Hook rate calculator

Three-second views is the shortest view count your report offers. Two seconds, three or six all work; what matters is that the same definition is used every time, because a rate built on one threshold cannot be compared to a rate built on another.

Impressions is the number of times the video was served, from the same report, the same dates and the same placements as the views above them. Mixing sources is the most common way this number comes out wrong, and the third output exists partly to catch it.

Ad spend is optional and turns the rate into the price of a stopped scroll.

Video performance is usually discussed as one thing and is at least two. A creative can fail because nobody starts it, or because everybody abandons it, and the two failures have nothing in common except the report they appear in.

Hook rate measures the first. It answers whether the opening frame, the first motion and the first silent second did their job on a surface where the default behaviour is to keep scrolling. Everything else about the ad — the offer, the demonstration, the call to action — is only reachable by people this number counted.

That is what makes it the first thing to look at and the first thing to fix. Improving the back half of a video with a hook rate of five per cent improves an experience that nineteen impressions in twenty never reach.

It is also the fastest signal a creative test produces. A hook rate stabilises after a few thousand impressions, long before a cost per result has enough conversions behind it to mean anything, which makes it the metric that can kill a bad opening in a day rather than a fortnight. The usual mistake is to let a creative run to statistical comfort on the outcome when the first second had already answered the question.

Cost per hooked view converts the rate into the currency the media is bought in, and it is the figure to track between periods rather than the rate itself.

The relationship is the same one that governs every rate on this site: the price of a stopped scroll is the price of the impression divided by the share of impressions that stopped anybody. So a cost per hook that rose has two possible causes — inventory that got more expensive, or an opening that got less effective — and only comparing it against the cost per thousand impressions separates them. A creative team handed a rising cost per hook with no auction context will rewrite a perfectly good opening in a week when the market was simply more expensive.

The impressions that produced no view at all is the same fact as the rate, stated as a count, and it does two useful things.

It makes the scale legible. Twelve per cent sounds like a number to nudge; 704,000 impressions that stopped nobody sounds like what it is, which is most of the media budget.

It also catches a common mistake. If the views exceed the impressions, the two figures came from different reports, different date ranges or different placements, and the rate above them is meaningless. The count goes negative and the page says so rather than printing a confident percentage over mismatched data.

Whether the video is any good. A hook rate is a measure of the first second. A creative can stop everybody and hold nobody, and the pair of rates is the only way to see it.

Whether the audience is right. A poor rate across every creative in a campaign is usually the audience, not the openings. The rate cannot distinguish the two on its own; running one known creative against two audiences can.

Anything comparable across placements. Autoplaying full-screen surfaces produce high rates for everybody, muted small-frame feeds produce low ones. A blended hook rate over mixed placements describes none of them.

Anything about intent. A stopped scroll is attention, not interest. Accumulated three-second viewers are a weak retargeting audience precisely because many of them never chose to watch.

Test openings, not ads. The discipline that makes this number useful is to hold everything constant except the first second: the same product, the same offer, the same body, three different first frames. Anything else confounds the test.

Look at what the frame shows in the first instant. Motion that starts immediately, a face, a product in use, text large enough to read at a glance, a visible change between frame one and frame two — all of these beat a title card, a logo or a slow fade, which are the three most common ways to open an ad and the three most reliable ways to lose one.

Assume no sound. The opening has to make sense muted, because on most feeds it will be.

Check the framing. A wide edit letterboxed into a vertical placement spends the first second showing bars, and the bars are the part that was supposed to stop somebody.

Then read the hold rate beside the result. A new opening that lifted the hook rate and dropped the hold rate has recruited people the rest of the video does not want, and the pair of numbers will say so long before the cost per result does.