The Luxury AI Report
alexandre@eliaatelier.ch Genève · Switzerland
The Findings Edition one · Swiss watchmaking · 46 independent houses

What the engines saywhen a buyer asks


14,256 recorded answers · 10 to 25 August 2026

Alexandre Olive · Founder & Publisher
The instrument

9 QUESTIONS A BUYER TYPES, PUT TO 3 AI, IN 3 MARKETS EVERY DAY FOR 16 DAYS.

7 questions went to the category and 2 to each house by name. Every answer was archived on the day it was recorded, with its engine, its market and its date, and read against all 46 houses.

46independent Swiss houses, every eligible house exhibiting at Geneva Watch Days 2026
14,256answers in the census, of which 13,248 concern a single named house
3engines recorded every day: ChatGPT, Gemini, Google AI Overview
3markets, the United States, the United Kingdom and Singapore

Every recorded answer receives one of five verdicts: 1 recommended, 2 named without endorsement, 3 named and discounted, 4 substituted, 5 absent.
The verdicts were named on 20 August 2026, before any figure was computed. No verdict is assigned by hand.

The roster
  • Akhor
  • Alpina
  • Alto
  • Amida
  • Ba111od
  • Beauregard
  • Beda'a
  • Bianchet
  • Bimbu
  • Breitling
  • Byrne
  • Claude Meylan
  • Corum
  • Czapek
  • Cédric Johner
  • David Candaux
  • De Bethune
  • DOXA
  • Emmanuel Bouchet
  • Ferdinand Berthoud
  • Frédérique Constant
  • Furlan Marri
  • Gallet
  • Gerald Charles
  • Girard-Perregaux
  • Greubel Forsey
  • Hautlence
  • Kerbedanz
  • Keris
  • Krayon
  • L. Leroy
  • Laurent Ferrier
  • Lederer
  • Louis Erard
  • Louis Moinet
  • Maurice Lacroix
  • Mauron Musy
  • MB&F
  • Oris
  • Perrelet
  • Renaud Tixier
  • Singer Reimagined
  • Speake-Marin
  • Stollenwurm
  • Universal Genève
  • Urwerk

46 houses. Every eligible Swiss house exhibiting at Geneva Watch Days 2026. Houses under majority group control and non-Swiss houses are excluded. None was invited and none was set aside by judgement.

0 · Where the buyer now starts

Your next client asked a machine about you before he asked anyone else.

82 of every 100 high-value buyers used an AI on their last luxury purchase.

82% of high-value buyers used an AI assistant on their last luxury purchase In 70% of those conversations the buyer named no brand at all

Comité Colbert and Bain & Company, June 2026. 35 houses and their clients.

440,000 Americans became millionaires in one year through the AI boom.

440,000new American millionaires in 2025, on the back of AI equities
17%of Swiss watch exports go to the United States

The people who made their money with AI are the people who will ask an AI what to buy.

Reuters and BCG, 2025. Fédération de l'industrie horlogère suisse, 2025.

2 engines carry 4 fifths of every AI conversation on earth.

81.8%of AI assistant use sits on 2 engines10.4 billion visits a month
ChatGPT53.9%
Gemini27.9%
Claude9.2%
DeepSeek4.1%
Grok2.4%
Perplexity1.3%
Copilot1.3%

Similarweb, May 2026, via Momentic. Google's AI Overview is absent from this ranking because it is not an assistant a buyer opens. It is the answer above the results.

The machine reads the page and keeps the reader.

Visits from Google, in 1 year−34%
Small publishers, in 2 years−60%
Clicks under an AI answer−37%
Traffic the engines send back<1%

The link is no longer the channel. The citation is.

Chartbeat via Axios, March 2026. SEER Interactive, April 2026.

Most worrying finding
1 · What the engines do with price

Nobody asked for a discount. The engines offer one anyway.

01

A buyer with $10,000 asks which Swiss watch to buy. The engine tells him to spend less.

82.5% of those answers steer him to the second-hand market or tell him to wait 475 of 576 answers · 4 price levels · 3 markets · 16 days
82 answers in 100the other 18
02

He asks what the watch will be worth. He is told to buy it used.

58.0% of answers about value send the buyer to the pre-owned market 3,842 of 6,624 answers about a named house
58 answers in 100the other 42
03

For most of the 46 houses, that is the usual answer.

32 of 46 houses whose most frequent answer on value is the pre-owned steer 144 answers per house
32 houses14 houses
04

Across every answer about a named house, the discount is the most common of the five verdicts.

