If you woke up to a cliff in your charts, start here: Barry Schwartz just published “Search News Buzz Video Recap: Google September 2026 Spam Update Is Here, Goodbye Discover Traffic, Multimodal Search Reporting & CrUX Ad Metrics In Page Experience.” Watch it: YouTube. It’s a clean read on what changed and why your traffic might be down, and it separates algorithmic hits from Discover repricing. 🚨 (seroundtable.com)
Why this is worth your time: Google confirmed a global “September 2026 spam update” kicked off Sep 24 at 09:15 PT with a rollout that can take up to two weeks (Search Status Dashboard). Barry also flagged new “[Web multimodal]” reporting in GSC (Search Central blog) and fresh ad metrics in CrUX (Chrome Developers). That combo explains a lot of noisy charts, so don’t treat a feed swing like a spam penalty. Different problems. Different fixes. (status.search.google.com)
- Segment firstSeparate Discover vs. Web Search in GSC; confirm by device, country, and page set.
- If spam-hitRemove thin/duplicative/scaled pages; consolidate near-dupes; halt link schemes.
- If Discover downRe-check 1200px+ images, headlines, cadence, policy compliance; expect volatility.
- Prepare for AI searchMake answers extractable and citeable; track where AI overviews cannibalize demand.
What Barry reported and why it matters
Google’s “September 2026 spam update” is live (Sep 24, 09:15 PT; global; up to two weeks) and enforces existing spam policies: scaled content abuse, link schemes, cloaking, doorway patterns, and so on (status dashboard; Search Essentials spam policies). Barry’s take is straightforward: another spam classifier is rolling, Discover is turbulent again, and reporting shifts (Web multimodal, CrUX ad metrics) are adding noise. 🔍 (status.search.google.com)
Two realities can be true at once:
- Your site tripped a spam classifier.
- Your Discover traffic got repriced because feeds are volatile by design.
Treat them differently. Conflating them wastes weeks and usually leads to gutting content that wasn’t the problem.
What to do if the September spam update hit you
If the drop is in Web Search (not just Discover) and lines up with the rollout window, clean house. 🧹
// My reaction
When the algorithm bites — time to tidy up
- Find and fix scaled low-value content
- Deindex or rewrite near-duplicate pages and templated variants that add no unique value.
- Kill doorway-like city or service matrices with token-swapped copy.
- Remove “zombie” hubs with 0 clicks over 90–180 days unless they’re strategically necessary. When unsure, noindex first and review impact.
- Check auto-generated or lightly edited AI content
- AI-assisted is fine; AI-at-scale with no editorial value is not. Add original analysis, data, or media. If you can’t add something new, consolidate or remove. (Scaled content abuse policy) (developers.google.com)
- Links
- If you ran paid guest posts, PBNs, or mass directory blasts, stop and start removing what you can. Disavow is a last resort and usually unhelpful without a manual action (Disavow guidance). (support.google.com)
- Technical signals that help the clean-up stick
- Make canonical intent obvious. Cut parameter bloat. Fix faceted crawl traps.
- Maintain a tight sitemap for index-worthy pages and exclude the rest. That’s index management, not decoration.
If you want a step-by-step, I wrote the exact play I’d run: <a href="/blog/google-rankings-drop-september-2026-triage">Google Rankings Drop — September 2026: Triage & Fixes</a>.
Discover traffic turbulence: what’s signal vs noise
Discover is option-value traffic. It surges and dies with topic interest, packaging, and imagery. It is not guaranteed. Treat it as bonus, not baseline.
Do these checks before you panic:
- Verify the drop is Discover-only in GSC’s Discover report. Compare to Web Search.
- Headlines: direct, accurate, and compelling without withholding key info.
- Images: use at least one 1200px-wide image and allow max-image-preview:large so Google can show large cards (Discover docs). 🖼️
- Freshness: publish on a steady beat. Update materially, not cosmetically.
