Google shipped its fourth spam update of the year with no completion date or impact data yet, Chartbeat’s publisher panel showed Google Search referrals down 40.2 percent over the past year, Search Console added a new multimodal traffic filter, and Google Ads reshaped Customer Match data collection and its budget tools.
Shipped But Unproven: A Spam Update Lands As Referral Losses Deepen
Google started rolling out a new spam update with no completion date and no impact data yet, the same day a major analytics vendor quantified how much search traffic publishers have already lost this year.
- Google Ships the September 2026 Spam Update Over Two Weeks Google released the update on September 24 and said the rollout, its fourth spam action of the year, could take up to two weeks to complete. That window means any ranking movement observed before early October cannot yet be attributed to this update with confidence.
- Google Search Referrals To Publishers Fell 40% In A Year, Chartbeat Finds Chartbeat’s 2026 Publisher Playbook found Google Search referrals to its client panel down 40.2 percent between July 2025 and July 2026, nearly double the prior year’s decline, while Discover referrals fell 34.3 percent over the same window. AI referrals from tools like ChatGPT and Gemini grew eightfold over two years but still made up just 0.01 percent of traffic on the same panel.
Tracked At Last, Explained Later: New Visibility Without New Answers
One rollout adds a data point Search Console never tracked before, while a months-old research paper resurfaces to describe how Google might make search retrieval faster.
- Search Console Adds A Multimodal Filter, No Query Data Yet Search Console now splits performance data into Web: text-based and Web: multimodal buckets, and Google Search Advocate John Mueller, after checking with the team, said the multimodal figures are newly counted data rather than a re-slice of existing totals. The new filter carries no queries dimension at all, since Google says most multimodal searches never generate a text query to log.
- Google Research Details a Faster Engine Behind Search Fan-Outs Search Engine Journal’s September 24 analysis revisits a March 2026 Google research paper and a September 15 Google blog post describing R4T-Diffusion, a compact model Google says generates search fan-out sub-queries 12 to 20 times faster than its prior method. Google has not confirmed the model is running in production, so the piece examines existing research rather than a new announcement or a live ranking change.
Ads Platforms Widen What They Collect And How They Spend
Two Google Ads changes from Thursday touch what data advertisers can gather and how their budgets get allocated across campaigns.
- Google Ads Customer Match Adds Unhashed IP and Timestamp Matching Google Ads Customer Match now accepts user IP address and interaction timestamp as unhashed matching keys, expanding the upload template to eight columns, but the feature excludes end users located in the EEA, UK, and Switzerland. Google also reworded several help pages to describe the underlying data as “collected” rather than “shared” in the same update.
- Google Ads Adds Holistic and Growth Modes to Investment Strategy Google Ads split its Recommended Investment Strategy tool into Holistic mode, which moves spend away from underperforming campaigns, and Growth mode, which only adds fresh budget to top performers. Advertisers on Target CPA or Target ROAS bidding should note that accepting a suggested budget change also updates the bid target.
Quick Hits
- Google Business Profiles Tests AI-Set Opening Hours Google is testing a Business Profile feature that lets owners describe their hours in plain text and has its AI convert the description into structured per-day hours for Search and Maps.
- Microsoft Ads Extends Optimization Experiments Past Search Campaigns Microsoft Advertising expanded its Optimization experiments tool to Performance Max, Shopping, and Audience campaigns, running four to eight week tests with a choice of confidence level.