Cognac and AI visibility: LVMH, Rémy Cointreau — and the independents left behind

Hennessy appears in 12 out of 15 Benchfolk prompts because LVMH has systematically invested in the structured data infrastructure that LLMs rely on: a rich Wikidata entity with 35+ properties, Wikipedia articles in over 20 languages, and decades of coverage in the premium media sources (Wine Spectator, Decanter, spirits press) that form the backbone of LLM training corpora. This is not simply a function of brand size — it is the result of deliberate structured data strategy that independent houses can replicate.

Yes. The gap between Frapin, Delamain and Pierre Ferrand (4–5 citations) and the top-4 (10–13 citations) is large — but it is driven by structured data gaps, not brand quality. Independent houses can close 30–50% of this gap within 3–6 months by creating a complete Wikidata entity, building English and French Wikipedia articles, and obtaining mentions in Decanter and Wine Advocate. Benchfolk's agent automates the first two levers.

LLM visibility for cognac is driven by four factors in order of impact: (1) Wikidata entity completeness — houses with 20+ properties and multi-language sitelinks are cited 3x more often; (2) Wikipedia presence in English and French — the two dominant languages in cognac training corpora; (3) citations in premium spirits and wine media (Decanter, Wine Advocate, The Whisky Advocate, Club Oenologique); (4) Schema.org Winery and Product markup on the brand's own website. Marketing spend does not directly drive LLM citation — structured data does.

Benchfolk tests three LLMs for every cognac benchmark: OpenAI GPT-4o (the most widely used consumer AI), Perplexity Sonar Pro (the dominant AI search tool for premium and trade buyers), and Anthropic Claude Sonnet (increasingly used by business and trade audiences globally). Each LLM has a different training corpus and retrieval behavior — a brand can rank well on GPT-4o and be invisible on Perplexity. The Benchfolk composite score averages across all three.

GEO (Generative Engine Optimization) for cognac brands is the practice of improving a house's visibility in AI-generated responses from tools like ChatGPT, Perplexity and Claude. When a buyer asks ChatGPT 'What is the best cognac for a gift?' or 'Which VSOP should I buy for a connoisseur?', GEO determines whether your brand appears in the answer. Unlike SEO, which targets Google rankings, GEO is driven by Wikidata entity richness, Wikipedia presence, structured data markup, and citations in premium spirits media.