.
How to calculate hold rate for video ads: the basic formula
The simplest version of how to calculate hold rate for video ads uses average watch time against total video length:
- Hold rate (%) = (average watch time in seconds ÷ total video length in seconds) × 100
- Example: a 30-second video with an average watch time of 12 seconds has a hold rate of 40%
- Example: a 15-second video with an average watch time of 9 seconds has a hold rate of 60%
This single number is useful as a quick health check and for comparing creatives of the same length. It gets less useful when you're comparing a 15-second ad against a 30-second ad, because shorter videos naturally post higher percentage hold rates — there's simply less time for attention to drift. When comparing across lengths, look at absolute watch time in seconds, not just the percentage.
Using platform view-through metrics instead
Most ad platforms report retention in bands rather than a single average — for example, the percentage of viewers who reached 25%, 50%, 75% and 95% (or completion) of the video. This is actually a better way to calculate hold rate because it shows you the shape of the drop-off, not just one average point.
To build your own hold rate curve from these bands:
- Pull the percentage of viewers who reached each quartile (25%, 50%, 75%, 95% or completion) for the ad.
- Plot those four numbers against time (not against percentage) using your video's actual length — a 30-second ad's quartiles sit at roughly 7.5s, 15s, 22.5s and 28.5s.
- Calculate the drop between each point: quartile 1 minus quartile 2, quartile 2 minus quartile 3, and so on.
- Flag the single biggest drop. That's the exact moment in the ad where attention is leaving fastest, and it's the first place to edit.
This quartile method is what most experienced creative teams actually use day to day, because a single average hold rate number can hide a brutal early drop-off behind a decent average.
Reading the shape of your hold rate curve
Two ads can have the same overall hold rate number and completely different problems. A curve that falls sharply in the first few seconds after the hook usually means the hook made a promise the next scene didn't deliver on. A curve that stays flat through the middle but collapses near the end usually means the offer or CTA section feels like a sales pitch that people were happy to scroll past.
| Drop-off pattern | Likely cause | What to fix first |
|---|---|---|
| Steep drop right after the hook (0–3s to ~25%) | Hook over-promised, or the transition into the next scene is slow or confusing | Rewrite the first 2–3 seconds after the hook so it pays off the opening line immediately |
| Gradual, steady decline across the whole video | Normal and expected — not necessarily a problem | Compare against your own benchmark ads before changing anything |
| Drop in the middle (around 50%) | Pacing sags; one scene runs too long or explains too much | Cut or compress the slowest middle scene; move proof earlier |
| Sharp drop near the CTA (75%–95%) | Offer or CTA feels like a hard pitch after a softer story | Soften the CTA language, shorten the pitch, or move a stronger benefit closer to the end |
| Flat, high retention almost to completion | Strong structure throughout | Test this ad's hook and offer against other variants to find your next benchmark |
Hold rate benchmarks: what to treat as a starting point
Avoid chasing a fixed industry number — hold rate varies by platform, placement, video length and category, and the only benchmark that matters long-term is your own account's history. As a starting point to test against your own data: for a 30-second vertical ad, treat anything holding under roughly a third of viewers to the halfway point as worth investigating, and treat a video that holds most viewers to around 75% of its length as a strong result worth reusing as a structural template. Run this check against a handful of your own past ads first, then set your real benchmark from that.
A simple workflow for using hold rate in creative testing
Hold rate is most useful inside a repeatable testing process, not as a one-off report you check after a campaign ends.
- Pull hook rate (3-second view rate) and hold rate (quartile drop-off) for every active creative on the same schedule.
- Separate ads into two buckets: strong hook / weak hold, and weak hook / strong hold. These need completely different fixes.
- For strong hook / weak hold ads, locate the steepest drop using the quartile method above and rebuild only that scene.
- For weak hook / strong hold ads, keep the body of the ad and test new opening lines or visuals against it.
- Re-test the edited version as a new creative, not as a replacement of the old one, so you can see the before/after retention curve side by side.
- Log the winning structure (where the proof sits, how long the problem section runs, where the CTA starts) so your next batch of ads starts from a pattern that already held attention.
If you're producing several variants at once to run this kind of test properly, see how to batch produce video ads for testing for a structured way to generate and organize multiple versions without losing track of what changed between them.
