Blog
What is a hold-rate decay curve for video ads?
What is a hold-rate decay curve for video ads?
A hold-rate decay curve for video ads is your ThruPlay (or 15-second / nearly full completion) divided by 3-second plays, plotted over time on one asset. Meta documents the ingredients, not a 2026 curve shape. If hold falls while hook holds, the body stopped paying off the opening. Fix payoff and pacing; Do not only recut frame one.
What is a hold-rate decay curve for video ads?
A hold-rate decay curve for video ads is your ThruPlay (or 15-second / nearly full completion) divided by 3-second plays, plotted over time on one asset. Meta documents the ingredients, not a 2026 curve shape. If hold falls while hook holds, the body stopped paying off the opening. Fix payoff and pacing; Do not only recut frame one.
Plot your asset, not a blog’s curve
Export weekly hold rate on the control. Compare to launch week. Placement splits matter. There is no official “hold dies on day seven” law.
ThruPlay counts how much of the video played, not wall-clock while paused. Use Meta’s definition, then do your division.
Act on the shape you see
Hold decay with stable hook → rewrite body promise within the first seconds after the hook. Hook decay with stable hold → new openings. Both down → new structure.
Make those cuts in eonik. Read again the same way.
How to keep the curve honest
One asset per plot. Do not average ten ads into a fake industry curve. Label whether the numerator is ThruPlay or a 15-second proxy.
Ignore invented “strong hold” bands. Relative shape is the point.
Hold-curve shape vs next cut
| Shape | Meaning | Make |
|---|---|---|
| Hold down, hook steady | Payoff failed after the stop | Rewrite body pacing |
| Hook down, hold steady | Opening saturated | New hook family |
| Both down | Footprint exhausted | New structure |
What usually breaks
- Importing a blog’s day-seven hold-death law.
- Averaging placements into one fake curve.
- Only recutting frame one when hold is the leak.
When the readout is fatigue, hook rate, or velocity, the next move is consider what to make. Then make the cut in eonik and approve it. The loop is how to generate AI ads. It never touches spend.
“Almost any question can be answered, cheaply, quickly, and finally, by a test campaign. And that’s the way to answer them, not by arguments around a table.”
That principle still holds. The next on-brand cut is yours to approve.
Sources
Questions
What is a hold-rate decay curve for video ads?
Your ThruPlay or 15-second completion divided by 3-second plays, plotted over time on one asset. Meta documents those play metrics, not a 2026 curve shape. If hold falls while hook holds, the body stopped paying off the opening. Fix payoff and pacing; Do not recut only frame one; There is no official day-seven death law; Plot your control versus launch week.
What is a good hold rate in 2026?
Meta does not publish one. Hold rate equals ThruPlays divided by 3-second plays. Compare to your baseline on the same placement. Ignore invented “strong hold” tables; Label whether you used ThruPlay or a 15-second proxy so the curve stays comparable week to week; A drop versus yourself is a signal; A number versus a roundup is noise.
Is hold rate the same as video plays at 25 percent?
No. Video plays at 25 percent is a quartile metric Meta documents separately. Hold rate as operators use it usually uses ThruPlay or a 15-second proxy. Do not mix them on one plot without labeling; Quartiles can still help you see where drop-off happens inside the body; The decay curve question is hold over calendar time on one asset, not a single quartile column.
Free, no card · macOS 15+ · Apple silicon