"Are we showing up in AI search?"
It is the question every accommodation operator is asking in 2026, and almost nobody can answer it. The usual response is a shrug, then someone types the property name into Google, sees the AI summary mention it, and concludes "yeah, we’re there".
That is not a measurement. That is a vibe.
Why the brand-name check tells you nothing
Your own name is the easiest query in the world to win, and the least useful one to check. A traveller who already typed your property name was always going to find you. AI search did not win you that guest. You already had them.
AI search matters for the traveller who does not know you yet. The one typing "best family apartments near the beach", or "where to stay in town for a long weekend", or "dog-friendly accommodation on the coast". Those queries decide whether an AI answer sends you a guest you would never otherwise have reached. And those are exactly the queries the brand-name check never looks at.
There is a second problem. You are checking from your own browser, logged in, sitting in your own region. The result you see is personalised to you. It is not the result a traveller two states away sees when they ask the same question. You are not measuring your visibility. You are measuring your own reflection.
What a real audit looks like
A real audit starts with the full set of queries a traveller actually uses when they plan a stay in your region. Not your brand name. The whole journey: location intent ("accommodation in the area"), accommodation discovery ("best apartments near the beach"), things to do, events, food, family travel, seasonal trips, and the brand and brand-variant searches as one small slice rather than the whole test.
For a single region, that is a few hundred prompts. In a recent audit we ran for a property we manage, the full set came to 257. Each one run through an AI-search presence report, not eyeballed from the operator’s own laptop.
Not every query counts, and that is the important part
Of those 257 prompts, only 166 were pursuable. A pursuable query is one a real in-market traveller would plausibly use and one you could realistically be cited for. The other 91 were noise: disambiguation (the town shares its name with somewhere else entirely), out-of-region intent, and searches that have nothing to do with booking a stay.
This distinction is the whole game. If you measure your AI visibility across all 257, you either flatter yourself or punish yourself with queries that were never winnable. You measure across the 166 that were. Everything else is thrown out before you score a thing.
The number that replaces the vibe
Your AI search share of voice is the number of pursuable queries where an AI answer cites you, divided by the total pursuable queries. That is a real percentage. It is repeatable. You can run the same set next quarter and know whether you went up or down, and on which categories.
Suddenly "are we showing up in AI search" has an answer that is a number, not a feeling. And a number is the only thing you can actually manage.
What the gaps tell you to do
The audit does not just score you. It shows you the exact queries where the AI cites a competitor and not you, which is where the real value sits. Each gap is a build instruction.
"We are invisible on every things-to-do query in our area" is a content brief. "We are cited on accommodation searches but never on family-travel ones" tells you precisely what to write next. "A competitor with worse rooms outranks us on the seasonal queries" usually means their pages carry the schema and the entity signals that let the engine read them, and yours do not. The audit turns a vague anxiety into a short, ordered list of things to fix.
You can start this yourself
A lighter version is within reach for any operator. Write down the real queries a traveller uses to find a place like yours, the honest ones, not your brand name. Check a sample from a clean browser that is not logged into your account. Note where an AI answer cites you and where it cites someone else. That alone will tell you more than a year of gut-feel checks.
The structured version, the full query set, the pursuable filter, the quarterly re-run, the gap-to-content mapping, is tedious and easy to fool yourself on, which is why it is one of the things we run inside Boost Direct. But the rule does not change whether you run it yourself or hand it off. Pick the real queries. Count the citations. Track the number.
A vibe cannot be improved. A number can.