Search Console
Google's numbers, turned into a list of decisions
Connect Search Console for six analyses over your own data, including whether the last thing you fixed actually worked.
- analyses over your own data
- 6analyses over your own data
- extra API quota - it is all stored
- 0extra API quota - it is all stored
- against the 28 before it
- 28dagainst the 28 before it
- measured on your property
- Your curvemeasured on your property
The six
Named here exactly as the product names them.
Striking distance
Queries at positions 5-20 with real impressions
Ranked by the clicks one step up the page is worth, not by how many impressions they have - a query at position 19 with 5,000 impressions and one at position 6 are not the same opportunity, and sorting by volume puts the hopeless one first.
Getting seen, not clicked
Results that rank fine and are chosen anyway
Position is not the problem, so the title and description are. Measured against what this property itself achieves at that position - not an industry table.
Losing clicks
Pages that earned materially fewer clicks than the 28 days before
Gated on the earlier period having had real traffic. Without that gate, three clicks falling to one is a 67% collapse and the top of the list is every page nobody visits.
Competing with yourself
One search, several of your own pages splitting it
Two pages with impressions describes almost every query on almost every site. The signal is the split, so each page has to carry a real share before it counts - and you are shown which page Google actually prefers.
Crawled, but never shown
Pages our crawler reached that Google has not put in front of anyone
Only pages the last crawl found indexable and answering 200 - a noindex page missing from Google is the system working, not a problem.
Did your fixes work?
The 28 days after a change against the 28 before it
For every finding you marked done and every page you rewrote. A change younger than the window says so instead of comparing half an 'after' to a full 'before', which would show a fabricated collapse at exactly the moment you go looking.
The two that need both sides
Crawled, but never shown and Did your fixes work? are not analyses of Search Console data. They are comparisons between what our crawler found and what Google did about it - which is a question neither system can answer on its own.
One tells you a page is reachable, indexable and invisible. The other tells you what happened in the 28 days after you fixed something, against the 28 before, so a change you made has a verdict instead of a hope.
Indexation is asked, not inferred
A page with no impressions might be unindexed, or indexed and never shown, or indexed and shown to nobody searching for what it answers. Those look identical from outside and only one is worth your morning.
So the URL Inspection API is asked directly and Google’s own verdict is reported - indexed, crawled but not indexed, excluded, discovered and not yet crawled - with the reason it gave and the date it last looked.
- IndexedGoogle is holding it
- Crawled, not indexedfetched and rejected - the one to read
- Discovered, not crawledknown, not yet fetched
- Excludeda rule of yours, or a duplicate
All four look the same from outside: no impressions.
The queries one place short of the fold
A query averaging 11 to 20 is already being shown to people. It is not a content gap - it is a page that ranks, one position below where anybody looks, and the work is an edit rather than an article.
Weighted by the impressions the position was measured over, so a rank held once on a quiet day does not outweigh one held all week.
“memewars crypto” — position 11.7
26 impressions, no clicks. Two places is the distance between being shown and being seen.
What it refuses to tell you
A query with nine impressions that moved three places
Under 50 impressions nothing is listed. That is not information, it is noise with a number on it.
That half your pages are below average
The curve is a median, so half the property sits under it by construction. A row has to be at half its band's median before it counts as underperforming.
A collapse that is really a recent fix
A change younger than the comparison window is marked not ready rather than compared against a partial period.
The (other) bucket
Google's anonymised remainder has no query and no page to act on. It stays in the totals and out of every list.
No extra quota, and read-only access
The connection is Google’s own consent screen, read-only, and every analysis here is a query over rows already synced - so opening this page costs nothing, and none of it can change anything in your Search Console account.
Questions
- Search Console shows you what happened. These six answer 'so what do I do' - and two of them cannot exist inside Google at all, because they compare Google's reality against our crawl of your site.
- From your property, not from a published industry table. Median click-through per position band, measured on your own queries. A site whose searches all carry a featured snippet has a genuinely different curve, and measuring it against somebody else's would flag every page it has. Too little data means no curve and nothing shown, rather than a guess.
- No. Every one of these is a query over data we already synced. No Google call, no credit, no extra column.
- Each analysis compares 28 days against the 28 before it. The first sync pulls 16 months, which is as far back as Search Console keeps, so those comparisons exist on the first day rather than in two months.
- 50 impressions for a query to be considered, 100 for a cannibalisation split, 10 clicks in the earlier window before a decline counts, and a 30% drop to call it a decline. Blunt numbers on purpose: a threshold you can be told is a threshold you can argue with.
Start free
One site and 100 credits - up to 200 pages a scan, about 5 scans, and the whole report. No card. Paid plans add more sites, bigger crawls and the AI rewrites.