Before launching a study at scale (coming soon: sector by sector, market by market), we used our own scanning engine on a pilot sample of 8 sites chosen for being large, globally recognizable names — not a sample representative of SMBs, but a first honest test of the question: are even the world's biggest brands ready to be read by AI assistants?
Methodology
8 domains (marriott.com, ihg.com, dominos.com, etsy.com, notion.so, stripe.com, zillow.com, redfin.com), scanned with the AIVIS engine: deterministic checks only (no LLM calls), max 10 pages per domain, robots.txt always respected. No site was picked to "make headlines" — these are simply the first globally recognizable names across different industries (hotels, food delivery, e-commerce, SaaS, real estate) that came to mind for a diverse test.
The numbers
- Average score: 43/100 — on a scale where 90+ is an "A".
- Only 1 site out of 8 (13%) has a published llms.txt.
- 2 sites out of 8 (25%) limit or block at least one relevant AI bot (GPTBot, ClaudeBot, PerplexityBot and similar) in their robots.txt.
What this means (and what it doesn't)
It doesn't mean these sites are "badly built": many have likely invested heavily in classic SEO, performance, accessibility. It means optimizing for AI answer engines is still, today, work that's almost never done, regardless of the resources available — consistent with AIVIS's founding thesis: this is a new era, the standard hasn't been written yet, and whoever moves now has a real advantage.
A stated limitation: our crawler does static HTTP fetches and doesn't execute JavaScript. Some of the low scores in this sample likely also reflect content (schema.org, meta tags) injected only client-side, invisible to our scan just as it is to a real AI crawler that doesn't execute JS — a real AI-visibility problem, not a measurement artifact.
What's next
This is a pilot sample of 8 global sites, not a market study. The next step is a targeted study by sector and country, on a larger, curated sample — we'll publish it here as soon as it's ready.
Want to know where your site stands? Run a free scan: same engine, same methodology, results in under 90 seconds.