The short answer
Hook rate is 3-second video views divided by impressions. Practitioner benchmarks for Meta feed video cluster around 20% to 35%, with 30%+ usually called strong, but treat these as directional. On our board of 1,752 winning video ads, the shortest openers (4 words or fewer) reach the top tier 11.9% of the time, above the ~10% board average.
Hook rate, defined
Hook rate is the share of impressions that turn into a 3-second video view:
Hook rate = 3-second video plays ÷ impressions × 100
Some teams call it thumbstop ratio. It measures one thing: whether the first moment of the ad stopped the scroll. It says nothing about whether people kept watching (that is hold rate) or bought.
Meta reports 3-second video plays as a standard column. Hook rate itself is not an official Meta metric, so you build it as a custom metric in Ads Manager.
Benchmarks: what practitioners publish
There is no official benchmark. The numbers below come from agencies and tool vendors. They broadly agree with each other, but they are opinions built on their own client accounts, so read them as directional.
| Source | Published range for hook rate |
|---|---|
| Motion | Below 25% needs work, 25% to 35% solid, above 35% strong. Suggests 30% to 40% as a baseline target |
| Vaizle | Aim for about 20% to 25%, with top performers above 30% |
| Kickbite | Feed video prospecting commonly sits between 15% and 35% |
Put together: under 20% is usually a creative problem, 25% to 30% is a reasonable middle, and above 30% is good for cold feed traffic. Reels and Stories behave differently from feed, and warm audiences hook more easily than cold ones, so compare like with like.
The better benchmark is your own. Pull the last 90 days of video ads, sort by hook rate, and use your median as the bar every new opener has to clear.
What we can and can't measure
The Meta Ad Library doesn't publish impressions or view counts for commercial ads, so nobody outside an account can see a competitor's hook rate. Any tool that claims to show it for ads it doesn't manage is guessing.
What we can see on the 2,443 ads on the Ad Radar board (all live, all with $50k+ in estimated spend) is the opener itself, how long each ad has run, how many variants it carries and its estimated spend. An ad that keeps its opening line for 200 days is an ad whose first three seconds the brand has stopped worrying about. That makes the openers worth studying even without the view data.
How winners use the first three seconds
The median spoken opener on the board is 11 words. At a normal speaking pace, that sentence is still running when the 3-second mark passes. So for most winners, the 3-second view is earned by the first few words plus the picture, not by the full sentence.
We grouped the 1,752 video ads with transcripts by the length of their first sentence:
| First sentence | Ads | Share | Reach top tier | Median est. spend |
|---|---|---|---|---|
| 1 to 4 words | 193 | 11% | 11.9% | $305K |
| 5 to 8 words | 414 | 24% | 8.0% | $222K |
| 9 to 12 words | 417 | 24% | 10.1% | $241K |
| 13 to 20 words | 525 | 30% | 8.2% | $213K |
| 21 words or more | 203 | 12% | 10.3% | $246K |
The shortest group stands out on both measures: 11.9% reach the top tier (against about 10% for the board) and the median estimated spend is $305K, the highest of the five. These are not complete sentences. They are reactions, numbers and fragments that make sense only with the picture.
Oh, no way!
- Est. spend
- $9.7M
- Days live
- 306
- Format
- Video, 48s
- Variants
- 2
Three words of reaction while the product is already in frame. The line explains nothing, so the eye has to look at the chessboard to find out why.
$2,300 for that.
- Est. spend
- $4.9M
- Days live
- 233
- Format
- Video, 57s
- Variants
- 2
A price with no context. The viewer has to stay to learn what costs $2,300 and why that matters, and the script answers in the next line with the at-home kit.
Okay, 5'10, 200 pounds.
- Est. spend
- $4M
- Days live
- 154
- Format
- Video, 58s
- Variants
- 8
Height and weight before anything else. The viewer instantly compares the body on screen to their own, which is the whole hook. Eight variants suggests the brand keeps re-cutting around it.
The same shape shows up across niches. The Farmer's Dog opens on a name, "Meet Beverly." (est. $5.44M, 89 days). Omnipod opens on a pun, "Nobody likes a prick." (est. $4.50M, 69 days, 7 variants). Melinda Maria Jewelry opens on "Excuse me." (est. $2.57M, 191 days, 5 variants). In each case the words are an invitation to look, and the picture carries the information.
Long openers can also work (21+ words sit at 10.3%), but they work differently: the first words are a full claim or story setup, and the rest of the sentence pays off after the 3-second mark. The weak zone is in between. Openers of 5 to 8 words and 13 to 20 words both sit around 8%. Many of those are generic setups ("If you have X, watch this") that are too long to be a reaction and too empty to be a claim.
Six ways to raise your hook rate
These come from what the winners on the board do, not from your account data, so test them against your own median.
- Put the product or the problem in frame on frame one. A reaction line ("Oh, no way!") only works if the thing being reacted to is visible. Don't spend the first second on a logo or a face looking at nothing.
- Try a 1 to 4 word opener. A number, a price, a reaction or a name ("This is Riley.", "Four days."). It forces the viewer to look for context in the image.
- Lead with a specific number. Specific-number openers have the highest top-tier rate among common hook patterns on our board (12.9%). See number hooks.
- Cut the throat-clearing. "Hey guys, so today I want to talk about..." spends the whole three seconds saying nothing. Start the edit on the first word that carries meaning.
- Add on-screen text that says something different from the voice. Sound-off viewers only get the text. If it repeats the voiceover, you've wasted half the hook. More in text overlay hooks.
- Test openers on one body. Change only the first 3 seconds and keep the rest identical. Then a hook rate difference is caused by the hook, not by something later in the edit.
Reading hook rate with the other numbers
| Hook rate | Hold rate | What it usually means | What to change |
|---|---|---|---|
| Low | High | The few who stay love it | The opener. Keep the body |
| High | Low | The opener promises something the ad doesn't deliver | The 3 to 15 second section |
| High | High, poor CPA | Attention is fine, the offer or landing page isn't | Offer, landing page, audience |
| Low | Low | The concept isn't landing | New concept |
A high hook rate on its own is not a goal. A shock frame can stop everyone and sell to no one. The ads on our board that run longest pair a strong opener with a body that keeps the promise the opener made. The first 3 seconds of a video ad breaks down what happens visually and verbally in that window, and the pillar list of real openers has 60 lines to test against your own median. To pull the openers your competitors keep running, you can search transcripts on the Ad Radar board.
Figures marked as estimated spend come from Ad Radar's model of engagement on public Meta Ad Library ads. They are estimates, labeled as such, and are best used to rank ads against each other.