Problem
Luxury shoppers increasingly ask AI assistants what to buy. 82% of high-spending luxury clients used AI in their last purchase, and 70% of luxury questions asked to AI name no brand at all (Bain & Company × Comité Colbert, 2026). The AI replies with a shortlist of two or three houses. Brands that are not on it never enter the decision.
Fashion and jewelry houses have no way to see or manage this. They can't tell whether AI recommends them, which of their pieces it suggests, or whether it describes their price, materials and heritage correctly. They also don't know which sources drive those answers. AI builds its shortlist mostly from magazines, creators, newsletters and forums, not from brand websites (official sites are only 10% of sources AI cites for watches). So houses keep optimizing their own sites and paying for creator partnerships without knowing if any of it makes AI recommend them.
Solution
We run real, unbranded shopper questions across the main AI assistants, repeatedly and by market, and show each house how often it is recommended, with which pieces (including image search), which sources drive those answers, and whether each partnership moved the needle.
Business model
B2B SaaS with annual contracts, entered through a paid diagnostic.
Diagnostic (one-off): €4,900, a full AI-visibility audit across 3 markets, credited if the client subscribes. Founding houses pay €2,500.
Subscription (monthly fee, billed annually): Atelier for independents at €1,200/month; Maison for mid-size houses, our focus, at €3,500/month; Groupe for multi-brand groups from €12,000/month.
Agency partners: PR and digital agencies resell under their own brand at a 20–30% discount.
The monthly index and mini-reports are free and act as our acquisition channel.
Expected average ticket: about €30,000 per client per year (≈US$34,000), weighted toward the Maison plan. Target gross margin above 70%; the main variable cost is AI model queries. All prices are hypotheses being tested in customer interviews.
Market
TAM: the global GEO software market was about US$390M in 2025 and is projected to reach US$4.25B by 2032, a 41% annual growth rate (MarketsandMarkets). Other estimates are higher (US$1.09B in 2026, Market Decipher).
SAM: GEO for fashion, leather goods, jewelry and watches, a €358B personal luxury market (Bain-Altagamma, 2025). Bottom-up, about 8,600 brands (100 groups, 2,500 mid-size houses, 6,000 independents) at €14k–150k per year each gives about €204M (≈US$232M) per year at full adoption.
SOM: our beachhead is mid-size houses (€20M–€1B revenue) in Paris, Milan, London and New York: 40 clients and €1.2M ARR in 18 months, about 200 clients and €7M ARR in 36 months.
Competitors
Fashion-specialized: Launchmetrics (the fashion PR standard, owner of the MIV metric) launched an AI Visibility score in 2026 that tracks brand presence in ChatGPT, Claude, Gemini and Perplexity and links it to press coverage. Meikai runs enterprise GEO in Europe and produced the AI-visibility analysis for Bain & Comité Colbert's 2026 luxury report.
Competitive differentiation
Product-level and visual, not just brand-level. Competitors report whether a brand is mentioned. We show which bag, shoe or jewel AI recommends, at what price and with what description, including in photo search.
Proof of ROI per source. Houses already pay for editorials, newsletters and creators (like Balenciaga's Substack deal). Our before/after tests with control questions show whether each partnership increased AI recommendations.
Built for luxury, priced for mid-size houses. Luxury vocabulary, price bands and multilingual prompts. Enterprise tools are priced for the Fortune 500; cheaper ones are generic.
Distribution before product. Our monthly Luxury AI Visibility Index builds an audience of brand, PR and e-commerce leaders before we sell, lowering acquisition cost.
Entry barrier
Historical data that can't be backfilled. AI answers change every month and are not archived. Each month we record what AI said about luxury houses, pieces and sources. A new entrant can't recreate that history, and it is what makes trends and before/after proof possible.
Partnership outcome data. Every test we run links a specific editorial, newsletter or creator to its effect on AI recommendations. That dataset gets more accurate with each client and cannot be bought.
Becoming the industry benchmark. Our public Luxury AI Visibility Index aims to be the reference metric the sector quotes, as Launchmetrics' MIV became for PR. A standard is hard to displace.
Switching costs. Once a house tracks its baseline, KPIs and board reports with Citelle, moving loses its history.
Traction
Not yet in revenue or users: Citelle is at the validation stage, started in September 2026.
Done so far: market and competitor research (Bain-Comité Colbert, Adobe and funding data on 9 GEO players), a validation plan with 6 testable hypotheses, a target list of 33 fashion and jewelry houses, brand identity and website.
Next 90 days, with clear go/no-go criteria: 25 interviews with brand, PR and e-commerce leaders; first edition of our public Luxury AI Visibility Index; 3 paid diagnostics (€2,500 founder price); 500 industry readers of the index.