Category: Uncategorized
-
The Predicted Google Update Window Opened on Schedule. Rank Trackers Spiked. Google Still Isn’t Confirming Anything.
On October 24, 2026, Mozcast, Semrush, Wincher and other rank trackers all spiked at once, and SEOs reported “collapsed” traffic — right inside the late-October window a prediction model flagged on October 13. Google’s Search Status Dashboard still shows nothing. We walk through what’s confirmed versus modeled, and reuse our own September blind test to…
-
The Pre-Registration We Sat On for 5 Days: 24 Queries, 3 Engines, 4 Intent Buckets, and the Two Predictions We Locked Before Collecting Anything
On October 21 we locked two hypotheses before collecting a single data point — H1: Local-intent queries show under 10% cross-engine citation overlap, far below Product/Informational/How-to. H2: last week’s hygiene-score reversal (cited pages scoring LOWER on content quality than uncited ones) replicates across ChatGPT and Perplexity, not just Google AI Mode. We’re publishing the full…
-
AI Mode’s Source Count Fell 74% in Two Months. The Citations That Survived Were Mostly Ranked Top 3.
A third-party tracker measured Google AI Mode’s average sources per answer falling from 16.3 to 4.3 between summer and mid-October 2026 — a 74% drop across 50,527 prompt-page pairs. The same window overlaps a Gemini model swap we already flagged as an untracked variable, and pages outside Google’s top 20 lost citations far more often…
-
ChatGPT and Gemini Agree on Local Business Picks ~30% of the Time. We Predicted Under 10% for Citations and Never Ran the Test.
Searchable tracked 28-44 brands across two 2026 windows and found ChatGPT and Gemini agree on local business recommendations only 26-33% of the time, versus a stable 46% for B2B software. GeoParrot pre-registered a stricter, related prediction on October 21 — under 10% cross-engine citation overlap for Local-intent queries — and didn’t run it for five…
-
Semrush Reported AI Overviews Dropping This Week. Google Had Just Deleted a Label, Not an Answer.
For roughly two days (October 20-22, 2026), some desktop AI Overviews lost their “AI Overview” label while mobile kept it — and Semrush’s tracker read that as a real decline in AI Overview frequency. It was the fifth AI-visibility measurement artifact we’ve logged this quarter, not the fourth time Google actually pulled back. Here’s the…
-
Google’s AI Overview Citation Cards Now Show Text That Isn’t On Your Page. Nobody’s Checking If the Paraphrase Is Accurate.
SEO consultant Brodie Clark spotted Google AI Overview citation cards displaying AI-generated summaries instead of verbatim page text on October 23, 2026, corroborated by Barry Schwartz. Google hasn’t confirmed it’s a rollout, and it doesn’t appear on every query. We separate what this actually changes about GEO from what it doesn’t, and what to check…
-
AI Brand Recommendations Flipped 79% of the Time in 4 Weeks. Two of the Four Assistants Tested Never Touched the Web.
A new Envoyra study found 79% of AI assistant answers to the same buying questions changed their recommended brand at least once over 4 weekly tests. But Claude and Gemini were tested without web search — so some of that “instability” can’t be fresh search results. We separate the two different things the number is…
-
A Model Gives a Google Update 96.8% Odds in 30 Days. It’s Not Confirmed, and Last Time Rankings Moved, Our AI Citations Barely Noticed.
SEO analyst Dan Taylor’s gradient-boosted prediction model puts 96.8% odds on a Google update within 30 days of October 13, 2026, with late October to mid-November as the likely window, theorized to target scaled AI content Panda-style. Google has confirmed nothing. We check what the model actually measures, and what our own two prior update-citation…
-
Researchers Planted a Fake Protocol Into AI Search in a Week. The Rule Against It Doesn’t Start Until November 12.
A Fudan University paper posted October 8 shows 8 of 10 AI search tools cited a made-up “Cactus Gravity Protocol” within seven days of researchers planting it on nine sites those tools already cite often — Grok linked to it in 30 of 32 answers. The predictor wasn’t content quality, it was whether the platform…
-
We Built the Missing Control Group. Hygiene Score Predicted Who AI Does NOT Cite.
Tuesday we pre-registered a control group the “4.2x citation” claim never had: an 8-pillar hygiene score (G) on every page in the Google AI Mode candidate pool, cited and non-cited. The core bet (a small, positive gap) failed — the gap is negative. Cited pages scored 0.55 average G; non-cited pages scored 0.74. High-hygiene pages…