When someone asks AI about where to eat tonight — can it read your menu?
Diners ask for a place that fits a set of constraints at once: open now, walkable, takes a booking for six, has something for a coeliac. Every one of those is a fact about a venue, and a menu published as a PDF or an image answers none of them.
When someone asks AI about a place to stay with parking — is it your listing or theirs?
Travellers ask assistants for a place with parking, near a specific thing, that takes dogs, for two nights. Aggregators have this data in structured form and independent properties usually do not — so the assistant recommends the aggregator's version of you, with the aggregator's commission attached.
When someone asks AI about a venue for 100 people — does it know you fit them?
Someone planning a work function or a wedding asks for venues that hold a number of people, on a date, within a budget. Capacity, minimum spend and what a package includes are the three filters, and they are the three things venues most often reserve for an enquiry form.
We ask what your customers ask
Not keywords — the actual questions people put to ChatGPT, Claude and Gemini before they choose. Then we read back who got named, and whether it was you.
We put a representative set to each assistant and read back who got named — and whether it was you. See what an assistant does with one of these →
The checks that only matter in hospitality
Every scan covers crawler access, schema and content. These are the ones an Australian hospitality buyer — and the assistant answering for them — actually asks about.
Where you actually work
Almost every question of this kind put to an assistant has a place in it. We check that you've named your regions, cities or postcodes rather than implying them.
A real number, somewhere
One indicative figure — “6.6 kW from $4,990 installed”, or a c/kWh rate — is the difference between being usable in an answer about cost and being skipped. “Request a quote” cannot be quoted.
Plan terms a machine can compare
Exit fees, contract length, benefit period, and how you sit against the DMO/VDO — as text. In a downloadable PDF they're invisible, so you're left out of “which retailer has no exit fees?”.
A real number, somewhere
One indicative figure — “6.6 kW from $4,990 installed”, or a c/kWh rate — is the difference between being usable in an answer about cost and being skipped. “Request a quote” cannot be quoted.
Plan terms a machine can compare
Exit fees, contract length, benefit period, and how you sit against the DMO/VDO — as text. In a downloadable PDF they're invisible, so you're left out of “which retailer has no exit fees?”.
Where you actually work
Almost every question of this kind put to an assistant has a place in it. We check that you've named your regions, cities or postcodes rather than implying them.
A real number, somewhere
One indicative figure — “6.6 kW from $4,990 installed”, or a c/kWh rate — is the difference between being usable in an answer about cost and being skipped. “Request a quote” cannot be quoted.
Plan terms a machine can compare
Exit fees, contract length, benefit period, and how you sit against the DMO/VDO — as text. In a downloadable PDF they're invisible, so you're left out of “which retailer has no exit fees?”.
Where you actually work
Almost every question of this kind put to an assistant has a place in it. We check that you've named your regions, cities or postcodes rather than implying them.
Results with numbers attached
“Cut 480 MWh and $96,000 a year across nine sites” gets quoted. “We helped a client save money” cannot be. We check for the units.
Why good operators come back invisible
None of these are marketing problems. They're all a fact that exists in your business but not in a form a machine can read.
Scan, see it, fix it
The scan and the report cost nothing and need no account. You only pay if you want to keep editing after the fixes are in.
Scan your sitefree
Enter your URL and confirm your business email — one code, so a competitor can't scan you and read your gaps.
Read the reportfree
Your score, what AI currently believes about you, who it recommends instead, and every issue ranked by what it costs you. Download it as a document.
Fix themfree
Open your report as a working list. Each issue has the exact change to make — click one, we make it, and re-check that item on the spot.
Keep it current
Prices, rebates and accreditations change. A plan keeps your record editable and re-scanned, so what AI says about you stays true.
Before you scan
Why do you need my business email?
Because the report lists exactly where a business is weak, and that's not something a competitor should be able to pull on you. One 6-digit code proves you're on the inside before we show it. We don't add you to anything.
Is this just an SEO audit with a new name?
No. SEO asks whether you rank in a list of links. This asks whether an assistant can read your page at all without JavaScript, whether your accreditation and prices exist as facts rather than pictures, and what the assistants actually say when asked to recommend someone in your trade. We put the questions to them and quote the answers back.
What if the scan gets my segment wrong?
Then you've already learned something. We work out what kind of hospitality business you are the same way an assistant does — by reading your pages, with no inside knowledge of you. So if we read you as accommodation when you’re restaurants & cafés, that is what ChatGPT and Gemini are concluding too, and it's why you're missing from the questions you should be winning. Being unreadable and being miscategorised cost you the same customers.
Pick the right one above and scan again to see what you should look like. The segment decides which checks apply, so an installer is never marked down for having no tariff terms and a retailer is never asked which panels it uses — checks that don't apply are excluded from your score rather than counted against you.
Do I have to change my website to act on this?
No. The fixes write to your Agoria business record, which publishes the machine-readable version assistants read — your existing site stays exactly as it is. If you'd rather hand the whole thing over, our team can do it for you.
How current is this?
The scan runs live against your site and against the assistants each time — nothing is cached from an earlier run. Scores are weighted by published research on what actually drives AI citation, and every finding is tagged with how strongly it's evidenced. Read the methodology.