Google’s September 2026 Spam Update (September 24, 2026): Spam Demotion or Ordinary Volatility? How Michigan Sites Should Tell the Difference Before October 8

Google released the September 2026 spam update on September 24, 2026, and said the rollout “may take up to two weeks to complete,” which puts the finish line around October 8. That is the longest spam rollout Google has announced this year: PPC Land clocked the March 2026 update at about 19.5 hours, June at about 48 hours, and August at roughly three days.

For Michigan businesses, the timing is awkward. Rank trackers were already spiking on September 23 before the announcement, Discover traffic has been falling for many publishers, and AI Overviews keep eating clicks. If your traffic dips this week, the spam update will get the blame by default. Below: how to tell a real spam demotion from ordinary volatility, and why the fix for one harms the other.

What is the September 2026 spam update, exactly?

It is an improvement to SpamBrain, Google’s automated spam classifier, applied globally and in all languages. It enforces existing policies; Google has not said which techniques it emphasized.

PPC Land’s coverage lists the policies SpamBrain enforces: scaled content abuse, expired domain abuse, site reputation abuse, doorway abuse, fake or undisclosed incentivized reviews, and back button hijacking (enforced since June 15, 2026). Barry Schwartz at Search Engine Roundtable noted the update is not specifically a link spam update. It is the fourth spam update of 2026.

How is a spam demotion different from a core update or normal volatility?

A spam demotion hits a pattern of pages that share a trait; a core update or volatility moves individual queries by relevance. Shape is your best diagnostic, because Google sends no notice for algorithmic demotions.

Here is the mechanism. SpamBrain is a classifier. It looks at signals across a group of URLs (a template, a subfolder, a hostname) and decides whether that group matches a spam pattern. When it fires, the demotion applies to the group, not to one keyword. A core update works differently: it re-scores how relevant and helpful each page is for each query, so the losses scatter across your site in proportion to how contested each query is.

Say a Sterling Heights roofer has 45 templated city pages plus a blog. If SpamBrain flags the city pages as doorways, those 45 URLs drop together, by a similar percentage, on roughly the same day, while the blog and the homepage hold. If instead the site lost ground in a core update or to AI Overviews, you would see a scattered pattern: some city pages up, some down, informational blog posts losing clicks while impressions stay flat.

Which symptoms point to spam, and which point to something else?

Compare the shape of the loss against three rival explanations before you touch anything. Use this side-by-side:

• Spam demotion: losses cluster in one URL pattern or subfolder; impressions and position both fall; onset lines up with September 24 or later; brand and homepage queries mostly hold.
• AI Overview click loss: impressions steady or rising, clicks and CTR falling; concentrated on how-to and “what is” queries; no clean start date.
• Discover decline: the drop shows up in the Discover report, not the Search results report; organic Search is flat.
• Tracker noise: third-party rank tools move, but Search Console clicks for the same pages barely change.

We wrote a longer diagnostic for the click-loss case in our breakdown of how Michigan sites can tell a traffic drop from a ranking drop. The short version: if position did not move, SpamBrain probably did not touch you.

Does “longest rollout yet” mean this update is bigger?

Not necessarily, and treating it that way will lead you to misread your data. A longer rollout tells you about deployment pacing, not about how many sites get hit.

John Mueller explained on March 31 that long core update rollouts happen because “several ranking components are updated step by step.” Google has not said the same logic applies here, but with a 14-day window, your site could be evaluated on day two or day twelve. A flat week does not mean you are in the clear, and a dip on September 25 does not prove the update caused it, since trackers were already elevated on September 23.

Here is where we disagree with the standard advice. Google’s line is that sites following its policies have nothing to worry about, and most coverage repeats it. That is true on average, but the average hides two groups. Sites that knowingly scaled thin content are the obvious losers. The less obvious group is legitimate businesses whose templates look like spam patterns: franchise location pages, dealer locators, and parts catalogs with near-identical fitment pages. Search Engine Land acknowledged that collateral damage occurs. Google has published no numbers on how many queries are affected, so plan for the possibility that you are in the second group.

Why are Michigan service and catalog sites more exposed?

Metro Detroit’s economy relies on two site types SpamBrain looks at closely: multi-location service pages and templated product catalogs. Both can resemble doorway or scaled-content patterns even when a real business stands behind them.

