Hook rate vs thumb-stop rate: which one actually tells you something

6 min readArtem Makarenko — Founder & Art Director, 2a4 Studio

In most accounts both names describe the same sum: 3-second video plays divided by impressions. Arguing about which label is correct is a distraction. What actually matters is measuring two different things separately — whether the scroll stopped, and whether the viewer then stayed — because a creative can win the first and lose the second, and the blended number hides exactly that. Neither is worth anything without the cost per result underneath it.

The two names, honestly

Usage varies by team and by tool, so it is worth stating plainly rather than pretending there is a standard. Most practitioners calculate both hook rate and thumb-stop rate as 3-second video plays divided by impressions. Some reserve hook rate for a longer marker — 15 seconds, or ThruPlay — and use thumb-stop for the 3-second version.

Because the definitions drift, the number is only comparable inside one account, calculated one way, against its own history. A hook rate quoted from a case study or a benchmark deck is close to meaningless unless the formula travels with it.

This is why the naming argument is not the interesting part. Whatever you call them, an account needs one number for did they stop and a separate number for did they stay.

Stopping and staying answer different questions

Split them and each becomes diagnostic. Collapse them and you get a number that moves for reasons you cannot attribute.

What you measureRoughly howWhat a drop actually means
Did they stop3-second plays / impressionsThe first frame lost. Opening image, text overlay or the first motion beat.
Did they stayLonger plays / 3-second playsThe premise lost. They looked, understood the offer, and left.
Did they actCost per resultThe only one that pays. Healthy stop and stay with rising CPL points past the creative.
Calculate each the same way every time and compare an ad against its own first week — cross-account benchmarks for these are not comparable.

The trap: a great hook that loses the lead

The failure we see most often is an account optimising the number it can move fastest. Hook variations are cheap to produce, and they reliably lift the stop rate — a louder opening, a harder pattern interrupt, a bigger claim in the first frame.

Then cost per lead does not improve, and sometimes worsens. The reason is straightforward: a hook that overclaims buys attention from people the offer was never going to convert. You have paid for a larger, less qualified top of funnel and called it a creative win.

The tell is the shape of the numbers. Stop rate up, stay rate down, CPL flat or rising means the opening is writing a cheque the rest of the ad cannot cash. The fix is not another hook.

When the hook is genuinely the problem

The opposite pattern is real too, and it is the case where hook work pays. Stop rate falling against the ad's own first week, with stay rate holding, is fatigue in the most literal sense — the audience has seen this opening and is no longer stopping for it, but the ones who do still find the argument sound.

That is a replacement job rather than a rebuild. The premise still works, so new openings on the same concept are the cheapest available win, and this is exactly what hook variations are for.

Distinguishing the two cases takes about ten minutes with both numbers in front of you, and is impossible with one.

Where these sit against everything else

Both metrics are upstream diagnostics, not objectives. They are useful because they localise a problem to a part of the ad, which tells you what to brief next. They are dangerous the moment they become the target, because both can be moved without moving revenue.

It is also worth remembering what they cannot tell you. Neither number says anything about whether you have enough distinct concepts running to be testing at all, and that is more often the binding constraint. At a roughly 5% win rate, an account running two premises is not going to be rescued by a better opening on either of them.

So the order is: enough concepts first, then stop and stay to diagnose the ones that are live, then cost per result as the thing that actually decides.

Read stop rate and stay rate together

  1. Fix one formula and write it down

    Pick your definition of each metric, note it somewhere the whole team can see, and never change it mid-quarter. Comparability inside the account is the only comparability these numbers have.

  2. Baseline each ad against its own first week

    Not against a benchmark, not against a different ad. The question is always whether this creative is decaying relative to how it started, which is the only version of the question the data can answer cleanly.

  3. Put stop, stay and cost per result in one row

    Three columns per ad, top spenders only. Almost every diagnosis falls out of the shape of the three together, and none of them is readable alone.

  4. Match the pattern to the fix

    Stop down, stay holding: fatigue, so ship new openings on the same premise. Stop up, stay down: the hook is overclaiming, so fix the promise rather than the opening. Both healthy, cost per result rising: the problem is past the ad — offer, landing page or audience.

  5. Check the concept count before acting on any of it

    If the account is running two or three distinct premises, the metric work is optimising a sample too small to conclude from. Widen the concept pool first; the diagnostics get trustworthy once there is something to compare.

Questions we get asked

Is hook rate the same as thumb-stop rate?

In most accounts, yes — both are calculated as 3-second video plays divided by impressions. Some teams reserve hook rate for a longer watch marker such as 15 seconds or ThruPlay. Because usage is inconsistent, define it once for your own account and compare only against your own history.

What is a good hook rate?

There is no portable answer, and quoted benchmarks are usually not comparable because the formula behind them is rarely stated. The useful version of the question is whether a given ad is holding against its own first week. A drop of 20-25% from an ad's own baseline is the point most practitioners act on.

Our hook rate went up but leads got more expensive. Why?

Usually because the opening is overclaiming. A harder hook stops more people, including people the offer was never going to convert, so you buy a wider and less qualified top of funnel. Check whether stay rate fell at the same time — if it did, the fix is the promise the ad makes, not the first two seconds.

Should we optimise creative for hook rate at all?

As a diagnostic, yes. As a target, no. Both stop and stay can be moved without moving cost per result, and a team that reports on them as goals will eventually optimise into a well-watched ad that does not sell. Keep cost per result as the objective and use these to work out which part of the ad to brief next.

Sources

Keep reading