Hold rate vs. hook rate vs. click-through rate
These three metrics answer different questions, and treating them as interchangeable is a common mistake in creative testing.
| Metric | Question it answers | When to act on it |
|---|---|---|
| Hook rate | Did the first few seconds stop the scroll? | Low hook rate: rewrite the opening line or visual before anything else |
| Hold rate | Did people stay once they stopped? | Low hold rate with a good hook: fix the specific scene where the curve drops |
| Click-through rate | Did the video move people to act? | Good hold rate but low clicks: the CTA or offer itself needs work, not the retention |
A video can hold attention well and still convert poorly if the call to action is unclear or buried. If your hold rate looks healthy but clicks are flat, the problem usually sits in the CTA or offer, not the pacing — see how to improve click-through rate for ads and how to interpret click-through rate for video ads for that half of the picture.
Where hold rate problems usually come from
A handful of structural issues account for most low hold rates, and they're worth checking before you rewrite anything from scratch:
- A hook that's a question or a claim with no immediate visual payoff in the next 1–2 seconds
- A problem/setup section that runs too long before showing the product or benefit
- On-screen captions that lag behind the voiceover, making the ad feel slower than it is
- A proof section (reviews, before/after, demo) placed too late, after attention has already started to drop
- Pacing built for sound-on viewing when a large share of your audience watches on mute — design the visuals and captions to carry the story on their own, then test with sound
Checking your ad on an actual phone, not just a desktop preview, catches a lot of these issues early — see how to optimize video ads for mobile viewing for a practical checklist.
Fixing a low hold rate without starting over
You rarely need to scrap an underperforming ad entirely. Once you know the exact timestamp where the drop happens, the fix is usually one of these:
- Trim the scene before the drop point by a few seconds rather than rewriting it completely
- Move your strongest proof point earlier if the drop happens before proof appears
- Add a visual change — a new shot, a text callout, a zoom — right at the drop point to re-earn attention mid-video
- Shorten the CTA if the drop happens near the end, so the pitch feels less like a hard close
- Keep the hook and the first scene; only replace the section after the drop, then re-test as a new version
This is where having a fast way to produce and adjust variants matters. FrameNotion builds every ad from scratch with a defined structure — hook, problem, benefit, proof, offer, call to action — which makes it straightforward to pinpoint which section maps to a hold rate drop. Once an ad is live, you can request text and color changes without starting a new ad from zero, and Pro and Agency plans support A/B hook variants, so you can test a new opening against the same body once you know the hook isn't the issue. A finished ad takes about 10 to 20 minutes to generate, which makes it practical to test a fix the same day you spot the drop-off in your data.
Putting hold rate to work across your testing calendar
Treat hold rate as a diagnostic tool you check on a schedule, not a one-time grade. Pull it alongside hook rate and click-through rate every time you review a batch of creatives, flag the steepest drop on each underperformer, and feed that into your next round of edits. Over a few testing cycles, you'll build a working sense of where your audience's attention typically breaks — and a library of structures that reliably hold it, which you can reuse as the starting point for your next product or offer. If you're setting up a style guide to keep that structure consistent across a team, see how to create a brand style guide for video ads.
Frequently asked questions
Is hold rate the same as video completion rate?+
No. Completion rate is a single figure for the percentage of viewers who watched to the very end. Hold rate usually refers to the broader retention curve across the whole video, including the quartile drop-offs along the way, not just the final number.
What's a good hold rate for a 15-second ad versus a 30-second ad?+
Shorter ads tend to post higher percentage hold rates simply because there's less time for attention to drift, so comparing a 15-second and a 30-second ad by percentage alone is misleading. Compare absolute watch time in seconds, or compare each ad against its own past performance at that same length.
Can a high hold rate still mean a bad ad?+
Yes. A video can hold attention well through pacing or music and still fail to generate clicks if the offer is unclear or the call to action is weak. Always check hold rate alongside click-through rate before judging an ad a success.
How often should I recalculate hold rate during a test?+
Check it as soon as a creative has enough views to produce a stable quartile breakdown, then again at the midpoint and end of the test window. Checking too early on low volume can show a misleading curve.
Does hold rate differ a lot between platforms?+
Yes, because feed behavior, autoplay, and typical viewing context differ by placement. Use hold rate to compare creatives within the same placement and platform rather than across very different environments.