TikTok hook tests: kill ads on 3-second views or on 6-second holds?

Most people use 3-second view rate as the first filter on TikTok openers, and it makes sense: if the first frame doesn't stop anyone, the hook is weak. But it's a cheap metric. Someone who scrolls past a bright opener still counts as a view, so an ad can look healthy at 3 seconds and then die at the offer.

I'd use 3-second views to judge the hook, and 6-second views, average watch time and CTR to judge the concept. Don't kill an ad until it has enough impressions for the gap to be real. Say a supplement brand tests two openers. A fast face close-up wins on 3-second views, but a slower product demo holds people longer and gets more clicks. In that case the demo is the better ad, even though it loses the first filter.

So which number do you use as the kill switch, and how much spend do you need before you trust it?

CTR is a shaky tiebreaker for a store. Demo clicks tend to be curiosity clicks that bounce or never reach add-to-cart, so the demo can win on CTR and still lose on cost per purchase. Kill on whatever metric sits closest to revenue that your budget can actually reach, which is usually purchase or add-to-cart.

That holds for a store with volume, but B2B is different. A LinkedIn campaign might only get a handful of conversions a month, so a purchase-based kill switch never reaches significance. For that kind of account I'd kill on cost per qualified lead, then check it against pipeline once sales has worked the leads.

Agreed on CPL, but with only a handful of leads per ad, a 40% gap is basically indistinguishable from luck. I'd split the job. Let 3-second and 6-second views decide which hook survives, since those pile up thousands of impressions, and only use leads to confirm the concept.