I scanned 632 sites across 28 verticals for AI readability

I've now scanned 632 real business homepages across 28 verticals, live, using the same engine behind my free Report Card tool. That's roughly seven times the size of the corpus I reported on last month, and the two headline findings didn't just repeat, they sharpened. Most businesses clear the basic AI readability bar: median score is 87, and 57% score an A. But one format is still missing almost everywhere. FAQ schema, the exact structure AI answer engines quote most, is absent from 81% of sites in the corpus. That single gap is now the clearest, most fixable edge in the entire dataset.

Most sites clear the basic bar

Across all 632 sites, the median AI readability score is 87 out of 100. That's a genuinely good number: on the same 0 to 100 scale as my free Report Card tool, 87 sits comfortably in A territory. 361 sites, 57% of the corpus, score an A (85 or higher). Another 121 score a B, so nearly four in five businesses in this dataset are at least readable to an AI answer engine without major structural problems.

But look at the mean, and a different picture appears. It's 79.6, well below the median. When the median sits above the mean like this, it means there's a large cluster of strong performers being dragged down by a long tail of weak ones. That tail is real: 103 sites score a C, 28 a D, and 19 an F. So the honest read isn't "most businesses are fine." It's that a solid majority have already done the basics, and a meaningful minority are badly exposed, invisible to the AI tools their own customers are starting to use.

FAQ schema is the blind spot almost nobody fixes

513 of the 632 sites in this corpus, 81%, have no FAQ schema at all. This is the single biggest and most universal gap I've found across two rounds of this study, and scale hasn't changed it: it holds just as strongly at 632 sites as it did at 93. FAQ schema is not a niche technical nice to have. It's a small block of markup that states a question and its answer in a format search engines and AI answer engines can parse directly, and it happens to be the exact structure those engines quote most often when they answer a question about a business.

What surprised me is where the gap shows up. It isn't confined to weak verticals, it shows up inside the strongest ones too. Among immigration consultants, one of my top two verticals by median score, 61% still have no FAQ schema. Among education sites, it's 75%. Among the 173 digital agencies in the corpus, it's 82%. Five entire verticals, including hotels and law firms, have it on precisely zero sites. A business can have a fast, well-structured, schema-rich homepage and still be leaving the single highest-leverage format for AI visibility completely unused.

Structured data is basically a solved problem now

Only 138 of the 632 sites, 22%, are missing structured data (JSON-LD) entirely. Roughly four in five businesses now have some machine-readable markup on their homepage, whether that's organization details, local business data, or product information. A year or two ago that would have been the main line separating sites AI tools could parse cleanly from sites they couldn't.

At this scale, it isn't anymore. Structured data has quietly become table stakes, something most businesses (or their web developers, or their website builder) already handle by default. FAQ schema hasn't caught up. It sits in a different category: something a business has to add deliberately, usually by hand, and 81% of the corpus hasn't bothered. That's the real dividing line now.

Who's ahead and who's badly behind

The best performing verticals in the corpus are dietitians and immigration consultants, tied at a median of 91.5, followed by education and IVF clinics at 90. All four sit in health and professional services, categories where a prospective client typically researches heavily before ever making contact, and where the businesses seem to have responded by building out real informational content.

The worst performing verticals tell almost the opposite story. Gyms and fitness studios are dead last at a median of 53.5, followed by law firms at 59.5, vets and pet clinics at 66, and hotels at 76. My guess, and it is a guess, is that these are categories built around walk-ins, referrals, and platforms like Instagram or booking aggregators rather than a homepage doing the selling. When the website is treated as an afterthought, that shows up directly in how well an AI answer engine can read it.

What to fix, in order

1. Add FAQ schema to your homepage first: it's the single highest leverage fix in this dataset.
2. Write out the 5 to 8 questions real customers actually ask before they book or buy.
3. Mark up your existing homepage content with basic JSON-LD if you haven't already (organization, local business, or product schema).
4. Check your own score with the free Report Card tool before guessing what's missing.
5. Re-scan after you fix something. A schema fix that isn't verified is just a guess.

Method and limits

I scanned 632 real business homepages across 28 verticals, live, using the same engine that powers my free Report Card tool. This is homepage-only, not a full site crawl, and every number here is aggregate. No individual business is named. 27 of the 28 verticals are Indian businesses; one set of 24 spans the US, Canada, and Australia. The honest caveat: a scan like this can only find businesses with a real, standalone website, so it skews away from businesses that operate mostly through Instagram, marketplaces, or other aggregators rather than their own site.

Frequently asked questions