Meta video ads: judge a creative on hook rate or hold rate?

Meta gives you a few ways to judge a video ad early: 3-second plays over impressions (hook rate), thumb-stop ratio, and hold rate at 15 seconds. People argue about which one should decide whether a creative survives its first week. I think hook rate is good for ranking openings, but it tells you very little about whether the offer lands. Say a dental clinic gets a 40% hook rate on a before-and-after reel, but CTR is flat and nobody books. The hook worked and the body didn't. So I'd judge the creative on hold rate and CTR together and treat hook rate as a diagnostic, not a verdict. How do others split it, especially for lead gen, where the click is far from the real goal?

I'd keep CTR out of the pair for lead gen. It mostly measures curiosity, and a before-and-after reel can pull clicks from people who will never book. Hold rate has a similar problem, since a long, entertaining video holds the wrong audience too. My rule: use hold rate as a gate to decide whether the body is worth paying for, then judge the creative on cost per booked appointment or qualified lead once it has enough conversions. Until then, don't kill a creative on one week's cost per lead. It depends on volume, though. A small clinic might never reach that sample size, and then you're stuck judging on proxies.

Agreed on the gate. For small clinics I'd set the kill rule in money before launch instead of judging on the fly. Pick a spend limit, say two times your target cost per booking, and only cut a creative if it hits that with zero bookings. Below that, the numbers are mostly noise. If the clinic can't reach that volume at all, use call taps or calendar opens as a proxy, but check after a few weeks that they actually turn into bookings at a steady rate. A proxy nobody verified is just another guess.

Nova's kill rule is solid, but I'd set the spend limit by how many bookings you'd need before you trust the number, not by a fixed multiple of target CPA. For a clinic booking maybe 15 patients a month, 2x target CPA is often only a handful of leads, so a creative can look dead when it's really just unlucky.

Agreed, and the math is quick. If a creative really performs at your target CPA, expected bookings after spend S is just S divided by that CPA. At 3 expected bookings, you'd see zero only about 5% of the time (e^-3), so killing at 3x target CPA with nothing booked is a much safer line. At 1 expected booking, zero shows up 37% of the time, which is why a 2x rule kills so many good creatives.

Just count bookings in the same window you used to set the target, because a clinic's consult can land a week after the click.