The Luxury AI Report™
How AI engines describe the luxury industry
A complete census
What AI engines tell a buyer about Swiss watch houses, recorded across 46 houses and six engines.
The engines have already answered.
The Luxury AI Report is published by ELIA Atelier and measures how AI answer engines describe luxury houses. Edition one, published , records 13,365 answers in the census and 14,067 in total, across 46 Swiss watch houses, nine buyer questions, six engines and three markets.
46 houses · 9 buyer questions · 6 engines · , 07:00 CET
Edition one of The Luxury AI Report publishes on at 07:00 CET, from Genève.
Edition one · · 07:00 CET · Genève
46 houses assessed
- Recommended
- The engine puts the house forward as an answer to the buyer's question, by name, and in terms that favour it.
- Named without endorsement
- The house appears in the answer as a matter of record, listed or mentioned, with nothing said about it that would move a buyer towards the door.
- Named and discounted
- The house is named, and the engine steers the buyer towards paying less for it: pre-owned, or at 40 to 60 per cent below retail.
- Substituted
- The buyer asks about the house, and the answer arrives carrying another house instead.
- Absent
- The question sits inside the house's own territory, and the answer does not contain it.
Every answer receives one verdict, assigned by a rule fixed and published before sampling, never by hand.
Answers recorded
Seven questions to the category and two to each of the forty-six houses give ninety-nine prompts in every market, on every census engine, every day, for fifteen consecutive days. English throughout.
The arithmetic
- Prompts per market, per engine, per day
- 99
- Engines in the census
- 3
- Markets
- 3
- Answers per day
- 99 × 3 × 3 = 891
- Sampling window
- 15 days
- Answers in the census
- 891 × 15 = 13,365
- Beyond the census
- 702
- Recorded in total
- 14,067
Ninety-nine prompts per market, per engine, per day. Seven questions go to the category and two to each of the forty-six houses. The sources an engine cites change from one day to the next with no intervention at all, so a single pass measures the day and not the engine. The Report reads ninety-nine, in every market and on every census engine, every day for fifteen days, and every answer is read against all forty-six houses.
Six. The census runs on ChatGPT, Gemini and Google’s AI Overview. A second panel runs on Perplexity, Google’s AI Mode and Microsoft Copilot.
Claude is available only as an API channel, which returns answers without the search grounding, the commercial units and the market location that the six measured engines carry. An engine measured differently is not an engine measured. Edition one excludes it and records the exclusion.
13,365 answers in the census. 702 further answers beyond it. 14,067 recorded in total.
Edition one reads the United States, the United Kingdom and Singapore. Hong Kong was the intended third market and was withdrawn before sampling began: ChatGPT is not available there, so a house’s answers could not be compared across all engines in that market. Singapore, the second largest watch market in Asia, replaced it. What is added or changed is always announced before sampling, never after the answers are in.
The headline figure is reserved for the edition.
One number decides how this edition of The Luxury AI Report is read. The Report holds it until publication, with the ordering and the method that produced it.
Eligible: any watchmaking house domiciled in Switzerland and exhibiting at Geneva Watch Days 2026, excluding those majority-owned by LVMH, Richemont, Swatch Group, Kering, Chanel or Rolex. No other exclusion. Minority holdings, and ownership by a family, a founder, a management team, a group or an investor, are recorded as attributes and never used to exclude.
How AI describes the great houses · Editions to follow
The Report has its own address
luxuryaireport.com
Edition one publishes there on at 07:00 CET: the ordering, the nine questions, the method and the limits. This page holds the field record until then. ELIA Atelier is the AI visibility practice for the luxury industry, founded by Alexandre Olive and based in Genève, Switzerland.
Visit the ReportQuestions on the Report
What is The Luxury AI Report?
The Luxury AI Report is published by ELIA Atelier and measures how AI answer engines describe luxury houses. Edition one, published 27 August 2026, records 13,365 answers in the census and 14,067 in total, across 46 Swiss watch houses, nine buyer questions, six engines and three markets.
Which AI engines does the Report measure?
The Luxury AI Report measures six answer engines. The census runs on ChatGPT, Gemini and Google’s AI Overview. A second panel runs on Perplexity, Google’s AI Mode and Microsoft Copilot. Claude is excluded: it is available only as an API channel, which returns answers without the search grounding, the commercial units and the market location that the six measured engines carry. The full method publishes with the edition.
Who publishes the Report?
The Luxury AI Report is published by ELIA Atelier, the AI visibility practice for the luxury industry, founded by Alexandre Olive and based in Genève, Switzerland. The practice is a practice of USKALE SA.
How many houses are measured, and which?
Edition one measures forty-six houses: every watchmaking house domiciled in Switzerland and exhibiting at Geneva Watch Days 2026, excluding those majority-owned by LVMH, Richemont, Swatch Group, Kering, Chanel or Rolex. No other exclusion is applied, and no house is excluded by judgement.
When does edition one publish?
Edition one of The Luxury AI Report publishes on 27 August 2026 at 07:00 CET, at luxuryaireport.com, with the ordering, the nine questions, the method and the limits in full.
Every house measured in The Luxury AI Report may see its own record.
Request the letter from ELIA Atelier, without charge.
Edition one, · Edition two,