Take an HVAC company based in Livonia that serves Westland, Canton, Plymouth, Novi and 30 other cities, with one page per city where only the town name changes. Heating season starts in October, and furnace-repair calls are the year’s highest-margin work. If those city pages lose half their visibility on October 3 and nobody spots it until November, a company closing 20 service calls a week from organic search could lose several hundred jobs before the cold snap ends.

Automotive aftermarket sellers in Warren and Madison Heights face a similar risk with fitment pages generated for every year, make, and model combination. We covered where Google draws that line in our piece on Google’s “lost faith” comment on programmatic SEO. The test is whether each page carries information a buyer cannot get from its neighbors.

How should a small team run the two-week watch?

Set a baseline now, check it on a fixed schedule, and do not rewrite anything until the rollout ends. Each step takes under an hour.

1. Export a baseline today. In Search Console’s Performance report, export pages for September 10–23 with clicks, impressions, and average position. Label it “pre-update.”
2. Group URLs by template. Add a column that tags each URL by pattern: /locations/, /parts/, /blog/, product, homepage. SpamBrain’s signature only shows up at this level.
3. Check twice a week, not every day. Compare Monday and Thursday exports against the baseline through October 8. Daily checks during a staggered rollout create false alarms.
4. Apply the shape test. A template group down more than 30% on both impressions and position, while other groups hold, is a spam signal. A drop in clicks with steady impressions is not.
5. Check Manual Actions anyway. Algorithmic demotions show nothing there, but a manual action during the same window changes your response completely, so rule it out.
6. Draft fixes without deploying them. If a template is flagged, write out what you would merge, noindex, or enrich, but hold the changes until the rollout finishes so you can see where the update settled.
7. Fix by adding substance, not by cutting everything. For city pages, keep the locations you actually serve and add real proof to each: technician names, local job photos, permit-specific details. Consolidate the rest into one service-area page.

How long does recovery from a spam update take?

Months, not weeks, according to Google’s own documentation. Google’s spam-updates page says changes “may help a site improve if our automated systems learn over a period of months” that the site complies.

That is why diagnosis matters more than speed. If you misread AI Overview click loss as a spam hit and noindex 40 working city pages, you lose rankings that were never in danger, and it takes months to earn them back. For link spam specifically, the same documentation warns that any ranking benefit from spammy links “cannot be regained.” Disavowing will not bring that traffic back.

What dates matter over the next few weeks?

• September 24, 2026: September spam update begins (Google Search Status Dashboard).
• Late September 2026: Google Ads removes campaign-level language targeting for Search campaigns, per the Google Ads Developer Blog. Keep paid and organic data separate.
• October 1, 2026: Google’s Local Services Ads missed-call charge takes effect, which matters for home service companies already under pressure from any organic drop.
• On or around October 8, 2026: the end of the two-week rollout window. Take your final comparison export here.
• Mid-October onward: deploy template fixes, then measure monthly rather than weekly, since Google says recovery takes months.

Frequently asked questions

Will Google notify me if the spam update demoted my site?

No. SpamBrain demotions are algorithmic, so nothing appears in Search Console’s Manual Actions report. The only evidence is in your performance data: a group of URLs sharing a template or subfolder losing both impressions and average position together, starting on or after September 24.

Are location pages for each city I serve now considered spam?

Not automatically. Doorway abuse describes pages built to rank for variations of a query that funnel users to the same destination with little unique value. A Troy page with that office’s staff, reviews from local jobs, and service specifics is a real page. Thirty pages that differ only by the town name are the pattern at risk, whatever the business behind them.

Should I disavow links because of this update?

Probably not. Search Engine Roundtable reported that this is not specifically a link spam update, and Google’s documentation says that when its systems neutralize spammy links, any ranking benefit they provided is simply lost and cannot be regained. Disavowing doesn’t restore anything. Spend the time on templates and content instead.

Does the spam update affect Google Ads performance?

Not directly. Spam updates apply to organic Search rankings, not ad auctions. The indirect effect is real, though. If organic city pages drop, paid campaigns often pick up the slack at higher cost per lead, and Google’s language targeting removal lands in the same window. Compare organic and paid reports side by side before concluding that either one “broke.”