How to Check Hook Rate in Facebook Ads

Someone in a Slack group posts a screenshot: 34% hook rate on a new video, everyone piles in with congratulations. You open Ads Manager, work out your own number, and get 19%. Before you rewrite the opener, ask what the other person divided by. There is a decent chance you two computed different things and neither of you said so.

Short answer: Meta Ads Manager has no hook rate column, so you build it as a custom metric: 3-second video plays divided by impressions, formatted as a percentage. The denominator is the whole argument. Swap impressions for reach or video plays and the same ad reports a different hook rate.

The takeaways

• Meta gives you no hook rate column: you build it from 3-second video plays over impressions, so two accounts quoting hook rates are not necessarily measuring the same thing.
• A 3-second play is not a decision to watch: in autoplay placements the feed starts the video for the viewer, which makes hook rate placement-specific and a Reels number incomparable to a Feed number.
• Hook rate grades the opener, hold rate grades everything after it: together they separate "nobody stopped" from "they stopped and left", and neither is a target worth optimising on its own.

Where do you find hook rate in Ads Manager?

You do not. Meta Ads Manager ships no hook rate column, so the number has to be assembled by hand. Vixen Digital's hook rate guide puts it plainly: "Hook Rate isn't a standard metric you'll find in Meta Ads Manager, so, you'll need to set it up manually." The build takes two minutes. Open Columns, choose Customise columns, click Create custom metric, then divide the field called "3 second video plays" by "Impressions" and set the format to percentage.

The consequence is more interesting than the clicks. Hook rate is a homemade metric, so every version of it carries the assumptions of whoever built it. When a number gets quoted in a thread or a client deck, the formula behind it travels nowhere. That is why benchmark shopping goes wrong so often: you are comparing your arithmetic to a stranger's arithmetic with no way to check whether the two match.

Why does the denominator change the answer?

Because the three candidates count different things. Impressions counts every time the ad was served. Reach counts unique people, so anyone who saw the ad four times lands in the denominator once. Video plays counts video starts, which is a smaller set than impressions.

Take an ordinary made-up week: 10,000 impressions, 7,000 reach, 9,200 video plays, and 2,400 3-second plays. Against impressions that is a 24% hook rate. Against reach it is 34%. Against video plays it is 26%. Same creative, same week, three defensible numbers with a 10-point spread between the extremes.

The industry does not agree either. In the same comparison, AdSights defines hold rate against 3-second video views and ThruPlay rate against video plays, so even published definitions of neighbouring metrics sit on different bases. Pick impressions, write it down, and never change it mid-quarter.

What does a 3-second video play count?

Three seconds of continuous playback, with or without sound. That last part matters more than it looks. In feed placements the video autoplays muted, so a 3-second play can mean nothing more than the ad sitting in the viewport while somebody read the caption underneath it. Nobody chose to watch anything.

Placement mechanics then push the number around on their own. Reels opens full-screen with the video already running, so its structural floor sits well above an in-feed placement where the ad competes with everything around it. A Reels hook rate and a Feed hook rate are two different measurements wearing the same name.

The trap this sets: if the ad set runs Advantage+ placements, your blended hook rate is a weighted average across placements you did not pick. Shift the delivery mix and the number moves while the creative sits untouched. Segment by placement before you conclude anything about the opener.

What does hook rate tell you that hold rate does not?

Hook rate grades the first three seconds. Hold rate grades what happens to the people who got past them: AdSights calculates it as 15-second video views divided by 3-second video views, which deliberately restricts the denominator to viewers who already stopped scrolling.

Read as a pair, the two localise the problem. Low hook with high hold means a weak opener is starving an ad that the people who do stay quite like, and rewriting three seconds is the cheapest fix in your account. High hook with low hold means the opener promised something the rest of the ad did not deliver. When both are low, the concept itself is what needs replacing.

ThruPlay rate is a third thing again (ThruPlays over video plays) and it blends hook and hold into one figure, so it cannot tell you which half broke. If you want the fuller picture, read your 25/50/75/100% drop-off curve instead of a single retention percentage.

Should you optimise for hook rate?

No. Hook rate is a diagnostic, not a scoreboard. It is trivially inflatable: a loud pattern interrupt in second one will lift it without moving a single sale, which is exactly how accounts end up with scroll-stopping creative and flat revenue. Always read it next to cost per result. A hook rate climbing while cost per result stays flat means the opener earned attention that the rest of the ad wasted.

The other discipline is patience. Hook rate on a few hundred impressions swings hard, and a first-day number tells you close to nothing. Let it stabilise before you judge an opener on it, then go and work on the words in the first three seconds.

Knowing what a number counts before arguing about whether it is good is most of the job. That is the ad-intelligence work in Adscalr: hook rate is one of six weighted metrics in the composite score, so it cannot win a test alone, and Bayesian shrinkage with format-specific priors pulls a new ad's wild early hook rate toward what its format normally does, so a lucky first day does not get crowned.