• AI visibility

Guide · AI search

How do luxury brands win the high spenders who now ask AI what to buy?

By making sure the press, retailers and reference sources that AI assistants cite tell your story accurately, because most luxury questions put to AI name no brand at all. The biggest spenders adopt AI fastest, so the clients a house values most are the ones meeting it first. This guide covers where AI enters the luxury purchase, what decides which houses appear, and what a luxury ecommerce team can do about it without giving up control of its positioning.

By Underneath Team Guide Updated

The short version

  1. The biggest luxury buyers are the heaviest AI users: in Bain & Company and Comité Colbert’s 2026 study (opens in a new tab), 82% of large luxury purchasers used AI during their latest purchase, as did 54% of US and 64% of Chinese luxury buyers.
  2. The brand is often not chosen yet: about 70% of luxury searches on AI platforms mention no specific brand, and about 75% are about discovery or comparison.
  3. Houses do not control most of what AI reads: official brand sites made up only 10% of cited sources in watch searches and 45% in jewelry (opens in a new tab), yet only 26% of houses are working on external content, against about 60% working on their own sites.
  4. Size does not buy presence: 70% of brands with more than €5 billion in revenue had a smaller share of AI visibility than of revenue, while every brand under €1 billion in the ranking had visibility three to eight times its market share.
  5. The stakes per client are high: Mytheresa’s average order over the last 12 months reached a record €847, and LuxExperience aims for top customers who are about 4% of buyers to make 40% of sales (LuxExperience earnings call, May 2026 (opens in a new tab)).

Who buys luxury online, and what is one client worth?

A small group of high spenders drives most luxury revenue, and each one is worth repeated, high-value orders.

The personal luxury goods market closed 2025 at about €358 billion, down 2% at current exchange rates and flat at constant rates, according to the Altagamma–Bain Worldwide Luxury Market Monitor. The same study describes an active client base that is shrinking, with aspirational buyers pulling back while the wealthiest keep spending. Online is a real but minority channel: LuxExperience, owner of Mytheresa, NET-A-PORTER and MR PORTER, cited Bain and Altagamma’s estimate of a global online luxury market of €75 billion.

The economics are concentrated. Mytheresa’s average order value over the last 12 months rose 12.5% to €847; NET-A-PORTER and MR PORTER combined reached €865. Top customers were 9.7% of Mytheresa’s customers in the quarter, and the company described “the famous 4% making 40% ratio” as the target it wants for NET-A-PORTER and MR PORTER too. We infer that houses selling direct see a similar shape: a few clients who buy across categories, return each season and book private appointments matter more than a large volume of first orders.

That shapes how AI matters here. Losing a single high spender at the moment they form a shortlist is not the loss of one order; it is the loss of a client relationship that the whole luxury model is built to grow.

Where does AI already sit in the luxury purchase?

At the start, before the client has chosen a brand, and increasingly even for purchases completed in a boutique.

Bain and Comité Colbert’s fifth annual Luxe et Technologie report found that, as of April 2026, 54% of US luxury buyers and 64% of Chinese luxury buyers used AI during their most recent luxury purchase, against 27% in France. Usage rises with spending: 82% of high spenders used some form of AI, compared with 28% among lower-spending buyers (opens in a new tab). And 97% of luxury shoppers who had used AI said they intend to use it again.

AI is not only an online habit. Forty-seven percent of shoppers who ultimately bought in a store reported using AI during their journey. The report lists what they used it for: researching products and brands, styling advice, summarizing reviews, comparing prices and finding complementary pieces.

The tools themselves are built for this kind of browsing. OpenAI documents (opens in a new tab) that ChatGPT can show product options with images, details, review summaries and links to buy, and offers virtual try-on for clothing and accessories. Google says (opens in a new tab) AI Mode draws on a Shopping Graph of more than 50 billion product listings and lets shoppers try clothes on a photo of themselves.

Across all US retail, Adobe’s data (opens in a new tab) shows AI traffic converted 42% better than other traffic in March 2026. That figure is not specific to luxury, but it suggests the visitors who arrive from AI answers come ready to buy.

What do luxury clients ask AI assistants?

Mostly open questions about what to buy, what is worth the price and how pieces compare.