38.6% named and discounted 5,113 of 13,248 answers
Named and discounted38.6%
Named without endorsement34.7%
Recommended18.7%
Substituted0.3%
05

Take the 288 answers about any one house. For not one of the 46 is “recommended” the most frequent answer.

0 of 46houses whose most frequent answer is a recommendation288 answers per house, across both questions that name it
Most often discounted24 houses
Most often described with no view21 houses
Tied between the two1 house
Most often recommended0 houses
06

The same watch, priced twice by the same engine.

One house in nine is quoted at less than half its own price. Five are quoted above it.

38 of 45houses the engines quote below their own retail price9 of them below half. 5 houses are quoted above their retail price
−79%The same figure in both contexts+38%
9houses · quoted below half their retail figure
29houses · below, by less than half
2houses · the same figure in both contexts
5houses · quoted above their retail figure

Measured. Not published until two people have read it. The extraction rests on 588 answers, 4.4% of the set, and it mistakes a servicing cost for a price. Two readers are working through a sample of 200 separately. This chart publishes with their level of agreement printed beside it, or it does not publish.

2 · What the engines read

LESS THAN 10% OF THE ANSWERS FROM AI COMES FROM THE HOUSES’ WEBSITES.

07

Of everything the engines read about a house, this much came from the house itself.

7.3% of what the engines read about a house came from its own domains 9,314 of 128,173 citations in answers about a named house
7.3% the house’s own domains 92.7% everyone else

By house the figure runs from 2.7% to 14.8%. Not one of the 46 reaches a fifth.

Marketplaces, dealers, jewellers26.3%
Enthusiast media and blogs23.0%
The long tail, 6,435 sites22.2%
The houses’ own domains7.2%
Forums4.7%
Video platforms3.1%
Auction houses2.6%
The trade press1.6%

The individual sites, named, with the number of times each was read and the answers they fed, are in the source file.

08

In more than a third of answers, the engine shows its sources and the house's own site is not among them.

36.3%of answers about a house cite pages, and not one of them belongs to the houseThe engine had somewhere to look. It looked. It did not look there
36 answers in 100the other 64
09

The most cited source in a study about new watches is a second-hand marketplace.

5.4% of everything the engines read comes from one marketplace, across its 44 country domains Ahead of YouTube, ahead of every editorial title
One pre-owned marketplace, 44 domains5.40%
The largest video platform2.97%
The largest forum2.63%
The largest editorial title2.09%
10

When it recommends, it has been reading the press. When it discounts, it has been reading the market.

Specialist press and blogs31.7% → 20.8%
Marketplaces and dealers18.8% → 30.1%
The long tail, 6,435 sites24.9% → 21.5%
The house's own site8.7% → 6.8%

The press falls 10.9 points. The marketplaces rise 11.3. They change places.

Share of everything cited, in answers that recommend against answers that discount. 7,804 sites in all. The largest single site is 3% of what the engines read, and past the first 30 there is nothing to buy a place on.

11

The engines read the enthusiasts. They do not read the industry’s own press.

Two kinds of watch writing, and the machines treat them as if only one existed.

1.6%of all citations come from the press written for the industryAgainst 23.0% for the media written for enthusiasts
Enthusiast media and blogs23.0%
The trade press1.6%

A house can be covered by every trade title in Switzerland and remain invisible to the machine.

3 · Who fills the answer

When a buyer names a sum, the independents are not on the list.

12

576 answers to a buyer with a budget. A house of the study comes first once.

1 of 576 first places held by a house of the study 4 price levels, 3 markets, 16 days
Rolex137
Omega137
Patek Philippe114
Tudor97
Vacheron Constantin57
Longines19
A house of the study1

Asked with no budget at all, Tissot comes first in 65.3% of answers. No house of the 46 is ever named first.

13

The short list changes with the budget, and only 1 independent is ever on it.

$5,000 · Tudor, Omega, Longines1 of the 46
$10,000 · Omega, Rolex, Tudornone
$30,000 · Rolex, Vacheron, Pateknone
$80,000 · Patek, Vacheron, Rolexnone

One independent house holds a first place at the highest level. It is not in the study and it does not exhibit at the fair.

14

36 of the 46 are never named at all when a buyer states a budget.

36 of 46 never appear in the 576 budget answers, in any position Not first, not fifth, not once
36 houses, absent10 houses, named at least once
15

Asked which Swiss brands to consider, with no budget named, the engines lead with Tissot.

Rolex is in almost every one of those answers. It is simply not the name the engine puts first.

