Google's John Mueller told one site owner last week that Google's systems "have possibly lost faith in your site providing good value to users based on the old pages." That line is now circulating without the four posts of context around it, and the missing context changes what you should do about it.
Here is the full picture, then a diagnosis you can run on your own site.
What was actually said, and to whom
On 6 September 2026, a developer posted a numbered thread to Mueller on Bluesky. His site, launched in April, catalogues technology stacks used by websites — his own comparison was BuiltWith. Google indexed many pages at first. He concluded those pages were thin, noindexed them himself, and then hit a different problem: his core pages, including /about, sat at "Crawled - currently not indexed" for months. No manual actions. Real backlinks from Hacker News and Product Hunt. He had even built a Google-only dynamic sitemap limited to pages with more than ten technology detections, specifically to avoid submitting thin URLs.
Mueller replied the next day, 7 September, across five consecutive posts. The relevant one reads in full:
Programmatic SEO like this often leads to a site that's either spam, borderline spam, or low quality. It's easy to spin something up with many pages, it's hard to provide real value to users. Our systems have possibly lost faith in your site providing good value to users based on the old pages.
Two words in that quote are doing heavy lifting, and both get dropped in the version being shared: like this. Mueller had described the specific pattern one post earlier — iterating a large list of domain names, technologies and attributes to find every possible combination. He then added why the combinatorial version fails on its own terms: "You end up with tons of pages that have 'some' value, but the overall picture is not that exciting. Expanding it with even more generated data ('what does php mean?' etc) doesn't make it more useful for users."
So the target is not templated pages as a category. It is combinatorial expansion where each new page adds a row rather than an answer.
What the shared version leaves out
The circulating post reads as a verdict on programmatic SEO. The thread reads as a diagnosis of one implementation, hedged twice ("it's hard to say", "I worry a bit") and followed by a timeline and a caveat that most summaries cut.
Circulating version | What the thread says |
|---|---|
"Programmatic SEO often leads to spam" | "Programmatic SEO like this often leads to…" — referring to the domain × technology × attribute combinations he had just described |
Old pages are "poisoning your entire domain" | "Our systems have possibly lost faith… based on the old pages" — a hedged read of one site, not a stated mechanism |
Quality > quantity. Always. | "Resolving this tends to take time & significant effort to show the value," benchmarked against how teams work through spam and core update recovery — with the explicit aside "not to say that your site is spam" |
Mueller closed the thread by inviting others to write about it and telling the publisher to review those write-ups "with a grain of salt." That instruction applies to the summaries of his own thread too.
None of this makes the warning weaker. Site-level trust loss is the serious claim here, and Mueller states it plainly: removing the pages did not restore the site, and recovery is measured in effort and months. The distortion is in the scope, not the severity.
The part that should worry you
Strip the framing and one uncomfortable fact remains. This publisher had already done the obvious remediation. He noindexed the thin pages himself, before asking. His backlink profile was organic and earned. He had no manual action. And his /about page still would not index.
That is the failure mode worth planning around: the pages you deleted can keep affecting how Google treats the pages you kept. Mueller's phrasing — "based on the old pages" — puts the assessment at the site level and in the past tense. Cleanup removes the cause. It does not reset the assessment.
Diagnose your own set in an afternoon
The test is not whether your pages are generated. It is whether a reader arriving on one gets something they could not assemble themselves.
1. Pull the set and sample it. Export the URLs from one template from Search Console or your sitemap. Take twelve at random — not your best twelve.
2. Ask the substitution question on each. Could a reasonably informed person produce this page's content from a database schema plus five minutes of thought? If yes, the page is a row. If the page contains something only you could supply — measured data, a real screenshot, a caveat about when the combination is a bad idea, an observation from actually using the thing — it is an answer.
3. Count the answers, not the pages. If two of your twelve pass, the pattern is combinatorial and you have a scope problem rather than a page problem. If ten pass, your issue is more likely technical, and the indexing report is the better place to look.
4. Check whether your core pages are healthy. This is the signal that separates a thin-content problem from a trust problem. If /about, your pricing page and your main service pages index normally, the generated set is being judged on its own. If they are stuck at "Crawled - currently not indexed" alongside everything else, the assessment has widened past the template. Reading the page indexing report is the fastest way to confirm whether the problem is spread beyond the template.
5. Decide per cluster, not per site. Where the data genuinely differs per page and readers act on it — a real specification, a real price, a real availability status — that cluster stays and gets improved. Where the page exists because a loop produced it, consolidate the set into fewer pages that each answer something.
The recovery path is unglamorous: fewer pages, each carrying information a reader cannot get by inference, plus the patience Mueller flagged. If Google's assessment of a site can be shaped by pages that no longer exist, then the sequence matters. Prove value on what remains before you scale anything new on top of it.
What this does not tell you
Mueller was looking at one site's description on a social platform, without access to its Search Console data, and he said so twice. Nothing in the thread describes a mechanism, a threshold, a page count, or a timeline in weeks. Nothing in it says a template is disqualifying. Anyone converting five hedged posts into a rule about scaled content is adding the rule themselves.
Treat it as what it is: a Google spokesperson telling one publisher that site-level trust exists, that old pages can inform it, and that winning it back is slow.
FAQ
Does this mean programmatic SEO no longer works? Nothing in the thread says that. Mueller's criticism was aimed at combinatorial page generation where the pages carry "some" value individually and little collectively. A set of templated pages built on data readers act on is a different thing and was not what he was describing.
If I delete the pages, how long until the site recovers? Mueller gave no timeline beyond "tends to take time & significant effort," and suggested comparing it to spam and core update recovery work — which typically runs in months. The publisher in the thread had already removed his pages and had seen no recovery.
How do I tell a thin-content problem from a site-level trust problem? Check whether your non-generated pages index normally. Healthy core pages suggest the generated set is being assessed on its own. Core pages stuck at "Crawled - currently not indexed" alongside the rest suggests the judgment has widened.
Was this said on Reddit or Bluesky? Bluesky, on 7 September 2026, in reply to a thread posted the previous evening. It was then reposted to Reddit's SEO community and covered by industry press, which is where most people encountered it.
Author: Julian Mercer, 14-Year Technical SEO Practitioner at Auspia. Julian writes about crawlability, indexing diagnostics, and the technical foundations behind site-level quality assessments.
