The Luxury Web Index 2026

Before your next client ever types a query into Google, there is a growing chance they have already asked an AI assistant: which galleries should I visit in Chelsea, which studio should design our house, who charters yachts out of Monaco? The assistants answer from what they can read. So this study asks a simple question the luxury world has not yet asked itself: when a machine calls on the finest websites in the world, what does it actually find?

We measured 105 of them — the websites of the leading art galleries, interior design studios, architecture firms, photographers, maisons, real estate brokerages and yacht houses. Every site received the same visits: a neutral crawler requesting its homepage, robots.txt, sitemap and llms.txt, verified where necessary through a real Chromium browser. Every site is scored 0–100 by a deterministic formula across four pillars, published in full below, from public signals anyone can check. The raw data is downloadable. No panels, no taste, no opinions — just what the machines are given to read.

The result, in one sentence: the most beautiful websites in the world are, to machines, very often nearly blank. More than a third of measurable sites hand a crawler fewer than 1,500 characters of text. Nearly half carry no structured data of any kind. And twelve of the most famous names in luxury refuse machine callers altogether.

Yet in every category there are leaders proving the trade-off is false — that a site can be exquisite to look at and perfectly legible to software. They are not who you might expect.

The headline findings

Twelve of luxury's most famous websites are sealed to machines — Louis Vuitton, Chanel, Hermès, Dior, Gucci, Cartier, Rolex, Tiffany & Co., Bottega Veneta, Saint Laurent and Loro Piana — plus real estate brokerage The Agency — refused every measurement client we sent: a neutral crawler, a fully browser-headed request, and an automated Chromium browser. Whatever the intent, the practical effect is that software browsing on a client's behalf is turned away at the door.

37% of measurable sites give a crawler under 1,500 characters to read — Either everything is rendered by JavaScript — which most AI crawlers do not execute — or neutral clients are refused outright. OMA's homepage serves a crawler 3 characters of readable text; Perrotin serves 8; Sotheby's International Realty serves a JavaScript shell that reads as blank.

48% carry no structured data at all — No schema.org markup of any kind — the vocabulary search engines and AI systems use to understand who a business is, what it offers and where it stands. The sites that machines recommend most confidently are the ones that describe themselves in it.

Architecture is the least machine-legible category in luxury — Median AI Visibility just 19 of 40. Seven of fifteen world-renowned firms serve under 300 characters of readable text; Foster + Partners and Zaha Hadid Architects additionally block four major AI crawlers each in robots.txt, and Olson Kundig blocks all six we tested.

Yachting quietly sets the standard — Category median 80 of 100 — the study's best. Moran Yacht & Ship posts the highest score of all 105 sites (98/100), and four of the overall top ten are yacht brokerages. The businesses that live or die on inbound deal flow have built the most machine-legible sites in luxury.

Only 10% have adopted llms.txt — The emerging plain-text standard that tells AI assistants what a site is and where its important pages live — nine sites of ninety-two. Early adopters are capturing a channel their peers have not noticed exists.

A quarter of the luxury web is slow — 25% of measurable sites take longer than 1.5 seconds to serve their first byte, and 34% publish no usable sitemap — table stakes of being found, unattended.

How the Index works

Every site earns a deterministic score out of 100 across four pillars: AI Visibility (40 points — can AI assistants and crawlers access, read and understand the site: crawler permissions, machine-readable text, structured data, llms.txt), Performance (20 — server response, document weight, modern image delivery), Search Foundations (20 — titles, descriptions, canonicals, sitemaps, social metadata) and Craft & Accessibility (20 — alt text, mobile viewport, language declaration, transport security). Every point traces to a public, verifiable signal measured on 2026-07-08; the full formula is published in the methodology below. Tiers: Vanguard (80+), Refined (65–79), Emerging (50–64), Veiled (below 50).

The 2026 leaderboard — top 25 overall

† Site refuses neutral HTTP clients (security wall); on-page signals were measured through a real Chromium browser and its machine-readable text is scored 0 — see methodology.

