Read hook rate and hold rate

Hook rate is the share of people the ad reached who engaged with it at all — who stopped rather than scrolled past. It is a question about the opening: the first frame, the first line, the thumbnail, whatever the format's equivalent is.

Hold rate is the share of the people who stopped who stayed. It is a question about the body: whether what followed the opening was worth the attention the opening bought.

The second is conditional on the first, and that conditionality is the reason the order matters. An asset nobody stopped on has a hold rate computed over almost nobody, and reading it as a statement about the message is reading a statement about four people.

Did it get seen. Delivery first. An asset with almost no impressions has no diagnostic value at all, and the cause is usually structural — a tight budget, an audience too narrow, an entity still paused, a campaign whose money went elsewhere — rather than creative.

Did it stop anybody. Hook rate. A weak one points at the opening and at nothing else. The message may be excellent; nobody reached it.

Did it hold them. Hold rate, among those who stopped. A strong hook with a weak hold is a promise the body did not keep, which is one of the few genuinely actionable findings in creative.

Did it move anybody. Clicks, then results. A strong hold with nothing after it is almost always a problem past the click rather than in the creative.

Each step narrows the next. Skipping to the last one is how a good asset gets killed for a landing page's failure.

Weak hook, anything else. Fix the opening. Everything downstream is uninformative because too few people reached it.

Strong hook, weak hold. The opening wrote a cheque the body did not honour. This is the pattern most worth acting on, because the fix is specific and the asset is half right.

Strong hold, weak click-through. People watched and did not act. Usually the ask is missing, late, or unclear — or the audience is genuinely the wrong one and enjoyed the content without wanting the thing.

Strong through, no results. Look past the ad. The landing page, the offer, the price, the form. Replacing the creative here discards the asset that was doing its job, and it is the most common expensive mistake in this subject.

Whether the asset is good. They are rates on intermediate events, and intermediate events correlate with results imperfectly. An asset optimised for stopping people is an asset optimised for stopping people, which is not the same as an asset optimised for selling anything.

Whether one asset beats another. Two rates with no sizing behind them is a comparison of two samples. A difference that would not survive sizing the test first is a difference you should not act on, however clean the numbers look.

How you compare to anybody else. Definitions differ by platform and by format — what counts as a view, and at what point, is the platform's own choice — so the same-named rate from two platforms is not one metric. Each connection carries a capabilities object describing what that platform can express, and the honest comparison is within a platform, against your own history.

There is no table of good numbers in this guide. A threshold stated as a fact about the world needs a real source with a date, and the figures in circulation for these two rates are vendor claims rather than sourced facts. Your own previous asset is the benchmark, it is specific to your audience and your offer, and it is the only one that answers whether this asset is an improvement.

Tempio stores counts and money, one row per entity per day, and computes every ratio at the moment you look. Nothing that can be computed is stored, which is why a rate here cannot drift from the counts beneath it.

Two consequences for this subject.

A ratio whose denominator is zero comes back empty rather than as a zero, and the result says so. An asset with no impressions has no hook rate — that is undefined, not nought, and averaging it in as a zero would drag a report down for no reason.

And reach is not a count you may sum. Platforms report it de-duplicated, so daily reach added over a fortnight is the sum of daily reach rather than the reach of the fortnight. Any rate you build with reach as the denominator has to respect that.

The reading side, stated once in full: your connections read your accounts, a change is written as a draft on the record with the state it would change from and the checks it passed, applying it is gated on approval — by you, or by a rule you wrote — and no platform write reaches a customer account until that platform has granted write access. Applying a cleared change to a platform is in private beta.

The second use of these two rates is watching a winner rather than grading a challenger. An asset whose hook rate is falling while frequency rises is an asset the audience has seen enough of, and that shows up in these numbers well before it shows up in cost per result.

Ad fatigue and when to refresh is that reading, and Run a creative test that answers something is what to do once the diagnostic has told you which part to change.