Guide
Ask five marketers and you'll get five rules: 15 seconds, 21–34, under a minute. They're all answering the wrong question. The right length isn't a rule — it's a measurement.
TikTok's own guidance favors 21–34 seconds for In-Feed ads, and there's real logic behind it: feeds are impatient, completion rate feeds the algorithm, and most ads simply don't have 60 seconds of material. Short formats force discipline.
But the averages hide the mechanism. Short ads win on average because the average ad has dead segments — and cutting an ad short happens to amputate them. The variable that actually drives performance isn't duration. It's whether attention survives to the end.
Your ad has a natural length: the point where its predicted engagement curve sags and doesn't recover. Everything after that point costs you completion rate, algorithmic favor, and CTA visibility while delivering nothing. Everything before it is earning its runtime.
That point is different for every creative. A demo with three strong reveals can carry 50 seconds; a single-joke UGC clip exhausts itself in 12. Applying a universal length rule to both wastes the first ad's material and pads the second's.
Upload your full cut to PreTestAds and read the curve: Attention Drop marks the second predicted engagement falls meaningfully below peak — your candidate cut point. Purchase Signal tells you whether viewers are still engaged during your CTA window; if not, the CTA needs to move earlier, or the section before it needs to go. Trim, rescore, compare. Two or three iterations usually finds the length where the curve stays strong start to finish.
Start with a strong opening (see how to write a TikTok hook), and validate the whole workflow in TikTok ad testing. Length is one packaging choice; the best aspect ratio for video ads is the other.
Upload your cut and see exactly where attention drops — first analysis free.
See Your Attention CurveTikTok recommends 21–34 seconds for most In-Feed ads, but the honest answer is: as long as your ad holds attention and no longer. A 45-second ad that holds viewers beats a 20-second ad that loses them at second 5. Measure where your specific creative loses attention and cut there.
Often, but not because short is inherently better — because most ads have dead segments that shorter cuts naturally remove. If your engagement curve stays strong, longer ads deliver more message and stronger purchase intent. Length is an output of the attention data, not an input rule.
Score the full-length cut with an engagement-prediction tool. The attention curve shows the exact second predicted engagement falls — cut shortly after your last strong moment, keep your CTA inside the high-attention window, and rescore to confirm.