TikTok hook tests: five hook variants or one body with fewer, bigger swaps?

Most TikTok advice says to test a pile of hooks, and I think that's where small budgets quietly fail. Say a supplement brand has one solid body edit and makes five hooks. Each hook gets a few hundred dollars, the spend splits so thin that none of them gets enough impressions, and you end up with noise.

What tends to work better is running one hook and body long enough to read the 2-second and 6-second view rates, then swapping only the hook once the body is clearly holding attention. Changing the body is a separate test. If you change both at once, you can't tell which edit moved the number.

How many variants makes sense depends on daily spend. Below a few thousand a week, I'd test fewer, bigger differences. Has anyone seen the opposite on their account?

Sasha's point about not splitting spend too thin still holds, but I don't buy that you have to prove the body is holding attention before you swap hooks. The 2-second view rate is mostly a hook metric. Whether someone stops scrolling in the first two seconds comes down to the opening frame and line, not the edit after it. So you can run two or three hooks on the same body and read the 2-second rate directly, then use the 6-second rate afterward as a body check.

At a few hundred dollars per hook, waiting for a body that's "clearly holding" can cost you weeks, and by then the body may be fatigued anyway. I'd cap it at two or three hooks so the spend doesn't get spread too thin.

Nova's right that the 2-second rate mostly reads the hook. The catch is that plenty of people scroll with sound off, so the first second has to work on text or visuals alone. If one variant opens on a voiceover line and another opens on a bold on-screen claim, you're testing two different things, not two hooks. Keep the opening frame and overlay text as the only change between variants, with the same sound or lack of it. Then a gap in the 2-second rate points at the hook and not at how the audio happened to start.

Rowan's right about holding the audio constant. I'd add that a 2-second view rate can jump just because one variant's opening frame loads faster or is a cleaner shot, so make sure your variants start on the same quality of first frame. Otherwise the hook that looks stronger may simply be the one with the sharper first image. And at a few hundred dollars per hook, a gap of only a few points in 2-second views is usually still noise, so wait for a real spread before calling a winner.

I'm not sure waiting for a real spread works at a few hundred dollars per hook. You may never get one, because the budget runs out or the creative starts fatiguing while you're still staring at a three-point gap. I'd set the rule before launch instead: a fixed impression count per variant and a minimum gap you'd act on. Then pause the clear loser and move budget to the leader. You won't be certain, but a consistent lead on a small sample is still a reasonable reason to shift spend.

Setting the impression count and minimum gap before launch is the right call, otherwise you'll stop on a lucky first day. What the thread hasn't covered is that TikTok doesn't always split delivery evenly between ads in the same ad group. It favors whichever ad it expects to perform, so one hook can quietly take most of the impressions and you end up comparing a large sample against a small one. Either put each hook in its own ad group with the same audience and daily budget, or check impressions per ad before you read any rates. And compare 2-second views per impression, not raw counts, because a variant with a cheaper CPM will show more views for reasons that have nothing to do with the hook.

Per-impression rates fix the volume problem, but they don't fix the audience problem. When two ads share an ad group, the auction can send each one to a different slice of people, and a hook that wins a cheap, younger audience will look better than it really is, even after you normalize. So the separate ad group setup matters more than the math. If you can't split them, check the age and gender breakdown before you trust a gap of a few points.

Say a supplement brand runs hooks A and B in one ad group. B's 2-second rate is four points higher, but its impressions skew toward women 18 to 24 at a lower CPM. Rerun each hook in its own ad group with the same audience and budget, and that gap may shrink or flip. If it holds, you've got a real hook winner. If it doesn't, you were reading audience mix.