What CTR is, and why there is no normal CTR

There is no normal CTR. In the Yandex Direct glossary, CTR is the number of clicks divided by the number of impressions, as a percent. Google’s own definition gives the arithmetic: 5 clicks and 100 impressions make 5%, and the next sentence says a good CTR is relative to what you advertise and to the network the ad runs on (Google Ads Help).

A line that says “campaign CTR 3%” therefore says nothing until you know where the impressions were. Below is how to calculate yours, why a blend of search and the ad network lies, and when a rise in click-through with no leads is worth checking rather than celebrating.

CTR is clicks divided by impressions

Five clicks on a hundred impressions is 5%. That is Google’s example of the arithmetic, not a target for Yandex Direct. Your own percent is the same division: clicks from a report divided by impressions from the same report and the same period. The calculator below does only that division.

The Direct glossary adds that CTR can be said to show how well the ad works. That reading stops where the click is not from a buyer. An ad can be clicked often and still produce no leads. Click-through rises, and the benefit does not. That case is covered in clicks without leads.

One campaign number mixes different places

Suppose a week brought 800 clicks and 10,000 impressions on search, and 240 clicks and 80,000 impressions on the ad network. The numbers are round. This is a teaching division, not an industry average.

Together that is 1,040 / 90,000, about 1.16%. The figure describes neither search nor the network. A campaign that keeps both placements on one row shows a third number that never actually occurred.

Position on the page changes click-through before the quality of the ad does. Direct publishes traffic volume: a value directly proportional to how clickable the placement is. The glossary’s example for a premium block of four places gives the first place a volume of 100, then 85, 75 and 65. Volume can also exceed 100. These are not CTR percents. Treating them as percents means mistaking the weight of a position for a share of clicks.

Weighted CTR in the same glossary adjusts click-through for the traffic volume of the position. The formula on the glossary page is an image and is not copied here. The point is enough: you cannot compare raw CTR of ads that stood in different positions, even inside search.

Direct also asks you not to compare clicks of a media campaign with clicks of a performance campaign. Performance meets demand that already exists. Media advertising warms a colder audience, and a click there is a different event (how to read media statistics).

When a high CTR is a reason to check the traffic

A rise in click-through while leads stand still is a fork. People may have clicked an honest ad more often and not bought. Or the clicks were empty and stayed in spend. The second case has signs, and they are not the percent itself. Signs of click fraud start by checking whether the filter already removed those clicks.

Direct does not look at a single percent when it prices a click. Its bidding notes say predicted CTR enters the forecast price of a click, while the quality coefficient separately takes into account what the person does after the click. A clickable ad that leads to an empty site feeds the first input and hurts the second. It is early to celebrate a CTR rise until the leads of the same period are open.

What to compare, if there is no norm: your search with your search, your network placements with your network placements, the same campaign type, a neighboring period. Someone else’s “normal CTR” from an article with no sample and no split between search and the network does not belong in that comparison.

Visit protection does not assign a CTR and does not promise a “normal” click-through. It looks at the visit to the site: behavior, the browser environment, the network and the ad-click parameters. After it, CTR is calculated again, from the same two numbers, on the impressions and clicks that remained. There is nothing to wait for that would set the percent to some correct value by itself.