Google may still judge your site on the programmatic pages you already deleted

Google's John Mueller says its systems may still be weighing the thin, templated pages a site published years ago, long after that site has cleaned them up, a remark that arrives while publishers are still digging out from the August 2026 spam update and a parallel wave of manual actions.

Mueller made the comment in a Bluesky reply, reported on September 8 by Search Engine Journal. He was answering someone whose site generated a page for every combination of domain names, technologies and attributes. That is not an exotic setup. It is the shape of a great many software content programmes: a page per integration, a page per competitor comparison, a page per industry, all built from one template and one list.

What was actually said, and what was not

Mueller's line was about trust rather than about a penalty. He wrote that "our systems have possibly lost faith in your site providing good value to users based on the old pages," and that fixing it "tends to take time & significant effort to show the value." He was careful about scope. The category as a whole was not his target; his words were "programmatic SEO like this," pinned to the specific build in front of him. He did not name a ranking system, a threshold, a score or a timeline. The word he chose was "possibly."

Our systems have possibly lost faith in your site providing good value to users based on the old pages.

That hedging matters, because the useful part of the remark is not a mechanism. It is the sequencing. Deleting or rewriting the templated pages is one job. Convincing Google's systems the site is worth ranking again is a second, slower job, and the first does not automatically deliver the second. Search Engine Journal notes Mueller separately told another site that resolving bigger algorithmic issues can take "sometimes many months."

Google's own spam policies define scaled content abuse as generating many pages "for the primary purpose of manipulating search rankings and not helping users," and state plainly that the definition applies "no matter how it's created." The listed examples include using generative tools to produce many pages without adding value, stitching content from different pages together, and creating many pages whose text carries search keywords but makes little sense to a reader. Purpose and value decide it, not the tooling.

Why this bites software marketers specifically

There is an awkward tension here that the trade coverage has mostly skipped. Google's own documentation on AI features describes a "query fan-out" technique, in which AI Overviews and AI Mode issue multiple related searches across subtopics to assemble one answer. Plenty of growth teams read that and drew the obvious conclusion: cover more phrasings, ship more pages. The spam policy names roughly that behaviour when the pages carry no independent value. Both documents are live on the same site. The reconciliation is entirely on you.

The second problem is that this is a site-level judgment, not a directory-level one. If Mueller is describing what he appears to be describing, the pages that pay your bills are hostage to the templated directory the growth team shipped in a sprint. Pricing, the product tour and the security page all sit behind the same site-level judgment. A comparison-page farm that never converted anyone can quietly tax the pages that do.

A contributor column published in Search Engine Land on September 9 argues that the sites hit by the August update were not punished for using AI, but for shipping without enough human oversight to hold a quality floor. Worth noting: Search Engine Land is owned by Semrush, which sells tooling in this category, and this is a consultant column rather than a news report. It says manual actions are running concurrently, that they hit far fewer sites than the update but that a meaningful number of publishers are dealing with both at once, and that manual actions often land on a single subdomain or directory before spreading.

What to do differently this quarter

The expensive mistake, per Search Engine Land, is the panic response: deleting content wholesale, or filing a reconsideration request before the underlying issue is genuinely fixed. It reports that clearing the specific item flagged in Search Console is no longer sufficient on its own, and that a site now needs to be fully compliant with the spam policies before a manual action is lifted. Each rejected request makes the next attempt harder and more costly. Nothing in that is guaranteed to be fast, and Google publishes no turnaround time for reconsideration.

• Inventory your generated pages by template, not by URL. One decision covers a thousand pages.
• For each template, ask whether a buyer arriving from a link, with no search engine involved, gets anything the parent page does not already give them.
• Pull engagement and conversion data per template before you argue about SEO. A template with no assisted pipeline is not a hard call.
• Check Search Console for a manual action before assuming this is algorithmic. The two need different responses.

The broader point is a portfolio one. Search Engine Land argues that drawing half or more of your revenue from Google Search is itself a business risk worth actively managing. For software companies whose pipeline model assumes organic compounds forever, this update is a reasonable prompt to find out how true that is.