What the leaders do differently

Only two sites in the study earned a perfect 40/40 for AI Visibility: Moran Yacht & Ship and interior designer Beata Heuman — a Fort Lauderdale brokerage and a London studio famous for its painterly, decorative interiors. They share the same unglamorous recipe: welcome crawlers in robots.txt, serve real text in the initial HTML rather than building the page in JavaScript, describe the business in schema.org structured data, and publish an llms.txt. None of it is visible on the surface of the site. All of it decides whether a machine can repeat your name to a client.

The pattern holds down the leaderboard. Denison Yachting (93), Peter Lik (92) and Y.CO (91) pair fast, server-rendered pages with rich structured data. What the leaders prove is that machine legibility is not a design constraint: Beata Heuman's site is among the most visually exuberant in this study, and one of the most legible. The trade-off between beauty and being read is a myth.

The sealed twelve

These sites refused all three measurement methods — a neutral crawler, a browser-headed request, and an automated Chromium browser. We publish no score for them rather than guess. Their robots.txt files mostly welcome AI crawlers on paper; in practice, their security layers refuse any caller that is not demonstrably human. As assistants and agentic browsers do more of an affluent client's looking, a wall against machines becomes a wall against the clients who sent them.

Not scoreable by their own choice: Louis Vuitton, Bottega Veneta, Loro Piana, Tiffany & Co., Chanel, Saint Laurent, Hermès, Rolex, Dior, The Agency, Gucci, Cartier.

Art Galleries

The mega-galleries split cleanly. David Zwirner (79) and Pace lead with server-rendered pages and structured data; Perrotin's homepage offers a crawler eight characters. A collector's assistant asking about artists or openings can only relay what it can read. Median AI Visibility: 25/40. See our web design for art galleries. See our AI services for galleries.

Interior Design

Beata Heuman's 88 — with the study's only non-yachting perfect AI pillar — shows a studio site can be painterly and perfectly legible at once. At the other end, several of the most celebrated minimalist studios extend their minimalism to what machines may read. See our marketing for interior designers. See our interior design websites.

Architecture

The least machine-legible category in the study, and the one with the starkest gap between fame and findability: seven world-renowned firms serve under 300 characters of readable text, and three of the biggest names block AI crawlers by policy on top of it. SOM (78) leads by simply serving real HTML. See our marketing for architects. See our architecture websites.

Photography

The second-strongest category (median 75.5). Photographers who sell at scale treat the site as a storefront, and it shows in the plumbing: Peter Lik (92) posts the category's best score. One famous portfolio, Ellen von Unwerth's, was unreachable outright at measurement — its SSL certificate invalid. See our SEO for photographers. See our photography websites.

Unreachable at measurement (invalid SSL certificate): Ellen von Unwerth.

Luxury Brands

Eleven of fifteen maisons are sealed against machine measurement, so the category is scored on the four that answer; Patek Philippe (79) leads them. The fortress posture may be deliberate brand control — but as clients delegate more of their searching to software, it cedes the conversation to whoever machines can actually read. See our marketing for luxury brands.

Sealed against machine measurement, not scored: Louis Vuitton, Bottega Veneta, Loro Piana, Tiffany & Co., Chanel, Saint Laurent, Hermès, Rolex, Dior, Gucci, Cartier — see "The sealed twelve."

Real Estate

A strong middle field led by Hilton & Hyland (81). The notable absence: Sotheby's International Realty, whose listings business depends on being found, serves a crawler a blank JavaScript shell — and The Agency sits behind a verification wall no machine gets past. See our marketing for luxury real estate.

Sealed against machine measurement, not scored: The Agency — see "The sealed twelve."

Yachting

The quiet champion of the luxury web: category median 80, the study's best, with the overall winner — Moran Yacht & Ship at 98/100 — and four of the top ten sites overall. Brokerages built for inbound inquiry have made machine legibility a habit. See our marketing for yachting.

What this means for your firm

None of this is a question of taste. The sites in this index are among the most beautiful in the world, made by people who care about every pixel. The gap the Index reveals is legibility: whether all that craft is visible to the software that now stands between you and your next client. A collector asks an AI assistant for galleries showing a particular artist. A homeowner asks for interior designers with published pricing philosophies. A family office asks for yacht brokers in Monaco. The answers come from what machines can read — and much of the luxury web, this study shows, cannot be read at all.