- Policies: re-read Discover content policies and remove sensational framing. (developers.google.com)
And yes, even with 1200px plus large previews, some publishers still see small cards, practitioners suspect rendered width or image sizing quirks. That’s anecdotal. Test on your own site.
| Signal to check | Why it matters |
|---|---|
| GSC Discover vs. Search reports | Confirms which system changed |
| Device and country splits | Isolates feed-driven vs. query-driven exposure |
| Image resolution and meta robots | Large previews affect CTR; blocked images kill reach |
| Update cadence and topic mix | Evergreen alone rarely wins a novelty-first feed |
How long does recovery take after a spam update?
There isn’t a fixed timeline. If you remove the root causes, you can see partial improvements as Google recrawls, but bigger lifts often line up with future updates.
Should I use disavow?
Only if you have clear evidence of manipulative links you created or controlled. It’s not a cure-all, especially without a manual action.
Is Discover a ranking factor for Search?
No. Treat them as separate systems with separate reports in GSC.
Clean up scaled junk, make answers extractable, and stop treating Discover like payroll. That combo survives spam updates and earns citations in answer engines.
- Barry Schwartz: Google September 2026 Spam Update Is Here, Goodbye Discover Traffic, Multimodal Reports & CrUX Ads (YouTube)
- Google Search Status Dashboard: September 2026 spam update
- Google Search Essentials: Spam policies
- Get on Discover (Google Publisher Help)
- Search Central: Web multimodal in Search Console
- Chrome Developers: New CrUX ad metrics
- Search Central: Generative AI performance reports in GSC
- Google Search Status Dashboard: Feb 2026 Discover-only core update
- Google: Disavow links guidance
Multimodal and reporting notes
Google added a “Web multimodal” filter in GSC that groups Lens, Circle to Search, image uploads, and “Search this image.” Don’t average this into regular text search; it behaves differently (Search Central blog, Sep 24, 2026). 🧪 (developers.google.com)
What I’d do:
- Annotate your analytics with the spam update window and any UI or reporting changes.
- Track image and video clicks separately; don’t bury them in “organic.”
- Watch CrUX’s experimental ad metrics (ad count, density, network and CPU “ad weight” added Sep 15) to quantify ad-heavy experiences that can depress real-world engagement (Chrome Developers; Page Experience doc now references them too: Understanding page experience). (developer.chrome.com)
- Third-party rank trackers may show gaps or sampling artifacts where bot protections and anti-scraping measures bite; lean on GSC, GA4, and logs until coverage stabilizes (Google vs SerpApi legal filing). (blog.google)
GEO reality check: AI answers are eating easy clicks
Spam updates and Discover swings are symptoms. The disease is this: answer engines keep low-complexity queries on their screens. Your counter-move is to be the source that gets cited, not the site that gets summarized. 🧭
Do this next:
- Identify “AI-takeable” queries and measure displacement. Use GSC’s Generative AI performance reports and segment “Web multimodal” to see image-led exposure separately (Generative AI performance reports; Web multimodal in GSC). (developers.google.com)
- Identify “AI-takeable” queries and see whether you’re cited. Use the <a href="/tools/ai-visibility">AI Visibility Checker</a> to spot where overviews appear.
- Rewrite priority pages to be answer-first, citeable, and extractable. Here’s the pattern I use: question, 2–4 sentence answer, then depth. See <a href="/blog/build-content-ai-search">Build Content for AI Search</a> and <a href="/blog/ai-answers-seo-playbook">AI Answers and SEO: Protect Traffic</a>.
- Add original data, steps, or media the models can’t fake. That’s what earns links and citations long-term.
Takeaway
This update is a stress test. If you relied on scaled thin pages or wishful Discover spikes, the bill came due. Clean up page sets, make your best content answer-first and citeable, and annotate your reports so you can tell algorithmic shifts from reporting or feed swings. That’s how you protect and grow visibility across SEO, GEO, and AI-driven search—and it’s exactly where I’d spend my next two weeks.
// My reaction
