How many pages on a small site ever get a click

01The average hides an order of magnitude

The headline is the least interesting number here. Every site in this census was built by the same person, in the same way, on the same stack, and the click yield still runs from 0 to 50 percent. Sameness of method did not produce sameness of outcome. That spread, not the 9.04 percent average, is what the data is actually saying.

Shown means Google both indexed the page and put it in front of someone at least once in the window. A page that is indexed but never shown is invisible to this instrument, so this denominator is smaller and stricter than a sitemap count. It also separates two failures that a sitemap ratio blurs together. One is the page nobody was offered. The other is the page nobody chose.

02Place your own site in the distribution

Two numbers off your own Search Console are enough. Open the performance report, set the date range you want, and switch to the Pages tab. Google aggregates that tab by page rather than by property, which is why the per page counts live there. Read how many pages appear at all, then filter to pages with at least one click and read that count.

This tool runs entirely in your browser on the numbers you type. It reads no Search Console account, sends nothing anywhere, stores nothing, and is not a traffic forecast. It compares your ratio against the 34 measurable domains on this page and nothing else.

03The distribution across the measured domains

Every row is one domain, ordered from the highest click yield down. Yield is clicking pages divided by pages shown. Domain names are withheld because these are one operator's private properties. Withholding them changes no figure, because nothing on this page depends on which domain a row is.

5 of the 34 measurable domains were shown by Google and never clicked once, which is the flat run at the bottom of the chart. A further one of the 35 domains was never shown at all, so it has no yield to report and is marked not measurable rather than counted as zero. Every rate on this page is therefore taken over the 34 domains that were shown. Three other domains report more shown pages than they currently have URLs in their sitemap, because sitemap counts are read on the capture date while the click figures cover the previous ninety days. That is why no ratio here uses a sitemap count as its denominator.

04Three published figures over three populations

These rows are not interchangeable and the percentages cannot be subtracted from one another. Each counts a different population with a different instrument, so no row is a check on any other.

These three rows measure different populations with different instruments and are not interchangeable. Ahrefs counts pages a crawler found; this census counts pages Google actually showed to someone. A page that is indexed but never shown is inside the Ahrefs denominator and outside this one, which is the main reason the two any-traffic figures are close but not comparable.

05Methodology

Our method was to enumerate every Search Console property the operator's service account can read, confirm owner permission on each, then query the Search Analytics API at the page dimension over one fixed window. Two counts were taken per domain. Pages with at least one impression, and pages with at least one click. Where a domain property and a URL prefix property described the same host, the domain property was kept, because it is the superset.

The properties belong to the author, which is what makes this a publisher side measurement rather than a modelled one. Ownership is not asserted here. It is established by the API itself, since the collecting account holds owner permission on every property in the file.

No figure on this page is typed into the copy. Every number above is computed from the published snapshot when the page is built, so if the file changes the page changes with it. Where the page and the file disagree, the file is correct.

surfacedPages            rows returned at the page dimension, meaning at least one impression
clickingPages            those rows with clicks greater than zero
clickYieldPctOfSurfaced  clickingPages / surfacedPages x 100

# a domain with zero shown pages has no yield, and is held rather than called zero

07Reproduce this page from the raw file

The dataset is published unmodified beside this page. Every figure above can be recomputed from the perDomain array with the standard library alone. Sum the shown and clicking columns for the portfolio totals, then divide one by the other for the yield. For the median, sort the per domain yields and take the mean of the two middle values, because the number of measurable domains is even. Taking a single middle value from an even list overstates it, which is a mistake this page made before it was corrected on 2026-08-15.