65.3%of answers name Tissot before any other brandRolex first in 27.1% · no house of the 46, ever
Tissot, named first65.3%
Rolex, named first27.1%
A house of the study, named first0%

Named first and named at all are two different things. When the buyer has not said what he can spend, the engine opens with the affordable name.

16

37 of the 46 are named nowhere except in the answers that are about them.

37 of 46never appear outside their own questions9 houses have any presence beyond them
37 houses, inside their own question only9 houses, named elsewhere

A house exists inside its own question and nowhere else.

17

Rolex is inside 56% of the answers about somebody else.

ChatGPT79.8%
Gemini59.4%
Google AI Overview28.6%

1 of the 4 largest names appears in 62.1% of answers about an independent house.

18

1 answer in 4 carries a cue that sends the buyer to a different house.

25.0%of answers about a named houseThe quietest of the 5 verdicts, and the most expensive
25 answers in 100 carry a substitution cueThe other 75
4 · What else the machine does

3 behaviours nobody in the industry had counted.

19

1 answer in 4 names the company that made the movement, not the house that signs the watch.

28.2%of answers name at least 1 movement supplier33.6% on the question about value. For 1 house, 86.5%
Sellita12.9%
ETA12.2%
Soprod4.3%
Valjoux3.3%
La Joux-Perret2.0%
Vaucher1.7%

For a house that has always said who makes its movements, the engine repeats what the house already publishes. For a house that has let the market believe it is vertically integrated, the engine now answers that question for the buyer, in front of him, without being asked.

20

1 engine sells beside its own answer, and only in 2 markets.

ChatGPT · United States, budget questions64.1%
ChatGPT · United States52.9%
ChatGPT · United Kingdom12.3%
ChatGPT · Singapore0%

Gemini and Google AI Overview carry no advertising anywhere.

21

The engine is not always sure which house it is describing.

12.1%of answers open by working out which house is meantOn 1 engine, 32.6%. On the other 2, under 2%
12 answers in 100 open in doubtThe other 88

One house of the 46 shares its name with a small car sold in Asia. The engine asks the buyer which one he means, then describes the car. Another shares its name with a sewing machine, another with an arms manufacturer.

22

And when it has decided, it spells the name wrong.

16 of the 46 houses have their own name written incorrectly by an engine. Nobody told them.

596 answers in which the engine writes the house name wrong, or drops part of it 4.5% of all answers about a named house
House A86.8%
House B59.0%
House C25.0%
House D16.7%
House E10.1%
The other 11 affectedunder 4%

Share of a house's own 288 answers. The houses are not identified here. Each is told its own figure.

One house loses the initial in front of its name in 248 of its 288 answers, so the machine calls it by half its own name. Another is written four different ways, two of which belong to a bakery group and a rum. A third is spelled as a company that makes non-lethal weapons, in 46 answers. A house that cannot be spelled cannot be found.

What we refuse to claim

What this record does not say.

No causality. The record shows a movement. Attributing that movement to any action needs a controlled comparison, which does not exist before edition two.

No result from a single observation. Nothing seen once is presented as an engine's settled position.

No ranking of the 46 in any public document, under any name.

No engine reports which version of itself answered, and the model behind an answer can be changed by its publisher without notice.

One window, 3 markets, the English language.

The study does not measure what the buyer did next, and nothing in it says a watch was sold.

A figure that could not be computed honestly on the closed window is absent, not estimated.

The conclusion

A new paradigm, and the first loss of control.

AI engines are now where a wealthy buyer begins. That much is measured elsewhere and is not in dispute.

What this study measures is what happens next: the way an engine assembles an answer does not favour the house. It reads the market before it reads the maison, it reaches for the largest names when asked about the smallest, and it offers a discount to a buyer who did not ask for one.

For 200 years a house decided how it was described. It chose the words, the images, the shop, the salesperson and the moment. That is over. The description now happens somewhere else, before the house is involved, and for the first time the house is not in the room when it is written.

With the loss of control of the narrative of the brand, AI visibility becomes of strategic importance and will influence the goodwill of the house.

Questions

If a figure here concerns your house, or if you want to discuss what to do about it, write to Alexandre Olive.

alexandre@eliaatelier.ch

The record cannot be recreated. The 16 days of August 2026 will not be run again.

No house of the 46 is named in this page.

The full method, the prompt set, the definition cards and the intervals are published at The Method.


The Luxury AI Report, Edition One: Swiss Watchmaking. USKALE SA, Genève, 31 August 2026.

Published figures may be quoted with attribution. The underlying record, the answers and the dataset are the property of USKALE SA and are not licensed for redistribution.