Google September 2026 spam update: what should you do now?

What is confirmed about the September 2026 spam update?

The confirmed announcement is limited: a regular Google Search spam update, starting on September 24, 2026, with a worldwide rollout across all languages. There is no confirmed list of industries, websites or content tools singled out by this release. That distinction matters when trying to explain a falling traffic graph.

The useful question is not simply “Was I hit?” Ask “What can I demonstrate is wrong, and would I fix it even without this update?” An inaccurate price, a hacked page or an unintended noindex remains a problem regardless of algorithm news.

How does a spam update differ from a core update or manual action?

A spam update improves automated detection of practices that manipulate search results. A core update makes broader changes to how search results are assessed. A manual action follows a human review and is reported in Search Console. These are different processes, not interchangeable labels for losing rankings.

SpamBrain is one example of Google’s spam-prevention systems. Its existence does not turn every ranking change into a SpamBrain verdict on your website. Automated changes also do not necessarily produce a message explaining the cause. An empty manual actions report therefore does not rule out every algorithmic or technical issue.

Use Google Search Console to investigate your own pattern. A third-party volatility indicator can provide market context, but it cannot diagnose what happened to your particular domain.

Which content and link patterns deserve closer attention?

Look for situations where acquiring search traffic appears to be the main purpose, while visitors receive little useful information. The following are audit scenarios, not a confirmed list of targets or casualties of this September release.

More pages without more answers

Imagine hundreds of pages answering almost the same question. Each repeats the same explanation with a different title and a few synonyms. The concern is not word count or publishing speed alone. It is scaled production intended to manipulate rankings without enough value for the person searching.

For each cluster, document the question each URL answers. Consolidation may help when two pages support the same decision. When they solve genuinely different problems, make those distinctions useful and explicit. A content strategy with clear topic boundaries prevents every keyword variation from becoming another near-identical page.

Comparisons without meaningful judgment

An affiliate page that copies product descriptions and adds purchase buttons provides little decision support. Explain selection criteria, intended users, limitations, alternatives and verifiable differences. Claim to have tested a product only when that actually happened. Comparing published specifications is not the same as conducting a hands-on test.

Affiliate links are not automatically harmful to SEO. Disclose the commercial relationship and qualify paid links with rel="sponsored"; nofollow is also an acceptable qualification. Correct link attributes do not compensate for an unhelpful article.

Manufactured authority and misleading journeys

Paid ranking links, expired-domain abuse and content exploiting another site’s reputation deserve separate investigation. This does not make every contributed article or acquired domain spam. Purpose, implementation and actual value matter.

Check for cloaking, hidden keyword blocks, unexpected redirects, fake downloads and buttons that do something other than promised. A free checker that never checks anything and merely redirects visitors has a fundamental usefulness problem. A back button that traps people creates a deceptive experience. Rewriting the meta description will not solve either issue.

Is AI content being targeted, and where does SAFE fit in?

Using AI does not by itself establish spam. AI can support research, organization, translation and drafting. The risk arises when a workflow produces unreliable or barely useful pages at scale to manipulate search results. Human-written content can have exactly the same problem.

Assess accuracy, completeness and practical usefulness. Can the reader make a decision or complete a task afterward? Are prices, product names and conditions correct? Is there a concrete example that adds something beyond the other pages covering the topic? A “humanizer” that changes sentence patterns does not answer these questions.

Does SAFE power this update?

SAFE stands for Scaled Abuse Forensics Examiner and appears in research on investigating coordinated synthetic abuse. The existence of that research does not establish that SAFE powers this Google Search release. Treat a proposed connection as unconfirmed, not as a description of the ranking algorithm. It also does not support a prediction that every AI-assisted website will disappear.

Are location pages, business profiles and directories now a risk?

A location page can be useful when it serves a real local need: access details, service coverage, relevant projects, local constraints or an accurate account of work performed there. Merely replacing city names above the same text can create doorway pages that funnel visitors to one destination without providing distinct value.

A shared layout is not the problem. Do not invent an office, case study or review to make each page appear unique. A business serving several towns from one location can explain that honestly. Do not create profiles that imply staffed branches where none exist.

Check the name, category, contact details and actual services in your Google Business Profile. Measure organic website performance and Maps visibility separately. This Search update is not evidence of a separate update affecting every business profile.

A few relevant, accurate business listings may help customers identify a business. Creating 30 profiles, filling three pages of branded results or obtaining a knowledge panel is not guaranteed protection against spam systems. Correct outdated profiles before opening more accounts.

How do you investigate a drop in Search Console step by step?

Start with a comparison someone else could reproduce. Otherwise, you might compare mobile visitors in one country with every device worldwide, or an incomplete day with a complete one.

