Contents09 sections
Most studios don't come up at all. We checked 224 high-end interior design studios in ChatGPT and Perplexity, and 171 came up in none of 6 answers. The biggest difference was the city: in London, New York, Sydney, Melbourne and Paris, 4 of 69 came up. Blocked AI bots and missing technical files did not explain it.
What did we check?
Between 21 October and 6 November 2026 we asked ChatGPT, with web search on, and Perplexity three questions a client might type about each studio's own city:
- "Who are the best interior designers in Sydney?"
- "Recommend a high-end modern interior designer in Sydney."
- "Sydney interior designer for a full home renovation."
That makes 1,344 answers, 1,210 of them separate, because studios in the same city checked on the same day could share an answer. We added one instruction to every question, the same one our free AI check uses: "Answer briefly, as a short ranked list of studio names with one line each." A longer answer could name more studios, so these are counts of short answers. For 29 studios we changed one or more questions to fit the studio. A studio came up if an answer named it, by its name or its web address, in the words of the answer, and we read doubtful matches by hand. A link to its site without the name did not count. On 6 November we then read the 215 studio websites that we could open: the platform, the words on the home page, the project pages and the robots.txt file.
A few things to keep in mind. These are studios we looked at as possible clients: high-end, modern, and mostly with a website we thought could be better. So this is not a picture of all studios, and it cannot show how a very strong website does. Each studio was checked once, and a studio that did not come up that day might come up on another. We moved to a newer ChatGPT model during those two weeks. Google's AI answers were not part of it.
Most came up nowhere
171 of the 224 studios came up in none of their 6 answers. 25 came up in just one, and 1 came up in all six.
With that instruction, an answer names about 4 studios on average, so there are few places. The two assistants also differ: ChatGPT named 48 of the studios at least once, Perplexity 23, and only 18 came up in both. One question in one assistant tells you very little.
The city made the biggest difference
| City in the question | Studios | Came up at least once |
|---|---|---|
| London | 23 | 1 |
| New York, Brooklyn included | 14 | 1 |
| Sydney | 17 | 1 |
| Melbourne | 9 | 1 |
| Paris | 6 | 0 |
| All these together | 69 | 4 (6 percent) |
| Everywhere else | 155 | 49 (32 percent) |
The numbers only show the gap. Our reading of it: in a big city, a short answer to a broad question has room for about 4 names and many known studios to choose from, while in a smaller place the local names come through.
So if you are in a big city and don't come up for "the best interior designers in London", that says little about your work or your website. Almost nobody in our check did. We looked closer at what the London answers read and the New York ones.
What did not explain it
Blocked bots. 2 of the 215 sites blocked AI search bots in their robots.txt file, and both studios still came up. Look once, but it is rarely the reason.
An llms.txt file. 8 of the 38 sites with the file came up (21 percent), and 43 of the 177 without it (24 percent).
Structured data on the home page. 38 of the 156 sites with it came up (24 percent), and 13 of the 59 without it (22 percent).
These gaps are small and our sample is not random. They don't show that either thing hurts. They do show that neither is a shortcut.
The platform. Squarespace was the most common, with 15 of 77 coming up. WordPress had 18 of 58, Wix 2 of 20 and Webflow 1 of 10. These groups are small and mixed across cities, so we would not move a site to another platform because of them.
What the words on a site did
Outside those five cities, studios with more words on the home page came up more often:
- under 100 words: 10 of 48 (21 percent)
- 100 to 499 words: 21 of 68 (31 percent)
- 500 words or more: 16 of 33 (48 percent)
Inside the five cities there was no such pattern: 2 of 30, 1 of 19 and 1 of 17.
This does not mean longer pages win. A studio with more to say may also have more years and more press behind it, and our sample cannot separate those. But an assistant can only repeat what is written down. The typical home page in this check had 189 words in its HTML, and 78 of the 215 had under 100. That leaves little room to say who the studio is, where it works and what kind of projects it does.
We could read the project pages of 81 studios through their sitemaps. The typical studio had 17 project pages of about 182 words each, and only 8 studios reached 400. For these broad questions, the length of a project page made no visible difference: 11 of 42 came up under 200 words, 9 of 31 from 200 to 399, and 2 of 8 from 400 up. But these questions don't ask about a project.
A specific question reads other pages
On 4 November we ran a separate, smaller test with ten other questions, some of them much more specific, twice each. One was "Who is the best high-end interior designer in London for a full renovation of a Georgian townhouse?"
ChatGPT cited 5 pages for it across the two runs. 4 of them were studios' own pages about one townhouse or historic London house. The fifth was Historic England. For a French question about a Haussmann apartment in Paris, all 6 pages it cited were on studios' own websites, including a project with "Haussmannien" in its title.
Two runs of a question is a small sample. It does show what a project page can do that a home page cannot: match the words of someone who knows what they have.
Where else the answers read
When a studio was named, its own site was among the cited pages in at least 60 of the 117 answers: 48 of 79 for ChatGPT, at least 12 of 38 for Perplexity. In the rest, we found no citation of the studio's own site.
Counting each separate answer once, Perplexity had a Houzz page among the first 8 pages it looked at in 450 of its 605 answers, and cited one in at least 196. No site was cited more. ChatGPT cited Houzz in 26 of 605. Perplexity's counts are a minimum, because it cites by number and we kept only its first 8 pages per answer. Instagram did not appear once, cited or looked at, in 6,910 links. Houzz, Instagram or a website goes into this, and Press on an interior designer's website looks at the magazine lists.
What can a studio do?
- Ask the question your client would ask. Your area and your kind of project, not only "best in London". Do it a few times, in both assistants.
- Write your project pages for that question. The place, the building, its age, what was done. Our guide to the project page gives the order.
- Say it plainly on the home page. One sentence: who you are, where you work, what kind of projects you do.
- Keep Houzz and the directories correct. The same name, city and description as on your own site.
- Check robots.txt once, then move on. Don't expect a technical file to do this work for you.
None of this guarantees a place in an answer. No one can honestly promise that. But it gives AI assistants words they can use, and it helps the clients who read your site as well.
Where do you stand?
Run the free AI check: it asks ChatGPT and Perplexity three questions about your city, live, and shows whether your studio comes up and who is named instead. If you are in a big city, read the result with the table above in mind.
05 questions
Questions this piece answers
Q01Why doesn't my interior design studio come up in ChatGPT?
Most studios don't. In our check of 224 high-end studios, 171 came up in none of 6 answers. We asked for short answers, and those name about 4 studios, so in a big city most are left out. Blocked bots and missing technical files did not explain the gap.
Q02Is it harder to come up in a big city?
In our check, much harder. In London, New York, Sydney, Melbourne and Paris, 4 of 69 studios came up. Everywhere else, 49 of 155 did.
Q03Is my website blocking ChatGPT?
Probably not. Of the 215 studio sites we could read, 2 blocked AI search bots in their robots.txt file, and both studios still came up. It takes a minute to check, but it is rarely the reason.
Q04Does an llms.txt file help a studio come up?
We saw no sign of it. Of the 38 sites with the file, 21 percent came up. Of the 177 without it, 24 percent did. Our sample does not prove the file hurts, only that it is no shortcut.
Q05How can a studio come up for more specific questions?
Write project pages that say the place, the kind of building, its age and what was done. When we asked ChatGPT about a Georgian townhouse in London, 4 of the 5 pages it cited were studios' own pages about one townhouse or historic London house. That was two runs of one question, so a small sample.