Refreshed May 9, 2026, methodology v1.0
The mapou Panel
A monthly reading of which brands AI assistants name, category by category. In TV brands, AI effectively recommends just 5.5 of 20 tracked brands, and the top two take 46% of all citations.
Every category on the panel
Brand AI names most, and its Panel Score (0 to 100)The Panel Score is a 0 to 100 score per brand. It combines citation rate across 5 AI engines (ChatGPT, Perplexity, Gemini, Claude and Grok) on 20 fixed buyer-intent prompts, weighted across four buying phases, with Wilson 95% confidence intervals. We test at the segment level: banking is not credit cards, hotels are not airlines.
The panel in numbers
500
Brands tracked25
Segments8
Industries5
AI engines20
Prompts per brandDeep studies
Long-form reports. Each puts a category's real buyer questions to ChatGPT, Gemini, Claude and Grok, then measures what each one recommends, why, and whether the facts it quotes are current.
In June 2026, beauty-tech company Novi published an analysis of 10.7 million ChatGPT citations ranking the sources behind AI beauty recommendations. Counting where AI looks is the first step. These studies measure what sits above it: what AI recommends, for what reasons, and what the buyer does next.
Updated October 2026
How AI recommends allergy medicineWhat AI tells allergy shoppers to buy. Brand-to-generic substitution, private-label routing, and how cautious each engine is.
Updated October 2026
How AI recommends credit cardsWhich card AI tells your applicant to open. Acquisition fragmentation, fee accuracy and incumbent concentration across four engines.
Updated October 2026
How AI recommends streamingWhat AI tells your subscribers to keep or cancel. The keep/drop hierarchy, pricing drift and the confusion tax.
Updated October 2026
How AI recommends live TVWhen AI tells a household to cut the cord. Carriage-currency errors and the cut-not-keep default.
Updated October 2026
How AI recommends beautyWhich products AI recommends to which shopper. Persona divergence, legacy-brand residue and grounding gaps.
How much of an AI answer is the search engine behind it? Read Same model, different searcher, the methodology note behind these studies. Across 14,000+ resolved citations, Who does AI read? measures the creator share of AI shopping answers. Shorter findings go in Articles.
Is your brand on the panel?
The mapou Panel is the 500 brands AI search is making decisions about right now.
Either you're on it, or AI is recommending someone else. The mapou Panel covers 25 segments and 5 engines (top 3 refreshed monthly, the full panel quarterly). It is the tracked set behind every report on this page.
Latest run: May 9, 2026. Reports refresh on the first of each month. Methodology v1.0.
Monthly research newsletter
Get the monthly Panel Score report. Fresh data, two minutes to read.
First Tuesday of each month: the new canonical run, new emerging brands and the Panel Score movers per category. Unsubscribe in one click.
We never share email addresses. One unsubscribe link in every send.