Modaes, reporting on the Bain study, gives the clearest real example: “Which luxury handbag to buy for less than 4,000 euros for everyday use” leaves the brand open, while a question about the price of a Chanel 2.55 starts from a choice already made. Most luxury AI questions are the first kind.

The prompts below are illustrative, written by us to show the shape of these questions; they are not observed data:

  • Discovery: “Understated luxury handbags that work for the office and travel.”
  • Value: “Is a heritage watch worth the premium over an independent maker at the same price?”
  • Comparison: “How do these two houses compare on leather quality and repairs?”
  • Authenticity: “Where can I buy this bag new from an authorized seller online?”
  • Gifting: “A fine jewelry gift under $5,000 that will still feel special in ten years.”
  • Styling: “What to pair with a camel coat for a winter wedding.”

Each of these is a moment where an assistant decides which names to put in front of a client who has not yet committed. That is the battleground Bain describes.

How does an AI answer turn into luxury revenue?

Through the shortlist: the client takes the names an assistant gave to a website, retailer or boutique.

The path is longer and more varied than in mass ecommerce, so it is worth tracing step by step:

  1. The unbranded question. A client asks about a category, a budget or an occasion. The assistant names a handful of houses and pieces.
  2. The shortlist. Modaes puts it directly: the list of options the customer carries with them “may have been formed earlier,” before any visit to a brand’s site, a retailer or a store.
  3. Where the purchase happens. The client may buy on the house’s own site, on a multibrand retailer such as Mytheresa, or in a boutique after booking an appointment. The in-store figure above shows that much of the influence will never appear as an AI referral in analytics.
  4. The relationship. A first purchase at full price can lead to repeat orders, private appointments and, eventually, top-client status. This is where the revenue concentrates.

A reasonable expectation, which we infer from these figures rather than measure, is that luxury AI influence will be undercounted by traffic data and should be judged against new-client acquisition and full-price sales instead.

What decides which luxury houses an AI assistant names?

Platforms document some product signals; the luxury findings are observed in Bain’s study; the rest is our inference.

Documented by the platforms. OpenAI states that ChatGPT considers structured product data from first-party and third-party providers, such as price and description, plus other third-party content. When it lists sellers for a product, “merchants are ranked based on factors like availability, price, quality, and whether they are the maker or primary seller.” That last factor matters to houses that sell direct. OpenAI also says review summaries are “based on reviews from public websites” and are not verified by OpenAI.

Observed in the Bain and Comité Colbert study. Third-party websites supply most citations in AI answers to unbranded luxury questions. Official sites were 10% of cited sources for watches and 45% for jewelry. Jewelers will find more detail in how jewelry brands get recommended by AI. Revenue was a poor guide to visibility: smaller houses were far more visible than their size would suggest, while most of the largest were less visible than their market share. Bain’s researchers warn that brands “risk losing potential customers and narrative control.”

Our inference. For luxury, the sources that seem most likely to shape an answer are those a client would also trust: fashion and watch media, auction and resale references, authorized retailers, and long-form reviews by collectors. Our article on whether AI assistants favor big brands shows that in controlled tests a familiar name wins when everything else is equal, but a clearly stated advantage can beat it. Heritage, craftsmanship and service only help if they are written down somewhere an assistant can read and quote.

Trust factors specific to luxury. Authenticity and authorized selling, price consistency across markets, repair and after-sales service, craftsmanship and provenance, and a brand staying true to its values. Savanta’s MillionaireVue data found that 52% of high-net-worth shoppers prefer deeper connections with brands, against 42% who prefer faster, more transactional engagement, which is a reminder that AI visibility has to serve the relationship, not replace it.

What does a luxury house risk by staying absent from AI answers?

Clients, and control of how the house is described, including whether buyers are sent to authorized sellers.

The first risk is losing the client at the shortlist. With seven in ten luxury AI searches naming no brand, a house that is absent does not get the chance to be compared. The second is narrative. If most citations come from third parties, an assistant’s description of a house’s quality, price and heritage is assembled from what others wrote.

The third risk is specific to luxury: counterfeits. The OECD and the European Union Intellectual Property Office estimate that trade in fake goods reached USD 467 billion, or 2.3% of total imports (opens in a new tab), with counterfeit imports into the European Union worth EUR 117 billion, and they note that counterfeiters use online sales platforms. Savanta warns that AI platforms “can easily recommend lookalikes if protections aren’t in place.” We have not seen a study measuring how often assistants send luxury shoppers to unauthorized sellers, so treat this as a risk to check, not a measured rate. For the general case, see our article on fake reviews and fake brands in AI recommendations.

