Click-Through Rate (CTR)

Click-through rate (CTR) is clicks divided by impressions for the same set of results over the same period, expressed as a percentage. A result with 100 clicks on 1,000 impressions has a CTR of 10%.

This library argues about CTR on dozens of pages and has never defined it, which is a problem when the argument turns on what the denominator counts. The metric is simple; the ways it misleads are not.

How do you calculate click-through rate?

Divide clicks by impressions, then multiply by 100. The formula is the easy half:

Those three rows are arithmetic illustrations, not measurements of anything. They exist to make one point: rows 1 and 3 report the same CTR from a 5x difference in clicks, and rows 1 and 2 report the same clicks at a 5x difference in rate. A ratio hides both of its inputs, so a CTR quoted without its click and impression counts is a number you cannot act on.

The practical consequence shows up constantly in AI-era reporting. If impressions fall because fewer results are shown, CTR can climb while the traffic that pays for the site drops. Read both sides, always.

What counts as an impression?

Whatever the surface reporting it says counts — and the definitions are not the same. An impression can mean a result rendered on a page a user opened, a result present anywhere in a scrolled list, or a result that entered the viewport. Analytics platforms, ad platforms, and search consoles each publish their own rule, and a CTR computed under one rule is not comparable to a CTR computed under another.

That is not a technicality. Most CTR disagreements in this niche are denominator disagreements wearing the costume of a factual dispute, so the first question about any CTR figure is not "is it right" but "impressions of what, counted how, over which dates."

How is CTR different from conversion rate?

They measure different steps, and the denominators are what separate them. CTR divides clicks by impressions — how often people who saw a result chose it. Conversion rate divides completed actions by visits or sessions — how often people who arrived did the thing the page exists for. One measures selection, the other measures persuasion.

They also move independently, which is why reporting them as a pair is more informative than reporting either alone. A title rewrite that overstates what the page contains can lift CTR and depress conversion rate in the same week, and a program judged on CTR alone will call that an improvement.

How much does an AI answer change CTR?

Two closed-list measurements sit on either side of the question, and each one belongs to its own period and sample:

Read together, they describe a shrinking pie with an internal advantage: presence of an AI answer pushes CTR down on affected queries, and being cited inside that answer buys a relative advantage over the sites that are not. Neither figure promises anything about a specific site, and nobody can promise you either outcome.

The fuller datasets, including the user-panel evidence and the 2026 rebound reporting, belong to the pages that own them: zero-click search, with the numbers reconciled and what an AI citation is actually worth.

Why do two studies of the same period report different CTR effects?

Because they measure different populations with different denominators. A user panel counts what people did; a query corpus counts what happened to top-ranked URLs; managed client accounts count what a specific set of sites experienced. Those are three questions, and identical direction with different magnitudes is exactly what you would expect from them.

The failure mode is quoting the scariest number with no date, sample, or definition attached. Where the sources genuinely diverge, the reconciliation is the useful content — which is the whole argument of is SEO dead, read against the click data.

What is CTR bad at?

Attributing cause, and measuring value. CTR moves when position moves, when the query mix shifts seasonally, when a competitor changes a title, when the results page adds a feature — none of which you did. A change in CTR is a prompt to investigate, never a finding on its own.

Said against our own interest as people who sell measurement work: CTR is also the wrong scoreboard for AI visibility specifically. A citation that a reader consumes without clicking leaves no mark on it at all, so a program judged only by CTR will systematically undercount whatever it is doing on answer surfaces — and will occasionally reward a page for a change nobody made. Use it as one input among several, alongside the measurement approaches in our generative engine optimization guide.