Twenty-five of our URLs rank on page one of Google. In 28 days they earned one click.

What we measured, and how

One property, `sc-domain:askedabout.com`, queried through the Search Console Search Analytics API with `dimensions: ["page"]`, `dataState: "all"`, `rowLimit: 2000`, over the window 2026-07-03 to 2026-07-31 — Search Console's data lags roughly two days, so that is the most recent complete 28-day window available on 2026-08-02. Sixty-three URLs returned any impression at all. We then kept every URL whose average position was 10.0 or better, which is the conventional definition of page one, and made no other selection. No sampling, no exclusions, no smoothing. The property total over the same window was 1,103 impressions, 1 click, average position 44.0.

The 25 page-one URLs, in full

The arithmetic

To ask whether 1 click is low, you need an expectation. We used a deliberately conservative page-one CTR curve — 28% at position 1, then 15, 11, 8, 6, 4.5, 3.5, 2.8, 2.3 and 2.0% at position 10 — applied to each URL at its own rounded position, and summed. These are lower than most published curves on purpose: if the deficit survives a pessimistic expectation, it is not an artefact of a generous one.

The last column is a Poisson tail: the probability of seeing 1 or fewer clicks if the conservative curve were the truth. On the non-boilerplate cut it is 0.04%. Boilerplate is excluded in the second row because `/privacy-policy` at position 2.7 and `/terms` at position 2.3 are our two highest-ranked URLs on the whole property, and nobody clicks a privacy policy out of a result page no matter what it is called — leaving them in flatters the expectation and depresses the observed rate at the same time.

One week earlier, on the window ending 2026-07-25, the same measurement read 17 URLs, 140 impressions, 1 click, and at that size the deficit was not detectable — the curve predicted 1.5 to 2.5 clicks and we observed 1, which is noise. In seven days the page-one impressions more than doubled and the clicks did not move. The deficit became measurable through volume, not through anything getting worse.

The two explanations, and why we cannot yet separate them

• The snippets do not earn the click. Titles and meta descriptions written for a reader who already knows the site, not a stranger scanning ten results. Fixable, and testable one page at a time.
• The queries are zero-click by construction. An AI Overview, an answer box or a People-Also-Ask stack answers the question above our result, so position 7 is not the seventh thing on the screen — it is somewhere below the part of the screen anyone reads. Not fixable by titles, and far more important if true.

The honest state of the evidence on 2026-08-02: the deficit is now real, and the cause is not resolved. One asymmetry is worth recording because it will look obvious later. The zero is uniform — 19 page-one `/news` URLs, spread across positions 3.6 to 9.8 and written by different hands on different weeks, and every single one is at zero. If bad snippets were the mechanism you would expect variance, because some titles would happen to work. A flat zero across 19 independently-written pages looks more like something structural sitting between the ranking and the reader than like 19 separately bad headlines.

We are testing it the only way that settles it: rewriting one title per week, on the largest page-one URL, and reading the result. The first rewrite shipped 2026-07-27 and its read-out is not due before 2026-08-10. The second shipped with this post. If clicks appear on the rewritten pages and nowhere else, explanation one wins. If nothing moves on either, explanation two is the better story and the more expensive one.

Limits

• One property, one small site. 1,103 impressions in 28 days is not a market study. This is a first-party reading, published as one, and nothing here should be generalised to a site with real volume.
• Search Console discloses the query for a minority of impressions. Twenty-four of these 25 URLs return no query rows at all, so for most of them we cannot see what was actually typed — only that we were shown and not clicked.
• A CTR curve is an average over a market, not a prediction for a page. The 12.6 figure is an expectation under a model, not a number of clicks we were owed.
• Average position is itself an average. A URL at 7.1 was not at 7.1 every time; it was at 3 for some queries and 15 for others, and the curve is applied to the average, which is a known simplification.

The takeaway

If you have been reassured by a rank report, this is the number that report is not showing you. Ranking on page one and being read are now two different achievements, and the gap between them is measurable on your own property in about ten minutes with the API call above — no tool required, ours included. What we sell is the adjacent measurement: whether the answer engines that increasingly sit above those results name your business at all. You can check that in 60 seconds, and if you want the ranking side too, our own weekly self-audit is published with the same policy as this post — the bad numbers go up as well.