Hook rate in fitness ads: the metric nobody checks that decides everything else
Hook rate is the percentage of your video's impressions that reach 3 seconds of playback. Above 25%, your ad stops the thumb. Below 15%, it doesn't matter how good the rest of the video is, the offer, or the landing page: almost nobody will get that far. It's the single most decisive metric for a video ad, and it doesn't appear in the default columns of Ads Manager, so 95% of gym owners have never seen it.
The logic is brutally simple. The average user scrolls metres of feed every day and gives each piece a fraction of a second of pre-conscious attention. In that first second or second and a half, without actually deciding, they decide whether the thumb keeps scrolling or stops: the thumb-stop. Everything that comes after (your argument, your social proof, your offer, your CTA) only exists for the people who cleared that filter. A brilliant video with a bad first second is a beautiful shop with the shutters down.
As we covered in why video ads generate more engagement, video is the dominant format in fitness. This article is the missing piece: how to know, with a single number, whether your videos are capitalising on that advantage or throwing it away in the first second.
How to measure it in Ads Manager, step by step
Meta doesn't hand you the hook rate ready-made; it gives you the ingredients. To build it:
- In Ads Manager, ad view, click Columns → Customise columns.
- Add 3-second video views and Impressions.
- While you're there, add Video plays at 50% and at 100%, plus ThruPlays: you'll need them for the diagnostic in the last section.
- Save the column set as a preset ("Video - diagnostic") so you don't have to repeat this every week.
The calculation: hook rate = 3s views ÷ impressions × 100. Meta also offers a "2-second continuous video views" column, which is stricter and closer to the real thumb-stop; if you use that one, lower the thresholds by 20-25%. The important thing is to pick one and always compare against the same metric.
Thresholds for fitness video in feed and reels:
| Hook rate (3s/impressions) | Reading |
|---|---|
| > 30% | Excellent: the hook is an asset, clone it |
| 25-30% | Good: above average |
| 15-25% | Mediocre: half-working, clear room to improve |
| < 15% | Problem: most people don't even know you exist |
| < 10% | The video is dead on arrival; change the hook now |
Honest caveat: some of those "3-second views" are autoplay with the user looking at something else, so the absolute number is inflated equally for everyone. That's why hook rate is most useful for comparing your own ads against each other, not for boasting about the absolute figure.
8 hooks that stop the scroll in fitness
The hook is not your gym's logo with a wipe transition. It's what happens on screen between second 0 and second 2. Eight patterns that work in local fitness, with a concrete example each:
- Unexpected movement. Something that breaks the feed's physics: a medicine ball dropping into the lens, someone jumping into frame, a deadlift drop with the sharp sound. The reptilian brain turns its head at sudden movement; you can't train that away, you can only use it.
- Counter-intuitive on-screen text. "Stop doing cardio." "Don't come to this gym if you just want to scroll your phone." "Crunches are the worst exercise for your abs." The mismatch with expectations forces you to read the second line.
- Hyperlocal result. "Marta, from Chamberí, 11 kg in 6 months" over a shot of Marta training. A neighbourhood name stops the thumb of someone who lives there better than any special effect: it's talking about me.
- Direct question to camera. Someone (the owner, a trainer) looking straight into the lens: "Have you been paying for a gym you don't go to since January?" Eye contact triggers social attention; feeling addressed stops the scroll.
- The mid-process shot. Starting the video in the middle of something: a packed class at peak chaos, a transformation halfway through. Incompleteness generates the question "what's going on here?", and that question is worth 3 seconds.
- The big number. "€9,640" (the cost of a decade of unused gym membership). "47" (the people who started this month). A number without context in the first frame is a clean curiosity hook.
- The common mistake. Showing someone doing a familiar exercise wrong, with a caption: "this is destroying your back." Half the gym audience wonders "am I doing it wrong too?"
- Reversed before/after. Start with the after and rewind. The before/after format is so overused that the eye skips it; in reverse, it throws people off just enough.
The first four are the most reliable starting point. And a coherence warning: the hook has to connect with what comes next. A high-impact hook followed by an unrelated video achieves a high hook rate and a disastrous 25% retention: you stopped the thumb only to disappoint it. How to structure the rest of the ad to deliver on the hook's promise is covered in the anatomy of a gym ad that converts.
The diagnostic: each metric points to a different culprit
This is the real value of setting up those columns: the video's internal funnel tells you exactly which part to fix. Not "the ad isn't working", but "second 1 isn't working" or "the CTA isn't working".
| Metric | Problem threshold | Diagnosis | What to fix |
|---|---|---|---|
| Hook rate (3s/impr.) | < 15% | The first second isn't stopping the thumb | Only the hook: new first frame, new opening text. Don't touch anything else |
| 50% retention (50%/3s) | < 30% | Hook promises, development bores or disappoints | The body of the video: shorten it, cut the corporate intro, get to the point faster |
| Full completion (100%/3s) | < 10-15% | Video too long for its content | Cut to 15-30 seconds; in local, less is more |
| CTR with hook rate > 25% | < 0.8% | People watch happily and don't click: CTA or offer isn't compelling | The offer and call to action, not the video |
| High CPL with everything else healthy | , | The problem is post-click | Landing page or form; the video is innocent |
This table prevents the most expensive mistake in video marketing: rebuilding the entire ad when 80% of it was working. If your hook rate is 32% and your CTR is 0.5%, shooting another video throws away the good work; the broken part is the offer. And the other way round: with a 9% hook rate, optimising the CTA is rearranging the furniture in a shop nobody enters.
The same logic applies to wear-out: a hook rate that drops week on week in an ad that was working is the subtle version of the CTR signal in detecting creative fatigue. The hook is the first thing to burn, because it's the only part the entire audience has definitely seen.
How to test hooks without burning budget
The hook has a property that makes it perfect for small budgets: you can test three different hooks on the same video. Same body, same offer, same CTA; only the first 2 seconds change. Nearly zero production cost, and the comparison is clean because everything else is identical.
Using the mechanics from the testing framework for gyms: 2-3 hook variants, €50 per variant, 3-4 days, and hook rate as the primary decision metric (with a follow-up check that the hook rate winner also wins on CPL; they almost always match, but "almost" exists).
A typical winning hook rotation at a real gym: the same 20-second class video, with three different openings (question to camera, counter-intuitive text, local result), goes from a 14% to a 27% hook rate just by choosing the right opening. Same media spend, roughly twice as many people seeing your message. There's no other optimisation in the entire account with that effort-to-result ratio.
What about static image ads? The concept applies just the same even though the metric doesn't exist: in static, the "hook" is the dominant visual element and the headline, and its thermometer is CTR. Almost all the principles from the 8 patterns above translate directly. What angles and concepts to put behind that first impression is covered in the guide to creatives for gym ads.
The minimum routine
Once a month, with the saved column preset: hook rate and 50% retention for each active video, checked against the diagnostic table. Fifteen minutes. Videos with a hook rate below 15% get a new hook; those above 25% with a low CTR get a new offer; everything else is left alone. At Pilotium this video funnel read is automated per ad, but the manual version with five columns works perfectly well if you actually do it.
This week: set up the columns (two minutes), calculate the hook rate of your main video, and put a number on something you've been judging by feel until now. There's a decent chance the video you thought was good doesn't break 12%, and that, as much as it stings, is good news: you just found the cheapest fix in your entire account.