Reading hold rate as a creative signal
Hook rate says whether the opening worked. Hold rate says whether the rest of the video earned the attention. Together they show where a video loses people:
| Hook rate | Hold rate | Likely problem |
|---|---|---|
| Low | High | Good video, weak opening — test new first seconds |
| High | Low | Strong opening, but the middle loses people — tighten the pacing or pay off the hook sooner |
| High | High | Attention is there — if it isn't selling, look at the offer, call to action or landing page |
| Low | Low | Rethink the concept |
Read hold rate on videos of similar length and format, and use it to decide what to change in the next version: the opening, the pacing or the concept itself. For short product videos made from a winning still, see AI video ads, and for openings to pair with a strong middle, the hook bank.
How hold rate is calculated
Like hook rate, hold rate is usually a custom metric built from Ads Manager columns, and its definition varies more than most. The most common version:
Hold rate = ThruPlays ÷ 3-second video plays × 100
A ThruPlay is Meta's count of plays that reach the end of the video, or at least 15 seconds for longer videos. Dividing by 3-second plays asks: of the people the opening stopped, how many stayed?
Other versions you'll see:
| Version | What it measures |
|---|---|
| ThruPlays ÷ impressions | Hook and hold blended into one number |
| 50% video plays ÷ 3-second video plays | Staying to the middle — handy for longer videos |
| Average play time ÷ video length | The share of the video watched, on average |
None of them is wrong. Pick one, write it down and use it consistently, so you're comparing like with like.
What counts as a good hold rate?
It varies widely with video length, placement, category and the formula you use — a six-second clip and a sixty-second story will never be comparable. Compare against your own videos of similar length and format rather than a published benchmark.
How to improve it
- Pay off the hook quickly. If the opening asks a question, answer it soon.
- Cut the slow parts. Remove intros, logo reveals and pauses.
- Keep the picture moving without making it chaotic.
- Show the product early, so viewers know what the video is about.
- Keep test videos short. They're quicker to make and easier to watch to the end.
For very short clips, such as the few-second product videos from BulkAds' AI video ads, the opening does most of the work, so watch hook rate and sales first. Whatever the length, judge a video on purchases as well as attention — the scoreboard in creative analytics ranks every ad by spend, purchases, cost per purchase and ROAS.
Questions people ask
What is a ThruPlay?
A ThruPlay is Meta's count of video plays that reach the end of the video, or at least 15 seconds for longer videos. It's the usual top half of the hold rate formula.
Is hold rate the same as video retention?
They're closely related. Retention usually means the whole drop-off curve across a video. Hold rate boils it down to one number, such as the share of 3-second viewers who reach a ThruPlay.
What's a good hold rate?
It depends heavily on video length, placement and which formula you use, so there's no reliable universal benchmark. Compare against your own videos of similar length.
Related terms
Put it to work
- Features
AI video ads
Turn a winning still into a Reel
- Features
Creative analytics
Which ad actually sold — and why
- Guides
Static vs video ads: which to make, and when
Statics are cheap to make and quick to test, video earns its place in Reels and Stories. An honest comparison, the metrics for each and a plan that uses both.
BulkAds AI is an independent product and is not affiliated with, endorsed or sponsored by Meta Platforms, Inc. Facebook and Instagram are trademarks of Meta Platforms, Inc.