Inside one position band, combined click-through differed nearly tenfold

A click-through-by-position curve says what share of impressions a position should turn into clicks. Reach position three and expect some percentage; slip to position eight and expect a smaller one. Used in a forecast, it turns a ranking target into a traffic promise.

On this property the combined rates by position band do not follow that shape, and between two of the bands they invert.

The whole site, bucketed by position

The two emphasised rows are the ones worth arguing from. The 1 to 3 band is 99 impressions across 61 URLs and proves nothing in either direction; it is printed because leaving it out would be a choice about which data the reader gets to see. URLs are canonicalised before counting, and every share uses the sum of these page rows as its denominator rather than the property day chart.

As groups, pages averaging position 10 to 20 were clicked at a combined 0.580 percent, and pages averaging 5 to 10 at a combined 0.439 percent. The band with the worse average positions had more clicks per impression, not fewer.

Comparing pages within one position band

The two bands contain different pages, so the comparison below stays inside a single band.

The 5 to 10 band carries 95.0 percent of every impression this site earned, and all ten of its most-shown pages. Split it by that ranking, which reads impressions alone.

Same position band, same site, same window. Combined click-through differs by 9.8 times.

Position is not fully held constant here. A five-rank-wide band is not a fixed position: inside it the ten most-shown pages average 7.65 and the other 170 pages 6.43, a gap of 1.22 positions. So position is not eliminated as a contributor, only narrowed. What can be said is that the residual difference runs against the most-shown pages, which average more than a position worse. How much of the 9.8x that residual could account for is not established here: a difference in group means does not bound the effect of the underlying position distributions, and separating the two properly needs finer matching than a band, or a model fitted to those distributions.

Why the bands invert

Because all ten of the most-shown pages average positions 5 to 10. They are 84.9 percent of that band's impressions, their combined click-through rate is 0.189 percent, and they pull the band's combined rate to 0.439. The 10 to 20 band contains none of them. Why their combined rate is lower is not measured here.

This shows how page composition can affect a band's combined rate. A curve that maps position to click-through has no term for which pages a band holds, so it cannot distinguish the two groups inside this band - that is a property of the form, not a survey of what has been published.

What this means for forecasts

A forecast of the form "we currently rank eight, moving to four is worth this much traffic" multiplies impressions by a click-through rate for a position. This page does not test such a forecast: it has no curve, no predicted rate and no prediction error. What it shows is that inside one position band on this site, two groups of pages had combined click-through rates 9.8 times apart, so a single rate for that band describes neither group.

Before relying on a position curve for a site like this one, check its rates against the site's own clicks and impressions for the pages the forecast is about. The same caution runs the other way: a low click-through rate beside a good average position is not, on its own, evidence that a page is underperforming.

What we are not claiming

Not that position does not matter. Nothing here measures position within one query family, because these buckets mix families. Finding 10 stays inside one cluster and compares common and rare queries within average-position bands.

Not that any published curve is wrong, in general or for this site: none was tested here.

And page-level average position is a blend. A page reported at 7 may sit at 3 for some queries and 15 for others, so these buckets are softer than they look. That is a real limit. It is also the number the interface reports, which is why the buckets use it.

How to check your own

Export your pages with clicks, impressions and average position. Bucket by position, then sum clicks and impressions inside each bucket and divide. Do not average the per-page rates: that gives a page with nine impressions the same vote as a page with nine hundred thousand.

Then split your largest bucket by impressions alone: its most-shown pages against the rest, each group's clicks summed over its impressions summed. If the two combined rates are far apart, a single rate for that bucket describes neither group, and a forecast built on one should be checked against your own clicks and impressions before it is relied on.

Full method

Google Search Console pages export, canonicalised to 382 URLs, bucketed by the page-level average position Search Console reports. Clicks and impressions summed inside each bucket and divided, rather than averaging per-page rates, so a page with four impressions cannot weigh the same as one with four hundred thousand. The 5 to 10 band is then split into the site's ten most-shown pages, ranked by impressions alone, and all its other pages.