CTR

What it means

CTR is clicks divided by impressions: of the people shown an ad or a search result, how many actually clicked. It is quoted as a percentage and measures one thing - whether the offer looked relevant enough to that person.

In paid search CTR feeds the quality score and therefore the cost per click. In organic search it tells you how the page title and description are performing in the results.

CTR says nothing about cost. That depends on the pricing model: paying per click, you pay only for those who clicked, so a low CTR costs you nothing extra directly. Paying per thousand impressions it is the reverse: you pay for the showings, and every extra point of CTR makes each click cheaper. The same CTR can mean a cheap campaign or an expensive one.

How it works in practice

CTR is never read on its own. High CTR with a low conversion rate means the ad promises something the page does not deliver - expensive traffic with no business at the end of it.

It improves through testing titles and descriptions, tighter targeting, and cutting keywords the company does not actually sell on. The same logic applies to retargeting, where CTR decays as the audience tires of the ad.

For the website of an accounting firm or a sole trader, organic CTR can be tracked page by page in the search engine’s own tools. If a page gets many impressions and few clicks, the quickest fix is rewriting the title and description to match the question people are actually searching with. Structured data can enrich the result further.

What to watch out for

Comparing CTR across channels is pointless - search, display and e-mail have entirely different bases. What is worth comparing is one channel over time, and always alongside the cost per acquired customer.

Another mistake is drawing conclusions from too few impressions. At a hundred impressions the gap between 2 and 4 % is two clicks, which may be pure chance. Rewriting an ad or pausing a keyword on that basis is deciding on noise. Wait for enough data before changing anything, and compare variants in an A/B test under the same conditions.