Who this is for

This piece is written for founders, brand directors and marketing leads at luxury and high-value businesses who want to understand why an AI assistant names some brands and not others – and what can be changed about their own website in response.

It is not a shopping guide and it does not recommend brands. If you arrived looking for which luxury house makes the best leather goods, this will not help you. If you run one and want to be in the answer, it will.

How an AI assistant actually picks which brands to name

When someone asks an assistant for a recommendation, it does not consult a ranking. It breaks the question into several narrower ones, retrieves passages that appear to answer each, and composes a reply from the sources it can most confidently attribute. Being retrievable at the passage level therefore matters more than being ranked at the page level.

Two consequences follow, and both are uncomfortable for brands that have invested in conventional SEO. The first is that the set of sources an assistant cites overlaps far less with the top of Google's results than it used to – one analysis puts that overlap below 20%, down from around 70%. The second is that being named is largely a function of corroboration: a brand described consistently across its own site and several independent sources is one an assistant can commit to, and a brand that exists only on its own domain is not.

AI discovery begins with a website that can be understood

The practical foundations are familiar: accessible pages, semantic HTML, stable URLs, helpful copy, descriptive titles, internal links and technical reliability. New discovery interfaces do not remove the need for good web fundamentals.

For a luxury brand, specificity matters because the language of the category is often overused. Words such as exceptional, bespoke and world-class reveal very little unless they are supported by facts, process, place, authorship or a distinct point of view.

Make the entity unambiguous

A site should make it easy to answer basic questions: What is the organisation? Where does it operate? What does it offer? Who is behind it? Which places, properties or services belong to it? How can someone verify or contact it?

Consistency across the studio page, service pages, contact details and structured data reduces ambiguity. Schema supports the visible page; it does not replace it.

Build pages around real questions and decisions

A useful page has a reason to exist beyond capturing a keyword. It should answer a coherent question for a real audience, then provide enough context to be trusted.

For hospitality this could mean a substantive private-hire page. For branded residences it could mean a clear explanation of ownership, service and availability. For a design practice it could mean a detailed account of process, project fit and collaboration.

Original value is stronger than content volume

Publishing dozens of lightly varied pages creates noise. A smaller collection of precise guides, project narratives, field notes, definitions and answers can establish far more useful evidence.

The strongest material comes from the knowledge already inside the business: questions from buyers, operational judgement, place knowledge, design decisions, founder perspective and documented process.

Crawler policy is a real choice

Public sites can separately consider search discovery and model training. OpenAI identifies OAI-SearchBot for ChatGPT search and GPTBot for potential training use. A brand can allow the former while setting its own policy for the latter.

Private previews should remain noindex and inaccessible to crawlers until the content, legal pages, domain and tracking choices are approved.

Why this is worth the effort now

The traffic is small and the visitors are not. Referrals from ChatGPT to business websites grew roughly fourfold in the year to mid-2026, and independent measurement consistently finds that visitors arriving from an assistant convert at several times the rate of ordinary organic search – unsurprisingly, since they arrive having already been given a reason to come.

The gap worth exploiting is the measurement one. Around 43% of marketers say they are working on AI visibility; only about 14% track whether they are actually being cited. Most brands are guessing.

Measure evidence, not promises

No studio can guarantee inclusion or recommendation in an AI answer, and anyone who does is selling something they cannot deliver. Monitor the evidence you can observe: crawler access, indexed pages, organic landing-page performance, referrals from AI products where available, branded query movement and the contribution of useful content to qualified enquiries.

One cheap and underused signal: watch your Search Console query report for long, conversational strings that read like somebody's prompt rather than a search. Those are AI fan-out queries, and they tell you which questions your pages are being tested against – often before any ranking movement shows up at all.

Treat the result as an ongoing publishing and measurement discipline rather than a one-time technical switch.