Programmatic SEO after the September 2026 spam update

What did the September 2026 spam update target?

The update started on 24 September 2026 and finished on 8 October, a rollout of 13 days and 16 hours. Hans van Gent’s analysis of the rollout highlights two targets: programmatic AI pages, meaning templated content with minimal variation across many URLs, and self-ranking listicles, where a brand ranks its own product first. He reports that such brand lists were excluded from AI Overview recommendations 83% of the time, up from 69% in June.

The rollout overlapped with Google’s 1 October rewrite of its guidance on generative AI content, which calls manual fact-checking critical for page copy, titles, metadata, structured data and alt text, and moved fabricated author profiles under its deception policy language.

Read the two changes together: the update shows what Google demoted, and the new guidance shows what it expects instead.

Why did self-ranking listicles get hit?

A “best tools” article published by a company that puts its own product at number one is easy to make and hard to trust. Readers have learned to discount it, and the September 2026 rollout suggests search and AI systems have too. Hans van Gent’s analysis puts it plainly: the page built to rank your brand first can feed the case for recommending someone else.

If you publish comparisons, make them honest. Explain the criteria, test the products against them, say where competitors are the better choice, and disclose that you make one of the products. A comparison that sometimes recommends someone else is the only kind that earns trust over time.

What counts as scaled content abuse?

Google defines scaled content abuse as generating many pages for the primary purpose of manipulating search rankings and not helping users. Its examples include using generative AI to create many pages without adding value, scraping or synonymising other content, stitching pages together without adding value, and pages that make little sense to a reader but contain keywords.

The test is purpose and value, not the tool. A thousand pages built from a real dataset that answers a real question can be fine; a thousand pages that swap a city name into the same paragraph are not.

In practice, look at your templates through a reader’s eyes. If a person read three pages from the same template in a row, would each one teach them something new? If the honest answer is no, the template is a risk, however good the keyword research behind it was.

Which programmatic pages are still safe to build?

Ask of each page type: would a person who landed here from search be glad they did, and does this page say something no other page on the site says?

Free tools are a good example of the safer kind. A calculator page with a clear explanation, the formula, worked examples and answered questions gives a reader something a generic article cannot, and people link to tools they use.

How do you build programmatic pages safely?

This is the same discipline I use for AI-generated marketing content: drafts can be automated, but nothing ships until a person approves it, and quality is measured against a golden set rather than assumed.

• Start from a dataset with real variation; if every row says the same thing, do not make pages.
• Give each page a unique section a person wrote or checked: a finding, a recommendation, an example.
• Put a human review gate before publishing, and fact-check titles, descriptions, structured data and alt text.
• Publish in batches, watch impressions and engagement, and prune pages that earn nothing.
• Name a real author and keep author details truthful.

How do you audit an existing programmatic section?

Export every URL in the section with its impressions, clicks and engagement for the last six months. Group the pages by template and sort each group by traffic. In most programmatic sections a small share of pages earns almost everything and a long tail earns nothing. The long tail is where the risk sits.

For each template decide: improve, merge or remove. Improve the pages where you can add unique data or a real recommendation. Merge near-duplicates into one stronger page and redirect. Remove pages that cannot be made useful and return a 404 or 410. Then re-check the section a few weeks after the changes.

• Export URLs with impressions, clicks and engagement.
• Group by template; find the long tail that earns nothing.
• Improve, merge or remove each group.
• Re-crawl and re-check after a few weeks.

What should you do if traffic dropped?

First confirm the drop lines up with the rollout dates rather than a seasonal dip or a tracking change. Then segment by page template: spam updates tend to hit whole patterns, so look for the templates that lost the most. Improve or remove the thin ones instead of rewording them, and expect recovery to take time, often until a later update re-evaluates the site.

Keep a written log of what you changed and when. Recovery from a spam-related drop is slow and noisy, and without a log it is hard to tell later which changes helped. Pair it with weekly exports of impressions by template so you can see each group respond.