Four real hotels. One consistent pattern.

Why we're publishing all four, unedited

We could have picked our best-looking result and called it a case study. Instead we're showing all four, because the pattern across them is the story. Three of these hotels — a Prague boutique, a Prague design hotel, and a spa resort — scored 0% on Google AI Overviews for generic questions. Not low. Zero. The fourth, a family-run mountain hotel, got exactly one AI Overviews mention out of 12 checks — the only AI Overviews detection anywhere in this data set, and still a rare result, not a reliable one. Every hotel was named in the low single digits to low teens overall, even the best performers. And in every single case, AI engines cited someone else's website — a booking site, a travel blog, a local directory, sometimes a competitor — far more often than the hotel's own site. This isn't a fluke in one market or one segment. It's what's happening right now, across four very different hotels, in the exact channel more and more travelers are starting their search in.

None of this means these are bad hotels — three of the four score well or reasonably on the parts of AI visibility they can control, and one is proof that a smaller property can still outperform on the numbers that matter. What it means is that being findable on Google is no longer the same thing as being findable by AI. Read through the four hotels below, then run the same free check on your own property — it takes a few minutes and shows you exactly where you stand today.

Why these numbers aren't a precise ranking

Three of these hotels are tracked on Staylight's full 5-engine panel — 200 to 300 non-branded checks each. The fourth, Bouda Mama, is on our entry-level Radar plan, tracked on a smaller panel (24 checks) across just 2 engines. That's not a flaw in the data, but it does mean the headline "overall detection rate" numbers shouldn't be read as a precise ranking across plan tiers — on a 24-check panel, a single detection swings the rate by about 4 percentage points; on a 200-check panel, one detection moves it by roughly 0.5 points. Bouda Mama's 12.5% and Mosaic House's 12.5% are both real, but they carry very different amounts of statistical noise.

To make an honest, like-for-like comparison possible, the table below shows detection rates on just the two engines every hotel in this data set is tracked on — ChatGPT and Google AI Overviews — regardless of plan tier. It's still a small-sample comparison, but at least the engine coverage is equal across all four.

Even on this fair, same-engine basis, Mosaic House leads on ChatGPT (20%); Bouda Mama is the only hotel with any Google AI Overviews detection.