Does "URL is unknown to Google" mean Google never found the page? Four inspections of 14 unfetched posts within 8 minutes returned both "unknown" and "Discovered" on 11; the two pages Googlebot fetched once were indexed within 40 hours

The question

Google's URL Inspection tool help defines the verdict "URL is unknown to Google" as: "This means that Google hasn't seen this URL before." Its Page indexing report help defines "Discovered - currently not indexed" as: "The page was found by Google, but not crawled yet. Typically, Google wanted to crawl the URL but this was expected to overload the site; therefore Google rescheduled the crawl. This is why the last crawl date is empty on the report." Read literally, a page moving from "Discovered" back to "unknown" means Google forgot a URL it had found. We reported exactly that on 2026-09-15 and 2026-09-16, calling the moves regressions. This post checks whether the two labels are stable enough to read that way.

Method

• Population: the 14 posts this site published between 2026-09-09 and 2026-09-16, all in the sitemap, all HTTP 200 on every read, live 1.7 to 8.7 days at the read.
• Daily reads: the URL Inspection API (the same call as the Search Console tool) on 2026-09-11, 2026-09-13, 2026-09-14, 2026-09-15, 2026-09-16, 2026-09-17, one committed artifact per day; 54 inspections in total, each post entering on the first read after it shipped.
• Repeat reads: on 2026-09-17 every post was inspected four times within 8 minutes (the daily read plus three more), and the six older posts our news hub links to were read the same way, two of them indexed, as the control.
• Server log: every request whose user agent is a crawler is written to a ledger at the edge (live since 2026-08-16); the Googlebot rows for each path since its publish date sit beside every inspection.
• Request-indexing was never used on any of these URLs. IndexNow pings went out the day each post shipped, which the IndexNow measurement already showed does not bring Googlebot.

The daily reads

Across the 6 reads, 9 of the 14 posts changed label at least once: 5 moves from unknown to Discovered and 6 from Discovered to unknown. 6 posts were "forgotten" (Discovered on one day, unknown on the next), and 1 of those was Discovered again later. No post was ever indexed or "Crawled - currently not indexed", no read ever carried a last-crawl time, and the server log holds 0 Googlebot requests for the 14 paths since they shipped.

Four reads within 8 minutes

The same URL, inspected minutes apart with nothing changing on the site and no Googlebot request in between, came back "unknown" on one read and "Discovered" on the next. Every URL that was stable across the four reads was stable on "Discovered"; no unfetched URL held "unknown" on all four. The two indexed pages never wavered. The sequences per URL are in the artifact.

What one Googlebot fetch did

Both fetched pages had sat unfetched for three weeks and had read "Discovered" or "unknown" on earlier daily reads. Each got exactly one Googlebot request, both on 2026-09-16, and each read "Submitted and indexed" on the next day's inspection with a last-crawl time that matched our own log within a second. This repeats the single-fetch reading from 2026-09-06 and the hub read of 2026-09-11, where every indexed page had a fetch and every unfetched page stayed out. Googlebot has not requested this site's homepage since 2026-09-12; 5 consecutive days at zero, and 0 requests for the news hub in 7 days.

Reading

• For a page Googlebot has not requested, "URL is unknown to Google" and "Discovered - currently not indexed" are not two states of the page. They are two answers the API gives for the same uncrawled URL, and it can give both within 8 minutes. Our 2026-09-15 and 2026-09-16 reports of "regressions" from Discovered to unknown were reading this noise, and we retract that word.
• The reading that held on every row is binary: a Googlebot request in the server log, or none. 2 of 2 pages with one request were indexed within 40 hours; 0 of 18 pages without one were indexed, at 1.7 to 25.7 days live.
• The last-crawl time is trustworthy when present: it matched an independent server log on both fetched pages to the second. An empty one, with either label, means the page has not been fetched, and the label beside it carries no extra information.
• A page Google has not fetched cannot appear in Google Search or AI Mode, and the same "was it fetched" question applies to the AI engines; how to check whether your business shows up in AI covers the engine-side read.

Pre-registered for 2026-09-24

1. 1Any of the 14 posts that receives a Googlebot request in the server log reads indexed on the first daily inspection at least 48 hours after that request. A fetched page still out at 48 hours breaks this.
2. 2On a four-read run of the posts still unfetched on 2026-09-24, at least half return both labels. Fewer than half means the label is more stable than this read found.
3. 3The two indexed control pages read indexed on every read of the same run.

Artifact

`content/news/index-state-cut-2026-09-17.json` holds each post's label on every daily read, the flips counted, the four-read sequences for all 20 URLs with timestamps, the Googlebot ledger rows per path, the two fetched pages' last-crawl times beside the log rows, the homepage Googlebot trend, and 16 self-checks. `content/news/index_state_cut.mjs` rebuilds it; `content/news/trailing_inspect.mjs` produces the daily inspection artifact it reads. Paths only; no names, addresses or IPs.