Reusing TikTok videos on Meta Reels: same cut or re-edit for each platform?

You've got a TikTok creative with a strong hook and want it on Reels too. One camp says upload the same file, since the audiences overlap and reshooting burns budget. The other says the cut was built around TikTok's interface, and Meta's overlays can cover the parts that matter.

I think the edit usually survives but the framing often doesn't. Say a skincare brand opens with text in the top third. On Reels, that text can end up under the profile name and action buttons, so the hook dies before anyone reads it. Keep the cut, move the key text lower, and check the safe zones before you export. That's a re-export, not a reshoot.

Has anyone seen a real gap in hold rate between the two placements, or is it mostly the same numbers?

Sasha's safe-zone point holds, but I think hold rate usually depends more on the first second than on the overlay. Reels and TikTok both autoplay full screen, so the hook has to land before the viewer's thumb moves, and that's often the same on both. If you see a hold rate gap, look at whether the cut opens on a talking head. TikTok's feed may tolerate that, while Reels viewers watching with the sound off won't.

Agreed on the first second. Before you blame the opening, though, check whether the gap is in the numbers. With automatic placements, Meta shifts delivery between Feed, Stories and Reels, so the audience you're comparing changes too. Break the same ad down by placement in Ads Manager and look at 3-second plays divided by impressions, plus watch-through rates, for Reels alone.

I also wouldn't compare those to TikTok's figures. TikTok counts a 2-second view and Meta counts 3-second plays, so the same hold rate can look better on one platform just because of how it's measured. For a clean answer, run the original cut and the re-edit as two separate ads against the same audience and read the placement split from that. Otherwise you're mostly comparing audience mix and counting rules.

Agreed on the split. For a store I'd go a step further and judge placements on purchases or add-to-cart, not plays. A re-edit can hold attention on Reels and still lose money if the people watching don't buy. And if both ads run on automatic placements, Meta can shift the mix between them, so pin the test to Reels or use manual placements to keep the comparison honest.

Judging placements on purchases alone, at least early on, is shaky. Unless the store has real volume, Reels will probably only show a handful of purchases in a week, and two orders either way is just noise. Hold rate and 3-second plays give you a usable signal much sooner. Then let purchases decide once each cut has enough conversions to actually separate them. Pinning to Reels is fine, but on a small budget you may need a longer window before the purchase numbers mean anything.

Agreed, two orders a week is noise. When you do compare cuts on purchases, set the reporting to 7-day click only. With view-through included, a Reels ad that just autoplayed can take credit for orders it didn't drive, and on a small test that's enough to make the wrong cut look like the winner. Also check the learning status of each ad set. Meta's guidance is roughly 50 optimization events in a week to exit learning, so an ad set still in learning isn't really comparable to the other one yet.

Agreed on 7-day click and learning status, but the setup matters too. If the two cuts are ads in one ad set, Meta decides how to split delivery and tends to favor whichever looks cheaper early on, so you end up comparing unequal spend. If they're in separate ad sets, each learns on its own and audience overlap comes back. Meta's A/B test tool in Ads Manager avoids both problems. It randomly splits the audience so each person sees only one cut, and you read results on purchases with the same attribution window. The catch is budget. Each arm still needs enough events to get past learning, so on a small budget you'll probably have to run the test longer, not shorter.

Kai's point about A/B tests needing enough events is the part people skip. A decent rule of thumb is that if each arm can't hit roughly 50 optimization events in the test window, treat the result as a hint, not a verdict. A cut that wins on hold rate but only has a dozen purchases per arm hasn't really won yet.