The encouraging part: almost every point lost in this index is recoverable without touching a site's design. Structured data, crawler policy, server-rendered text, alt text, an llms.txt file — these are plumbing, not aesthetics. In our AffluentAllure™ approach to AI-powered luxury marketing this is the foundation layer: make the brand legible to machines without compromising how it feels to humans. If you would like to know your own score — and what it would take to move it — request your complimentary measurement.

Methodology & integrity

What we measured. On 2026-07-08 we requested each site's homepage, robots.txt, sitemap.xml and llms.txt the way any neutral crawler would, and recorded only public signals: robots.txt policy toward six major AI crawlers (GPTBot, ClaudeBot, anthropic-ai, PerplexityBot, Google-Extended, CCBot), machine-readable text served before JavaScript runs, schema.org structured data, response time and document weight, image delivery, metadata and accessibility basics. No private data, no logins, no scraping beyond four standard public URLs per site.

Scoring. AI Visibility 40 points (crawler access 15, readable text 10, structured data 10, llms.txt 5); Performance 20 (response 8, weight 6, modern images 4, lazy loading 2); Search Foundations 20 (title 3, description 4, canonical 3, sitemap 4, social metadata 3, robots.txt 3); Craft & Accessibility 20 (alt text 8, viewport 4, language 4, HTTPS/HSTS 4). The scoring is deterministic: the same measurements always produce the same score.

Security walls (†). Some sites refuse any client that is not a real browser — returning 403/429 to neutral, well-behaved HTTP requests. We do not guess at those sites: their on-page signals were measured through a real Chromium browser, their response timing comes from the browser's navigation timing, and their machine-readable text score is 0, because that is precisely what a neutral crawler receives. Their robots.txt policy is reported as written, and we describe these sites as refusing neutral clients rather than claiming any specific AI crawler is blocked, since security vendors may allowlist individual crawlers invisibly.

Selection. Sites were selected before measurement: galleries and interior studios from our published editorial rankings; architects from Pritzker- and AD100-tier international firms; photographers of international standing across fine-art, fashion, wedding and architectural work; maisons from leading brand-value rankings; brokerages and yacht houses of established global standing. Inclusion reflects prominence, not endorsement — and no commercial relationship with Nakada Design.

Corrections. A handful of sites run security walls that also block neutral measurement; they appear as "not scoreable" rather than being guessed at. If you run one of these sites — or believe any measurement is wrong — write to us via the inquiry page and we will re-measure and update this page. Raw data: download the CSV.

Our own score. Measured by the same engine on the same day, nakadadesign.com scores 86/100 (AI 30/40, Performance 20/20, Search 20/20, Craft 16/20). We publish it because an index you can't apply to yourself isn't worth citing.

Frequently asked questions

What is the Luxury Web Index?

An annual Nakada Design study that measures the public websites of leading art galleries, interior design studios, architecture firms, photographers, luxury maisons, real estate brokerages and yacht houses, and scores each 0–100 on how visible it is to search engines and AI assistants. Every score is computed deterministically from public, verifiable signals — no opinions, no panels.

Why does AI visibility matter for luxury businesses?

Affluent clients increasingly ask AI assistants — ChatGPT, Claude, Perplexity, Google's AI results — for recommendations before they ever search. Those assistants can only recommend what they can read. A website that serves no machine-readable text, blocks crawlers, or carries no structured data is effectively absent from the conversation, no matter how beautiful it is.

How were the sites selected?

By prominence within each category, fixed before any measurement: editorial rankings we had already published, award-tier standing, and established market position. Inclusion reflects stature, not endorsement, and no site in the Index has a commercial relationship with Nakada Design.

Our site scored lower than expected. Is that fixable?

Almost always, and usually without a redesign. Most points lost in the Index come from plumbing — missing structured data, JavaScript-only content, crawler blocking, absent metadata — that can be corrected while leaving the design untouched. That work is a core part of our AI and SEO services.

Can we request a re-measurement or correction?

Yes. If you believe a measurement is wrong, or you have fixed something and want the Index updated, contact us via the inquiry page. We re-run the same engine against your site and publish the updated score with the re-measurement date.

Cite this study: The Luxury Web Index 2026, Nakada Design (2026-07-08), nakadadesign.com/stories/luxury-web-index-2026. Journalists and editors: we are happy to provide vertical-specific breakdowns or re-runs for your story — get in touch.