6 min read
Four of our 25 pages had never been fetched by Google. One still hasn't.
Being in the sitemap is not the same as being crawled. We ran Google's URL Inspection API against all 25 URLs on our four-day-old domain and found four that had never been fetched at all — one of them the highest-value page we had published. Three were indexed within about 25 hours of a manual indexing request. The fourth is still unknown to Google a day later, and the difference between them is the interesting part.
The check nobody runs, and it takes one API call per URL
The Search Console URL Inspection API returns, for any URL on a property you own, whether Google has ever fetched it, when it last did, which canonical it chose, and what the coverage state is. Running it across an entire sitemap is a loop, not a project. We ran all 25 and got: 21 indexed, three never crawled, one queued but never fetched.
The assumption that breaks here is that a sitemap entry produces a crawl. It produces a candidate. On a domain with no external links, crawl budget runs out before the sitemap does, and the pages that lose are not the ones you would guess.
The most expensive miss was our comparison page for an expense tool — the page with the best commercial value on the site, targeting a keyword cluster with cost-per-click between $16 and $62. It had been live for days. Google had never fetched it once.
What the four had in common
All four were poorly linked from within the site itself. The comparison page had two inbound links from page bodies, and it linked out to five guides that did not link back — a sink. The two app pages that went unfetched were linked only by a card grid that repeats identically on every page, which is template furniture rather than an editorial signal. The support page had no body links at all.
We only learned this by building the graph: fetch every sitemap URL, extract internal links, and count inbound links per page while excluding anything inside a header, nav or footer. Without that exclusion every page on a 25-page site looks equally well linked, because the chrome links to everything.
The lesson is not new, but the size of the effect on a zero-authority domain was larger than we expected. Internal links were doing the entire job of discovery, because there was nothing else.
The timeline, exactly
On 15 August we requested indexing for all four through Search Console, one at a time, and separately added five contextual links to the comparison page from guides that Google was already recrawling every two or three days. The next morning, three of the four were fetched — timestamps of 18:49, 18:49 and 18:51 UTC — and all three were reported as Submitted and indexed. The fourth had not moved.
So a manual indexing request works, and it works quickly. What it did not do was fix the page whose problem it did not address.
Both things were done on the same day, which means we cannot cleanly separate what the request did from what the new links did. That is a real limit on this experiment, and we would rather state it than imply a cleaner result.
The page that stayed invisible
Our support page sits in the sitemap, is linked from the site-wide footer, serves a self-referencing canonical, has no noindex, and ships FAQ structured data. Search Console's answer is still: URL is unknown to Google. No referring sitemaps detected, no referring page detected.
Nothing on the page is telling Google to stay away. It is the one page of the four with no editorial link from anywhere — footer links did not count for it, and no article had a reason to point at it.
The honest conclusion at this point is that a request alone did not rescue a page that nothing links to, while a request plus new inbound links coincided with three pages being fetched within a day. That is an observation from a sample of four on one small site, not a rule.
What we would tell another small site to do
Inspect every sitemap URL rather than assuming. Build the internal link graph with chrome excluded, so you can see which pages have real editorial support. And treat publishing and being crawled as two separate events that can be days apart, because on a new domain they are.
The cheapest recurring check we have found is the sitemap URL count: it catches a broken route before anything else does. The second cheapest is running the inspection loop after each content deploy instead of occasionally.
Both of these are free, both use APIs Google already gives you for your own property, and neither requires a paid tool.
More from the studio
Looking for the practical material instead? Our guides cover organising receipts, choosing a scanner app and keeping business records.