What the free audit looks like
This is a representative example, built from the patterns we see across restaurant audits: an independent restaurant, about sixty covers, in a large European city. The numbers below are typical, the findings are the ones we write most often, and your report follows this exact shape with your name on it.
Visibility score
Readable in parts, invisible where it matters. The restaurant exists to every assistant and is recommendable to none of them: the menu, the identity, and the answers diners need are all missing at the moment of the question.
Representative example
Knows the restaurant exists and places it in the right city. Cannot name a single dish.
Reads the website live, finds a PDF where the menu should be, and cites a delivery platform instead.
Merges the restaurant with the similarly named place two districts over. Recommends neither.
Describes the place from a two year old review, opening hours included.
Five findings, none of them rare.
- The menu exists only as a PDF Finding 1
Dishes, prices, and dietary options are invisible at the exact moment a diner asks about them. AI assistants cannot reliably read PDF menus, so the kitchen's whole argument is missing from every answer.
- Three platforms, three different restaurants Finding 2
The website, the Google profile, and the delivery platforms disagree on the name, the hours, and the address style. Machines see several weak entities instead of one strong one, and confidence collapses.
- A platform outranks the restaurant for its own name Finding 3
Searching the restaurant by name lands on a delivery marketplace first. Recommendations route through a commission page, and the assistant reads that page's facts, not the restaurant's.
- No page answers a real diner question Finding 4
Dietary options, groups, the late kitchen, the reservations policy: all answered by phone daily, none stated anywhere a machine can quote. The assistant answers those questions anyway, using someone else's pages.
- Zero structured data of any type Finding 5
No JSON-LD on any page. To a machine the site is prose and photographs: pleasant, and unreadable as fact. Every competitor with markup is easier to recommend.
Three fixes, in the order that pays.
- Publish the menu as an HTML page with hasMenu markup Fix 1
One page, plain text, backed by structured data. This single fix answers the largest share of real diner questions and is why it goes first.
- Unify the entity everywhere, and say so with sameAs Fix 2
One name, one address format, one set of hours across the website, map profile, and platforms, with sameAs links tying the profiles to the site. The three weak candidates become one strong one.
- Add a diner FAQ in text, marked up as FAQPage Fix 3
Ten questions the phone answers every day, written down once, in the restaurant's own words. Assistants quote it verbatim, which is the point.
Who the assistants actually name.
Across the local dining questions in the audit, how often each restaurant appears in the answers. The gap is the lost tables.
Yours arrives with the real names: the competitors above you, and the reasons they are there.
Your restaurant, this report.
The free audit follows exactly this shape: score, platform breakdown, findings, fixes in paying order, and the share-of-voice table with real names. No obligation attached.