1. Select the correct property. Open the website in Search Console. Go to Performance and the search results report. Confirm that the search type is Web.
2. Record the filters. Note country, device, query and page filters. Keep them consistent across both periods. Export the baseline before making changes.
3. Compare complete periods. Open the date selector and choose Compare. Use equal periods containing the same weekdays. Record September 24 in your own change log. Treat the first rollout days as provisional observations, not a completed after-update measurement.
4. Enable all four metrics. Examine clicks, impressions, CTR and average position together. Compare the same queries as well: a different query mix can move the average without every existing ranking declining.
5. Open Pages and Queries. Establish whether the decline affects one URL, a directory or the entire site. Select an affected page, then inspect its queries. Separate branded and non-branded terms where useful.
6. Inspect representative URLs. Check indexing status, the selected canonical and the last crawl. A live test helps investigate current accessibility; it does not replace stored indexing information. Also check Manual actions and Security issues.
7. Compare the timing with your change log. Record migrations, deployments, redirects, content edits and measurement problems. Examine analytics separately. A broken analytics tag can reduce reported sessions while Search Console clicks remain unchanged.

Query rows do not always account for the full chart total, including because of privacy limitations. Avoid drawing firm conclusions from a row sum that does not reconcile with the graph. Save the filter settings with every export so future comparisons use the same population.

How do you turn a traffic change into a testable diagnosis?

A useful diagnosis connects a measurable signal to a hypothesis and a check. Do not begin with “spam” as the explanation for every anomaly.

Worked example: 30% fewer clicks is not automatically a 30% ranking loss

A hypothetical page receives 20,000 impressions and 1,000 clicks in the first measurement period: a 5% CTR. In a comparable second period, it receives 20,000 impressions and 700 clicks: a 3.5% CTR. Clicks fall by 30%, while average position moves from 7.1 to 7.3.

This example does not establish a spam issue. It first calls for investigation of queries, devices, snippets and search results. Even a similar average position can conceal underlying changes. The appropriate next step is segmentation, not automatically deleting 300 pages.

Download the blank spam update measurement worksheet (CSV). Use one row per URL or cluster and a fixed set of filters. Record observations, hypotheses and demonstrated errors separately. The worksheet contains no sample results that could accidentally be mistaken for your own measurements.

Which ten checks should come before major website changes?

Work from demonstrable errors toward editorial improvements. This is a practical prioritization, not an official recovery formula.

1. Check availability. Test affected URLs for error responses, unintended redirects and server failures. Produce a list of specific issues rather than relying on one overall speed score.
2. Check indexability. Compare robots settings, noindex directives and canonicals with the pages you intend to publish. Record deviations by template.
3. Build a content sample. Examine declining and stable pages from the same cluster. This helps distinguish an issue associated with the loss from something that exists throughout the site.
4. Test the answer. Ask someone to answer the main question using only the page. Record missing steps, criteria and exceptions. Adding more words is not automatically the remedy.
5. Verify experience claims. Check evidence for hands-on tests, client results, credentials and author roles. Remove inaccurate claims rather than inventing supporting evidence afterward.
6. Review local variations. Identify the actual service and useful differences for each location. Propose consolidation where pages have no distinct function.
7. Investigate link patterns. Examine links you purchased or placed through automation before rejecting unfamiliar domains. Our explanation of backlink assessment and disavow helps distinguish cases that may warrant intervention.
8. Keep business details consistent. Check names, websites, contact details and company-author relationships across existing profiles. A genuine, current listing is more useful than a fabricated one.
9. Test the promised action. Do forms, downloads, tools and back navigation work? Check on mobile. A visitor should be able to complete the task the page promises.
10. Assign an owner. Record who is responsible, when the change happens, its expected effect and the review date. Keep each intervention traceable.

When technical and editorial issues overlap, commission a focused SEO audit. Ask for evidence behind each finding, the affected URLs and an actionable priority. A document full of red scores is not enough to justify major changes.

What should you fix now, and when should you evaluate recovery?

Correct proven errors immediately. Do not wait for the rollout to finish before fixing a hack, accidental noindex, misleading claim or broken form. Preserve a baseline and document the change. Be more cautious about broad strategic conclusions while the rollout is still in progress.

Make a consistent comparison after completion. Use complete periods, matching filters and enough data. Distinguish technical recovery from the effects of content improvements. Include leads and revenue where measurement is reliable: recovered clicks without relevant enquiries are not a complete business outcome.

Allow time for sustained improvements. Spam-related reassessment can take months. There is no button that requests an automatic ranking reassessment. A reconsideration request belongs to a corrected manual action, not every decline in organic visibility.

What should you avoid?

• Deleting every page that receives less traffic for one week.
• Redirecting the entire website to a new domain without a diagnosis.
• Buying profiles, reviews or links as supposed update protection.
• Only making articles longer or trying to pass an AI detector.
• Promising recovery by a fixed deadline or guaranteed top rankings.

The next step should be specific enough to verify and important enough to solve a real problem. A broader effort to improve Google rankings starts with the same foundations: useful answers, accessible pages and quality you can demonstrate.

What else should you know about this spam update?