I scanned 15 Indian restaurant websites for AI visibility

I scanned 15 real, independent Indian restaurant websites (different cuisines, different cities, no chains) and scored each one, once, on 17 July 2026, with the same free Report Card I used for the hotel study. The score has nothing to do with the food. It measures whether an AI answer engine, the kind now fielding 'where should I eat' questions, can actually read the page: whether it can find the restaurant's name, cuisine, hours, menu, and location without a human clicking through five tabs first.

Structured data is the dividing line, again

Structured data is the line that splits this group in two. The 11 restaurants that publish some form of schema markup (LocalBusiness, Restaurant, Menu, whichever they chose) have a median score of 81. The 4 that publish none have a median of 49.5. That's a 31.5 point gap, and it's the widest structured-data gap I've measured in any vertical yet: clinics came in at 26.5, hotels at 27. Restaurants have the most to lose from staying illegible.

Here's what that gap means in practice. When a restaurant's own page carries no structured data, an AI answer engine still needs an answer, so it reaches for whichever source it can parse cleanly, usually the restaurant's Zomato or Swiggy listing, since those aggregator pages are heavily marked up by default. The restaurant gets recommended by proxy, its name and rating pulled from a platform, while the site meant to represent the business directly gets skipped. Structured data isn't decoration. It's the difference between being cited as yourself and being cited as a listing on someone else's platform.

Every single restaurant was missing one thing

Every one of the 15 restaurants, all 15, had zero FAQ markup. Not one used FAQPage schema anywhere on its homepage. It's the single cheapest fix in this whole study: diners already ask direct questions an AI engine could answer straight from a FAQ block if one existed. Do you have vegan options? Are you open on Mondays? Do you take walk-ins or only reservations? Most of these answers already sit somewhere in the site's prose. They're just not marked up in a form a crawler can lift and quote.

• 80% (12 of 15) had a missing or broken H1, so the page never states, in the one place crawlers weight most, what the restaurant actually is.
• 33% (5 of 15) had weak or missing Open Graph tags, so a shared link often shows no name, image, or description at all.
• 33% (5 of 15) had a title tag that ran too long or too short to work as a clean, quotable label.
• 27% (4 of 15) had no structured data of any kind, the same 4 sites dragging the median down to 49.5.

The menu-as-image trap

Hotels have the JavaScript booking widget. Restaurants have the menu as a photo or a PDF. It's the most common trap in this group: dish names, prices, and dietary tags sitting inside a scanned image or a downloadable PDF instead of plain HTML text. An AI crawler can't read pixels, and it doesn't run JavaScript either, so a menu rendered by a script is just as invisible as one locked inside a JPEG. If the words 'paneer tikka, 320, contains dairy' never appear as actual text on the page, no answer engine can ever quote them. The fix isn't a redesign. It's putting the same menu that already exists on the page as plain HTML text too, whether or not the photo or PDF version stays up for humans.

What to fix, in order

1. Add Restaurant or LocalBusiness JSON-LD with name, cuisine, address, hours, and price range. This is the single biggest lever: it accounts for most of the 31.5 point gap between the two groups here. The free schema generator builds this in a couple of minutes, no code required.
2. Add FAQPage markup for the questions diners actually ask: dietary options, timings, reservations, parking. Not one of the 15 restaurants in this study had this, so it's an easy way to stand apart from the rest of the field.
3. Publish the menu as plain HTML text, dish names, prices, and dietary tags included, even if a photo or PDF menu stays up for humans too. A crawler needs the words, not the image.
4. Fix Open Graph tags (title, description, image) so shared links and AI summaries show the restaurant correctly instead of a blank card.

None of this makes the food better. It makes the restaurant legible, so the engines already answering 'where should I eat' questions have something to read. More on why restaurants get skipped over in the first place at the restaurant AI visibility page.

Method and limits

All 15 homepages were fetched once, on 17 July 2026, and scored against a fixed list of machine-readability signals: structured data, FAQ markup, title and meta tags, Open Graph tags, H1 presence, and a few others. No JavaScript was executed, since AI crawlers don't execute it either, so anything rendered only by a script wasn't seen by the scorer any more than it would be seen by a real crawler. Only the homepage was checked, not the full site, and 15 is a small sample. The structured-data and FAQ findings are the ones I'd trust most here: they're binary, easy to verify by view-source, and consistent with what I've now seen across clinics, hotels, and restaurants alike.

Frequently asked questions