The state of AI demand intelligence.
Original research on how ChatGPT, Claude, Gemini, and Perplexity respond to life-moment conversations across 17 categories. Updated continuously, refreshed quarterly.
What the data is showing us.
Six structural patterns we keep seeing across categories. Drawn from live monitoring across ChatGPT, Claude, Gemini, and Perplexity through Q1 2026.
The top 3 brands take 78% of AI recommendation share.
Across 17 categories, AI engines consistently funnel recommendation share to a tight top-three. The fourth-named brand averages 9% share. The fifth averages 4%. After that, the curve flattens to noise. Brand-of-record is winner-take-most.
Models agree more than you'd think.
Across the 17 categories we track, ChatGPT, Claude, Gemini, and Perplexity name the same top-three brand in 64% of cases. Translation: when you win the brand-of-record slot on one model, you tend to win it on all of them. The work compounds.
Upstream moments drive 6× more downstream conversion than direct queries.
Brands named in the upstream life-moment conversation (“we just got engaged”) convert downstream traffic six times more reliably than brands named only in direct product queries (“best joint checking account”). Earning the moment compounds into the query.
Brand named on a direct product query. Conversion baseline.
Brand named when AI pivots from a life moment to a product category.
The same 5 sources drive 71% of AI citations in any given category.
AI engines aren't synthesizing answers from across the open web evenly. In every category we've audited, five domains contribute the majority of citations. Identify those five, land in those five, and you win the model's reasoning chain.
11.4 inferred categories fire from a single life moment, on average.
When AI receives a life-moment signal (“we just got engaged”), the downstream inferred-need graph averages 11.4 distinct purchase categories. The highest we've measured is a new-baby trigger that fans out to 23 categories. Every life moment is a multi-category opportunity, not a single funnel.
Brand-of-record changes hands roughly once per category, then defends for years.
When a new brand takes the #1 recommendation slot in a category, our data shows the slot holds for an average of 14 months before the next contested change. The window is narrow and the defense compounds. Acquisition costs in the brand-of-record era are front-loaded.
How we collect the data.
Aethon runs continuous synthetic monitoring across ChatGPT, Claude, Gemini, and Perplexity. We query each engine with a vertical-specific library of life-moment prompts (currently 850+ across 17 categories), capture the full answer text, parse brand mentions and citation chains, and aggregate the data into our moment-and-pathway dataset.
Findings in this report reflect data collected from Oct 2025 through Mar 2026. The full underlying dataset, including per-category breakdowns and the full life-moment library, is available to Aethon customers under the Business and Enterprise tiers.
Want the data for your category specifically?
Book a session and we'll walk you through the moment library, source authority map, and brand-of-record leaderboard for your vertical, live.