This was the week the AI search narrative shifted under everyone's feet.

On May 15, Google published its first official guide to optimizing for AI Overviews and AI Mode. The headline, in their own words: "Optimizing for generative AI search is optimizing for the search experience, and thus still SEO." No separate ranking layer. No special schema. The myth busting list inside called out, by name, half the tactics the GEO industry has been selling for 18 months.

The same week, Lily Ray published her analysis of 220+ sites running AI content programs. 54% lost 30% or more of their peak organic traffic. 22% lost three-quarters.

Two stories, same direction. The shortcuts are collapsing. The fundamentals are compounding. Here's what we're tracking.

The five tactics Google specifically called unnecessary for its AI features:

The honest read: Google's framing applies to Google's AI features. ChatGPT pulls from Bing and skews toward Wikipedia, Reddit, and editorial. Perplexity is real-time and Reddit-heavy. Cross-platform citation work remains distinct because the surfaces are distinct.

But the consolidation moves matters: Strong SEO is the floor for visibility everywhere. If you've been funding GEO as a parallel discipline with its own headcount and dashboards, May 15 is your permission slip to merge budgets. Brands cited in AI Overviews earn ~120% more organic clicks per impression than uncited brands (Seer, March 2026). That premium is the business case. The tactics that earn it are SEO.

While the industry argued about AI optimization, the most important data point of 2026 sat in plain view: AI Overviews trigger on branded queries only 13.1% of the time, vs 24.9% for non-branded (Ahrefs). Branded search is structurally protected from the zero-click problem eating informational traffic.

We just published research pulling together every credible study on this. The pattern is overwhelming.

The numbers that should change your 2026 budget:

The HubSpot misread to retire: HubSpot lost 60-80% of blog traffic and got framed as the SEO cautionary tale. Revenue grew from $883M (2020) to $2.6B (2025) at ~25% YoY. They rank 3rd in Semrush's AI Visibility Index with 15.4% share of voice, ahead of Salesforce and Adobe. What they lost was informational traffic ("how to make a pie chart"). What they kept branded queries, commercial pages, AI recommendations, was the actual growth engine.

Peer-reviewed evidence closes the loop: marketing investment causally explains 76% of variance in branded search volume (Filippou et al., 2025). Identical search results rated 15-25% higher when associated with a known brand (Jansen et al., 2009). Brand isn't a vanity metric. It's the leading predictor of competitive position.

Teams cutting brand to fund "AI optimization" are funding the wrong side of the equation.

Lily Ray spent months monitoring 220+ websites publicly identified as customers of AI content scaling platforms:

The trajectory has a name now: "Mount AI." Rapid page growth over 6-12 months. Traffic peak within 3-6 months. Then a steep decline that erases the gain and frequently drops below baseline within a year. Most declines happened after the case studies were published which raises the uncomfortable question of whether the case studies themselves are part of the signal Google flags.

The eight content templates showing up repeatedly in declining sites: comparison pages at scale, programmatic "What is X" glossaries, "Best [X] for [Y]" listicles, self-promotional listicles where the publisher names itself #1, competitor-vs-alternatives pages, programmatic location/language scaling, FAQ farms, and off-topic content at scale.

The unconfirmed Google update around January 20, 2026, took out a wave of sites running self-promotional listicles 40-95% drops over Jan-April. Subfolders frequently took the full domain down with them.

Ray's line worth sitting with: "The packaging is new, but the pattern is not." This is the third major cycle the industry has watched in five years. Helpful Content Update (Sept 2023). March 2024 Core Update. Now whatever Google is doing through 2025-2026 to flag AI-shaped fingerprints. The brands still growing don't match the eight templates. They publish less, with clearer first-party signal.

Most teams catch a Mt. AI pattern two quarters too late. Site-wide dashboards hide it. By the time anyone notices, the affected subfolder has dragged the rest of the domain down.

The view we run weekly: Revenue Drop Analysis. A single page comparing two date ranges across every URL, revenue, clicks, impressions, CTR, position, all with directional deltas.

A real example from a maritime publisher client this week: /collections/cornell-maritime-press lost 70.8% of revenue ($239 → $69) despite ranking better (position 12.5 → 6.9, a +45.2% gain). Higher ranking, lower revenue — classic AI Overview cannibalization, invisible in aggregate dashboards. Two more pages in the same view dropped 100% revenue while impressions were up 50-100%+. More visibility, zero conversion.

Aggregate organic traffic is a comforting metric that hides the pages quietly breaking your quarter. You can't fix what you don't see at the URL level.

Want this level of visibility on your pages?

One question to audit: When was the last time you compared revenue AND position deltas at the URL level (not site-wide)? If you only look at site totals, at least one page is silently losing revenue right now.

One quick win: Run your branded search volume against the LLM recommendation order for your top 5 category queries across ChatGPT, Gemini, Perplexity, and Claude. If you're not first, the gap between your volume and the leader's is your roadmap.

One thing to stop doing: Funding "AI content at scale" as a growth strategy. Google's May 15 documentation, Lily Ray's 220-site analysis, and our own SERP study all point at the same outcome: scaled AI content is a 12-month traffic loan with a balloon payment.