Hook Rate Test Sample Size Calculator
Two thumbnails, two opening seconds — which hook actually stops the scroll? This tells you how many impressions each creative needs before the difference in hook rate is real, not just whichever one Meta happened to favor early.
A hook test compares the hook rate — the share of impressions that become 3-second video views (the ‘thumb-stop’) — between two or more creatives. Because hook rate is a proportion, the sample size comes from the same two-proportion z-test as any A/B test, just measured in impressions instead of visitors. The smaller the difference you want to catch, the more reach each creative needs. Hook events are frequent, so these tests are usually cheap relative to conversion tests — but they still need real sample to settle.
Hook Test Sample Size Calculator inputs and result
How to use this calculator
- Enter your baseline hook rateUse the control creative’s 3-second view rate. Most healthy paid-social creative lands in the 25-35% range; pull the real figure from your platform.
- Name the difference worth catchingDecide the smallest relative gap in hook rate that would change your decision. Tiny gaps need large reach, so test creatives that are genuinely different.
- Set creatives, confidence, and powerTwo creatives is a head-to-head; each extra one needs its own sample. Many teams run hook tests at 90% for speed, then re-confirm winners at 95% before scaling.
- Add your CPM for a spend estimateEnter the blended CPM for the audience. The tool multiplies total impressions by CPM to estimate what the test will cost in media.
- Hold everything but the hook constantSame audience, placement, budget, and run window — only the opening seconds change. Then export the plan with the share, CSV, or print buttons.
RGM Expert Says
Hook rate is the first gate of paid-social performance: if the opening seconds do not stop the scroll, nothing downstream gets a chance. Teams test hooks constantly, but most do it by eye — they glance at the dashboard, see one creative ahead after 8,000 impressions, and kill the other. That is almost always too early, and the platform’s own delivery bias makes it worse, because the algorithm front-loads impressions toward whatever looks good first.
We size hook tests for the same reason we size any experiment: so the call is the creative’s, not the algorithm’s. The good news is hook tests are cheap. Hook rate is a high-frequency event — tens of percent of every impression — so the sample accumulates fast, and a clean head-to-head often costs a fraction of what a conversion test does. That makes powering them properly an easy discipline to keep.
The trap is reading a winner before each creative clears the learning phase. Early delivery is noisy and unevenly distributed, so we let the impressions reach the sized number, with each creative out of learning, before we trust the hook-rate gap. A hook that wins at 5,000 impressions and a hook that wins at the sized sample are not the same evidence, and only one of them is worth scaling.
How it works
Hook rate is a proportion, so the impressions-per-creative figure comes from the standard two-proportion z-test — the same math as a conversion A/B test, with impressions as the sample unit and hook rate as the rate.
- p₁ — baseline hook rate (3-second view rate) of the control creative.
- p₂ — target hook rate after the difference you want to detect.
- zα, zβ — critical values for confidence and power.
- CPM — cost per 1,000 impressions, used to estimate media spend.
Hook rate is a proportion, so this is the standard two-proportion z-test (the same formula as any A/B sample size). Hook-rate benchmarks (25-35% typical) are common paid-social rules of thumb, not a guaranteed range — pull your own figure. See Evan Miller on sample size and Georgiev on why early calls mislead.
Most creative ‘winners’ are called far too early
Paid-social platforms front-load delivery toward whichever creative looks strong in the first few thousand impressions, which makes early hook-rate gaps look more decisive than they are. Sizing the test in advance is how you tell a genuine thumb-stop winner from an artifact of the delivery algorithm — the same protection a sample size gives any A/B test.
Hook tests are unusually affordable, and that is the case for doing them right. Because hook rate is a high-frequency event, the impressions needed to separate two creatives accumulate quickly and cheaply. There is little excuse for eyeballing a winner when a properly powered head-to-head costs a fraction of a conversion test.
The discipline that matters most is the learning phase. Reading a hook-rate gap before each creative has exited learning — and reached its sized impression count — means reading noise. Let the sample build, then scale the hook that wins on real evidence, not the one the algorithm happened to favor on day one.
How the gap drives the reach you need
Impressions per creative rise sharply as the difference you want to detect shrinks. These figures use a 30% baseline hook rate at 95% confidence, 80% power.
| Difference to detect | Per creative | Read |
|---|---|---|
| 25% | ~1,600 | Bold creative differences are cheap |
| 10% | ~9,200 | Comfortable for most ad accounts |
| 5% | ~37,000 | Subtle hooks need real reach |
| 2% | ~230,000 | Likely not worth resolving |
What the experts say about creative testing
Let the platform exit the learning phase before you judge a creative. Early delivery is noisy and front-loaded, and reading it as a result will burn budget on the wrong winner.
The opening seconds are the whole game in social video. Test the hook deliberately, with enough reach to know the difference is real.