CTR

CTR stands for Click-Through Rate, also known as click rate. It answers a simple question: How often is something clicked on after being shown to someone? To calculate it, you divide the number of clicks by the number of impressions. If an ad banner is shown 1000 times and clicked 20 times, the CTR is 2 percent. This metric is over 25 years old and originates from the early days of internet advertising. Today it is far more than a measure of success: entire AI systems are trained to predict it.

Why the click rate matters to the money

With traditional television advertising, nobody knows exactly who was watching. On the internet, it’s different. Every impression and every click can be counted. This turned CTR into the hard currency of an entire industry. Ad placements are priced according to it, and campaigns are evaluated by it.

For search engines and social networks, the math is even more direct. Google and Meta earn most of their money from ads. An ad that nobody clicks on generates no revenue. Even an improvement of just a few tenths of a percent in average CTR can mean billions of dollars for these corporations. That’s why enormous computing power is devoted there to the question of which ad is shown to which user.

Typical values, incidentally, are lower than many people assume. A normal display banner often achieves only 0.05 to 0.5 percent. For search ads that match a query just typed in, several percent are possible. Such figures can therefore only be meaningfully compared within the same environment.

How models predict clicks

When you open a webpage with an ad slot, an auction takes place in the background. It lasts a few milliseconds. Multiple advertisers bid for the slot. But the winner isn’t simply the highest bid—usually it’s the bid with the highest expected return. For this, the platform needs an estimate of how likely you are to click.

This estimate is provided by a CTR prediction model. It has learned from huge amounts of past impressions which combinations of user, context, and ad lead to clicks. Features used include things like time of day, device, past behavior, and the content of the ad. The result is a probability, say 1.8 percent. Multiplied by the bid, this yields the expected return.

A common misconception: a high CTR doesn’t automatically mean success. Sensationalist headlines generate many clicks and many disappointed users. Experts call this clickbait. That’s why platforms combine click rate with other signals, such as time spent after the click or an actual purchase. Anyone who optimizes only for CTR ends up with a system that’s good at baiting.

From the ad to the news feed

CTR is most visible in advertising accounts. Anyone running a campaign on Google Ads or Instagram sees impressions, clicks, and click rate displayed side by side. Website operators also know this figure from Google Search Console, where it shows how often a search result was shown and how often it was clicked.

Less visible, but more influential, is its role in recommendation systems. TikTok’s feed, YouTube’s homepage, and Amazon's product suggestions are based on related predictions. The order in which content appears to you is the result of such estimates. CTR is thus one of the metrics that quietly helps determine what millions of people see every day.

The term comes up in business news when corporations explain their quarterly figures. Sentences like “increased click rates despite falling prices per ad” are common there. The debate over data privacy is also tied to this: less data about individual users means less accurate predictions and usually a lower CTR.