Most houses are not yet acting on this. Only 48% of groups and houses regularly monitor their performance in AI search, and only 10% consider their current position strong. Nearly a quarter of luxury companies now rank AI among their top three priorities, up from 5% in 2024, but Bain says most of that work is in the back office rather than with clients.

How does GEO work for a luxury brand without diluting it?

It shapes the facts and sources assistants rely on, in the house’s own voice; it cannot guarantee a mention.

Generative engine optimization (GEO) for luxury is closer to press relations and brand protection than to performance marketing. It usually covers six areas:

  1. External coverage where AI looks. Since official sites are a minority of citations, work with the fashion, watch and jewelry press, collector communities and reference sites to keep accurate, current descriptions of the house and its key lines. Our guide to building authority for AI search explains how to choose those sources.
  2. Authorized-seller clarity. Publish who sells your products, in which markets, and how to verify authenticity, so an assistant that is asked where to buy has a clear answer and the house stays the “maker or primary seller” OpenAI describes.
  3. Product facts written plainly. Materials, origin, craft techniques, sizes, care and repair policies, stated as facts on product pages and in feeds, not only as imagery. Our article on what drives AI product recommendations shows why concrete details matter.
  4. Consistent pricing and naming across markets. An assistant that quotes last season’s price or misnames a line undercuts the house; our guide to correcting wrong brand information in AI answers covers the repair.
  5. Multibrand partner alignment. Retailers such as Mytheresa describe your pieces too; make sure their product copy matches yours.
  6. Market-by-market monitoring. AI use differs sharply between the US, China and France, so track the unbranded discovery and comparison questions clients ask in each market, across ChatGPT, Gemini, Perplexity, Copilot and Google’s AI features. Our guide to GEO across languages covers the multilingual side.

None of this asks a house to discount, chase volume or change its tone. It asks the house to make its own story easy to verify.

What can’t a luxury house learn from current AI research?

How often an AI answer changes which house a client buys from.

The strongest figures here come from one consultancy study, reported through trade media; we could not access Bain’s full report directly. The adoption numbers are survey answers, and “used AI during the purchase” covers anything from a quick question to a long session. The citation shares cover watches and jewelry; other categories may differ. Mytheresa’s figures describe one multibrand retailer’s clients, not every house. Adobe’s conversion data covers all US retail. No public source yet links AI visibility to luxury client acquisition or lifetime value, and no platform documents how it chooses between luxury houses for an open question.

Where should a luxury ecommerce team start?

Start by finding out what assistants say about your house when clients ask open questions in your key markets.

A useful first step is an audit of the unbranded discovery, comparison, gifting and “where to buy” questions your clients ask, across the main assistants and in your priority markets, with the sources each answer cites. That shows where your narrative is being written by others and where buyers may be sent to the wrong sellers. To have us run that audit with you and plan the work, tied to new-client acquisition and full-price sales, arrange a private conversation with our team. Our generative engine optimization service page outlines how that work proceeds for a house, from press and authorized-seller facts to monitoring each market, in its own voice.

Frequently asked questions

Do wealthy luxury clients really use AI to shop?

Yes, more than other buyers. In Bain and Comité Colbert’s study, 82% of high spenders used AI during their latest luxury purchase, against 28% of lower spenders.

Can a luxury house control what AI says about it?

Not directly. It can make accurate facts easy to find on its own site and in the third-party sources assistants cite, which is where most citations come from.

Will AI shopping push luxury toward discounting?

Not necessarily. Assistants weigh price when a budget is given, but clients also ask about quality, craftsmanship and value. Clear facts about those let a house compete without discounting.

Should luxury brands block AI assistants from their websites?

Blocking may keep a house’s own pages out of answers while third-party descriptions remain. Our crawler-blocking study covers the trade-offs; decide deliberately rather than by default.

Sources

Free strategy call

Some questions are easier to answer about your own business.

Bring the one that matters most. On a free 30-minute call we’ll take a first look at it and send you a short written read afterward.

Ask about my businessHow we work