The Luxury Homepage Index 2026: what 60 high-end homepages tell machines
I ran the same audit on 60 luxury, fashion, beauty and hospitality homepages. The average score was 81, four in ten turned the audit away, and five gaps kept coming back.
A few days ago I read a handful of luxury homepages by hand and wrote about what I found behind the campaign images. Reading by hand doesn't scale, and a few memorable mistakes don't tell you how common they are. So this time I ran the same checks the way a machine would: one automated audit, the same rules, on 60 luxury, fashion, beauty and hospitality homepages in a single day.
These are the numbers, the brands that did best, and the five gaps that came up again and again.
The headline numbers
- 60 homepages audited on October 26, 2026.
- 36 could be read. The other 24, four in ten, couldn't be read at all; most refused the automated request outright.
- Average score: 81 out of 100. The median was 82, the range 53 to 96.
- Only one site in 36 passed every check that doesn't depend on where the visitor is.
An 81 sounds respectable, and in a sense it is: most of these sites get the visible things right. The gaps are almost all in the parts a shopper never sees and a machine reads first.
The leaders
These ten homepages scored highest. Each of them is doing the unglamorous work well, and it shows.
- Chopard: 96
- The Peninsula Hotels: 92
- Harry Winston: 91
- Vince: 91
- Tom Ford: 89
- Auberge Resorts Collection: 88
- Aesop: 87
- Khaite: 86
- Brunello Cucinelli: 85
- Marc Jacobs: 85
Chopard's homepage was the only one that passed every location-independent check: one clear main heading, a share image, described images, structured data and clean code. That's what "looked after" looks like.
The five most common gaps
Across the 36 sites that could be read:
1. No single main heading: 56%
Twenty sites didn't have exactly one main heading (H1), and sixteen of them had none at all or more than three competing ones. The H1 is how search engines and AI assistants learn what a page is about. On a homepage built entirely from images, it's often the only sentence that says who the brand is.
2. No share image: 53%
Nineteen sites had no share image (og:image). When someone pastes the homepage into a message, LinkedIn or Slack, the preview shows a blank box or whatever image the platform guesses. For brands that spend millions on a single campaign photo, it's a strange place to leave things to chance.
3. Images without descriptions: 47%
Seventeen sites had images with missing or meaningless alt text. Screen readers announce them as file names or nothing, and machines can't tell a handbag from a hotel lobby.
4. Weak or missing meta description: 39%
Fourteen sites had no meta description or one that was too short or too long to work. That's the line under the brand name in Google, and the first summary an AI assistant may lift.
5. No structured data: 31%
Eleven sites had incomplete structured data (JSON-LD), and nine had none at all. Structured data is how a site tells machines, in their own language, "this is an organization, this is its name, logo and official social profiles." Without it, assistants piece the brand together from whatever they find.
Close behind: subheadings that are missing or used for styling rather than structure (31%), page titles long enough to get cut off in search (19%), and broken character encoding (11%). Three sites still carried leftovers from the template they were built on.
Four in ten sites wouldn't open the door
The most striking number isn't a gap on a page. It's that 24 of 60 homepages couldn't be read at all. Nineteen refused the request outright, three returned server errors and two sent it around in a redirect loop.
Bot protection is a sensible thing for a large retailer to have. But the same kind of rule that turns away an audit can also turn away an AI assistant fetching the page for a shopper who just asked about the brand. When that happens, the assistant answers from memory or from someone else's website. It's worth checking that the door is closed only to the visitors you mean to keep out.
What the leaders do differently
Nothing exotic. They write one plain sentence for the main heading. They set a share image on purpose. They describe their images. They tell machines who they are with structured data. None of this changes how the homepage looks, and every item on this list is a few hours of developer time, not a redesign.
Method
Each homepage was checked once on October 26, 2026 with the free Luxury Homepage Audit, which reads the page's HTML and scores 19 checks out of 100. The audit runs from a server in Europe, so some brands may have served a European version of the site. For that reason, the gap figures above leave out the three checks that depend on the visitor's location: the US privacy opt-out link, spelling conventions and discount wording. The overall scores are as the tool reported them. Sites that block automated requests are reported as unreadable, not scored. Homepages change daily, so a site may score differently today. I have no commercial relationship with any of the brands named here. If you'd like a brand added to the next edition, write to me.
Check your own homepage
The Luxury Homepage Audit runs the same 19 checks on any homepage in about twenty seconds. To see what AI assistants actually know about your brand, try the AI Visibility Check.
Journalists: the figures are free to cite with a link to this page. For the full data or questions, write to info@valevlaube.com.
Selling to American buyers? I help Estonian, Nordic and Baltic companies adapt their websites, messaging and launch materials for the US market.
Run the free US-Readiness Check
