Hook rate and hold rate: what good actually looks like on Meta.

Most advertisers judge a video ad by its cost per purchase and stop there. That number tells you the ad failed. It does not tell you where. Hook rate and hold rate do, and they are the two metrics that turn a dead creative into a brief for the next one.

This guide covers what each metric actually measures, the thresholds worth acting on, how to read the two together as a diagnostic, and the traps that make both numbers lie to you.

What hook rate actually measures

Hook rate is 3-second video plays divided by impressions. It answers one question: did the opening frame stop the scroll. Nothing else in the ad matters until this number is healthy, because everything downstream is gated by it. A brilliant offer inside an ad nobody stops for is an offer nobody sees.

The important thing about hook rate is how narrow a slice of the ad it judges. You are not measuring the concept, the script, the product or the price. You are measuring roughly the first frame and a half: the opening image, the first motion, the first word of on-screen text. When hook rate is weak, rewriting the body of the ad is wasted effort.

What hold rate actually measures

Hold rate is typically measured as ThruPlays, or 15-second plays, divided by 3-second plays. It asks the second question: once someone stopped, did the ad earn the next ten seconds.

The denominator matters. Dividing by impressions instead of by 3-second plays gives you a blended number that moves when either the hook or the body changes, which makes it useless as a diagnostic. Keep hold rate conditional on the hook having worked, and the two metrics stay independent.

Working thresholds

Benchmarks vary enormously by vertical, placement, audience temperature and country, so treat published numbers as orientation, not as targets. As a working heuristic on cold Meta traffic, a hook rate around 25 to 30 percent is healthy, and below 15 percent is a problem worth fixing before anything else. For hold rate, somewhere around 15 to 20 percent of hookers reaching 15 seconds is a reasonable bar for a 30-second ad.

The honest version of this advice: the only benchmark that means anything is your own account. Pull the last 90 days, take the median hook rate across all video creatives that spent more than a token amount, and use that as your line. An ad in the top quartile of your own account is a winner regardless of what a blog post says the industry average is.

Reading the two together

The diagnostic power comes from the combination, and there are four cases.

Low hook, low hold. The creative is not landing at all. Do not iterate on it, replace the concept. Something about the angle is wrong for this audience.

Low hook, high hold. This is the most encouraging failure there is. The few people who stopped stayed, which means the message works and only the opening is broken. Keep everything after the first second and test five new openings against it. These are often the cheapest wins available in an account.

High hook, low hold. The opening is writing a cheque the ad does not cash. Usually a clickbait hook disconnected from the product, or a strong visual followed by a slow second act. Tighten the first five seconds after the hook, or make the hook honest about what follows.

High hook, high hold, still no conversions. The creative is doing its job and the problem is downstream: the offer, the landing page, or the tracking that reports the result. Do not touch the creative.

Why the averages lie

Two traps. First, placement blending. Reels, Stories and Feed produce very different hook rates for the same asset, because a Reels view starts full-screen with sound and a Feed view often does not. An account-level average across placements can hide a creative that is excellent in one and dead in another. Break the numbers out before you judge.

Second, the early-life distortion. In the first day of delivery, Meta shows an ad to the users it considers most likely to engage, which flatters both metrics. Wait for meaningful spend and a stable delivery pattern before you read either number as real. Killing an ad on day-one metrics is one of the more expensive habits in paid social.

What to do with each result

A weak hook is the easiest thing in performance creative to fix, because you can rebuild it without reshooting. Change the first frame to something with visual tension. Start mid-action rather than with a logo or an establishing shot. Put the promise in the first two words rather than the first sentence. Test the same body against several openings, which is exactly the isolate-one-variable discipline in our creative testing framework.

A weak hold is harder, because it usually means a structural problem: the ad takes too long to get to the point, the pacing sags, or the middle section explains something the viewer did not ask about. Cut the ad down rather than adding to it. Most 30-second ads with a hold problem are 15-second ads with 15 seconds of throat-clearing.

The takeaway

Cost per acquisition tells you an ad failed. Hook rate and hold rate tell you where, and where is the only part you can act on. Measure hold conditional on hook, break both out by placement, benchmark against your own account rather than an industry table, and wait for stable delivery before judging. Do that consistently and every dead creative starts producing a brief for the next one instead of just a bill. If your metrics look healthy and the conversions still are not there, the problem is usually further down the funnel, and a 30-minute audit will normally find it faster than another round of creative.