Data Study
"Keep it under 15 seconds" is the most repeated piece of TikTok ad advice. We scored 76 verified Creative Center top ads — 5 seconds to 4½ minutes — on predicted brain engagement. Correlation between length and engagement: r = 0.04. Effectively zero.
| Runtime | Ads | Median engagement index | Note |
|---|---|---|---|
| Under 15s | 26 | 25.2 | lowest median of any bucket |
| 15–30s | 28 | 36.7 | the sweet spot by sample size and score |
| 30–60s | 20 | 36.1 | statistically identical to 15–30s |
| 60s+ | 2 | 23.7 | tiny sample — read nothing into it |
The one real signal: sub-15-second ads underperform everything else by a wide margin. Once an ad clears ~15 seconds, more runtime neither helps nor hurts — the 30–60s bucket matches 15–30s almost exactly.
The 20 highest-engagement ads in the set run a median of 28 seconds; the bottom 20 run 14. That's not "longer is better" — it's selection. An ad that keeps earning the next second can afford 40 seconds: the #1 ad in the benchmark is a 42-second cleaning demo that builds to its peak at 26s. An ad with one idea and no build has to stay short, and even then it bleeds: the under-15s bucket is where most of the weakest curves live. Watch both extremes — the 5-second ad scored 8/100 and the 4-minute ad scored 9/100. Length was never their problem.
What actually separates the top of this leaderboard from the bottom is the shape of the attention curve, not its width. Top ads build to a mid-video peak (median peak position: 57% of runtime) and avoid sustained crashes. Weak ads either open at their peak and decay, or crash early — of the 20 ads with a detected attention crash, the median crash lands at 7.6 seconds. If you're deciding how long your cut should be, the practical move is to score both versions and compare curves — see how to read an attention curve and the opinion-based companion piece, best TikTok ad length.
All 76 ads are verified top performers from TikTok Creative Center, scored by the fMRI-trained AdCortex™ model (methodology). Every ad, curve, and score in this study is public in the benchmark library. The model predicts attention, not conversions.
Score both before you spend — same model, same benchmark, up to 8 free credits.
Compare My CutsNot according to brain data from 76 proven top ads: runtime and predicted engagement were essentially uncorrelated (r = 0.04) across a range from 5 seconds to over 4 minutes. Ads from 15–60 seconds scored a notably higher median than sub-15-second ads, but within that band, length itself didn't predict anything. Pacing decides, not duration.
In this dataset the opposite pattern appeared: the 20 highest-engagement ads ran a median of 28 seconds, while the 20 lowest ran a median of 14 seconds. Under-15-second ads had the lowest median engagement of any duration bucket. Short isn't a strategy — it's often what's left when an ad has nothing to build toward.
Yes, if it keeps earning seconds. The 30–60 second bucket in our data matched the 15–30 second bucket almost exactly (median engagement 36.1 vs 36.7). Length only hurts when attention crashes partway — which is a pacing failure you can detect before spending by scoring the ad and reading its attention curve.