ToolsThatRank / measured data
First party census
Search Console names the search behind a third of your impressions
Filter the Performance report to one of your pages and you expect the list of searches that brought it traffic. On the typical page in our census, Google names the search behind 29.0% of its impressions and withholds the rest. Not because anything is broken. Rare queries are dropped on purpose.
Verified 2026-08-24 against the Search Console searchAnalytics endpoint. Window 2026-05-24 to 2026-08-21, 34 properties.
Apply it to your account
How much of your own report carries a query
Type how many pages Google shows for your site and their total impressions. You'll get our measured rates applied to your numbers, so you can see the rough scale of what is missing from a query level view.
Our rates, your numbers. This does not read your account and cannot know your query mix, so treat it as a scale check rather than a measurement of your site.
The measurement
Two gaps, and only one of them matters
Start with the one that sounds dramatic and mostly is not. We pulled the page dimension for every property, then the query and page dimensions together, and asked which pages ever appear with a query beside them. 3,608 of 4,974 never did, which is 72.54% of pages.
That number is far less useful than it looks. Those 3,608 pages hold only 2.96% of all impressions and 2.2% of all clicks. They are a long tail of URLs Google barely showed, not a blind spot over traffic that matters.
The gap worth knowing is the second one. On the 1,366 pages that do get queries named, those query rows sum to 654,152 impressions against 1,824,935 at page level. So even on a page that looks fully reported, roughly 64.15% of its impressions carry no query you can read.
The number that actually helps
Whether missing queries are normal depends on your impressions
Split the same pages by how many impressions each one got and the picture becomes usable.
| Page impressions | Pages | No query named | Share | 95% interval |
|---|---|---|---|---|
| 1 | 679 | 653 | 96.17% | 94.45 to 97.37 |
| 2 to 9 | 2,030 | 1,825 | 89.9% | 88.51 to 91.14 |
| 10 to 49 | 1,270 | 907 | 71.42% | 68.87 to 73.83 |
| 50 to 199 | 466 | 184 | 39.48% | 35.15 to 43.99 |
| 200 or more | 529 | 39 | 7.37% | 5.44 to 9.92 |
Read the bottom row first. Past about 200 impressions, missing queries stop being normal. Only 7.37% of those pages had none, so an empty query list on a page at that volume is genuinely unusual and worth a look.
Read the top row second. A page with a single impression had no query named 96.17% of the time. Nothing to diagnose. The threshold has not been cleared, and no amount of investigation changes that.
Highlighted rows are the bands where attribution is mostly present. The intervals do not overlap between neighbouring bands.
A trap we nearly fell into
Unattributed pages are not worse pages
Pages with no named query earn clicks far less often than pages with one. Raw, it looks like a gulf. But unattributed pages are overwhelmingly small pages, so that comparison is measuring size, not attribution. Match the two groups on impressions and most of the gap disappears.
| Page impressions | Query named | No query named | ||
|---|---|---|---|---|
| Earned a click | Pages | Earned a click | Pages | |
| 1 to 1 | 0.0% | 26 | 0.92% | 653 |
| 2 to 9 | 1.46% | 205 | 0.82% | 1,825 |
| 10 to 49 | 5.51% | 363 | 3.42% | 907 |
| 50 to 199 | 19.86% | 282 | 15.76% | 184 |
| 200 or more | 57.14% | 490 | 30.77% | 39 |
In the 50 to 199 band the two groups sit at 19.86% and 15.76%. A page having no named query tells you Google withheld the search terms. It does not tell you the page is failing.
Every property
It happens on every site we measured
Across the 23 properties with at least 20 pages, the share of pages with no named query ran from 19.44% to 93.79%. Removing the largest property moves the fleet figure from 72.54% to 68.69% and the named impression share from 35.85% to 39.18%. Not one property escaped the effect, which is what makes it a property of the report rather than a quirk of one site.
| Property | Pages with impressions | Pages with a query named | No query named |
|---|---|---|---|
| bestchromeextensions.com | 885 | 55 | 93.79% |
| belikenative.com | 2,167 | 487 | 77.53% |
| lockml.com | 26 | 6 | 76.92% |
| zovo.one | 895 | 256 | 71.4% |
| ml0x.com | 45 | 15 | 66.67% |
| claudkit.com | 30 | 11 | 63.33% |
| recipetool.net | 34 | 13 | 61.76% |
| claudhq.com | 101 | 39 | 61.39% |
| epochpilot.com | 57 | 25 | 56.14% |
| claudflow.com | 33 | 15 | 54.55% |
| gen8x.com | 24 | 11 | 54.17% |
| kappafy.com | 23 | 11 | 52.17% |
| invokebot.com | 28 | 15 | 46.43% |
| kickllm.com | 59 | 33 | 44.07% |
| abwex.com | 39 | 22 | 43.59% |
| heytensor.com | 102 | 58 | 43.14% |
| gpt0x.com | 31 | 18 | 41.94% |
| enhio.com | 34 | 20 | 41.18% |
| ml3x.com | 31 | 19 | 38.71% |
| lochbot.com | 64 | 43 | 32.81% |
| lakelevelnow.com | 77 | 54 | 29.87% |
| b2berp.com | 65 | 52 | 20% |
| ingredientcalculator.com | 36 | 29 | 19.44% |
Highlighted rows are above 70%. Properties with fewer than 20 pages are left out because a handful of pages moves the share too far to read.
Why it matters
What this breaks in everyday reporting
Summing query rows to get a page total undercuts it. Across our properties that sum reaches 35.85% of the page level figure, so a report built on query rows alone is reading about a third of the traffic and presenting it as the whole.
It also explains a pattern that wastes a lot of time. A page shows impressions and an empty query list, and the natural conclusion is that something is broken. Below 10 impressions that is simply the normal case.
The rule we follow. Use the page dimension for totals, the query dimension for direction, and never present a query level sum as a complete account of anything.
Method
How we measured this
We measured this on 2026-08-24, over 2026-05-24 to 2026-08-21, against the Google Search Console Search Analytics API on 34 verified properties the operator owns. Two pulls per property, one with the page dimension and one with the query and page dimensions together, both paged through to completion.
Checks that could have broken this
No property came within five rows of a pagination cap, so nothing was truncated. Matching URLs loosely on scheme, www, trailing slash and fragment rescues 32 pages, 0.89%, so the result is not a string matching artefact.
Why the median and not the mean
Withheld impressions are wildly concentrated. One page alone carries 24.4% of the fleet total and the top ten carry 45.5%. The per page median of 29.0% describes a typical page. The fleet figure describes our portfolio mix.
What counts as a named query
The URL appears at least once in the query and page pull. One named query is enough, which is the most generous reading available and still leaves most pages unnamed.
Why two pulls
The gap only appears when the two dimensions are compared. Either pull on its own looks complete.
Limits
What this cannot tell you
Your share will differ from ours. A site living on a few high volume searches loses far less to the threshold than one built on rare long tail phrasing, and our own range of 19.44% to 93.79% shows how wide that spread gets.
We measured what the API returns, not why. Google's stated reason is privacy protection for rare queries. We take that at face value because this data cannot test it.
These are 34 small sites in mixed niches run by one operator, and one property supplies most of the impressions. Every fleet figure here is printed with that property removed as well, so you can see how much depends on it.
No ranking, traffic, or revenue outcome is promised.
The operator sells the software described here and owns every site in the dataset.
Search Console is authenticated, so no third party can verify these readings from outside.
Figures are read on the stamped date and change as Search Console reprocesses data.
Related reading. Zero clicks is usually an impressions problem and what pages below position 30 actually earn, both measured on the same properties.