ChatGPT names four hotels in your city. We get you on that list.
Google gave guests ten links to browse. Modern AI platforms suggest four property names and stop there. When your hotel gets skipped, no alert appears on your reports; room bookings simply drop off during the dates you expected to sell.
Three steps, and you can stop after any of them. Start with where you stand today, then decide the rest once the evidence is in front of you.
The 15-minute check
We ask what you sell and who you compete with, put three questions to three AI platforms while you watch, and tell you what we would do about the answers.
The audit
Twenty-five questions across five platforms, read by someone who has run rooms. Every gap costed and ranked, with the two we would fix first and why.
The build
We make the fixes ourselves. No new website. Settings, listing fields, room names, and the two or three pages nobody got round to writing.
We ran this on a real Sydney property in August 2026,
name blanked out.
Top-rated on paper, yet invisible to AI.
Number one on TripAdvisor. 4.7 stars. 580-plus Google reviews. None of it reached the AI platforms that a growing share of travellers now ask instead of scrolling a search page. Yours takes ten minutes to check.
Here is what actually changed for you.
You used to get ten shots at this. Now you get one.
Open your phone, search your city and a hotel, and look at what sits above the links before you scroll. That is the shift, right there.
How it used to work
Ten links on the page. If you were not first, you could still be third, or seventh, and a guest would scroll and still find you. There was room to be found without being the best in your comp set.
How it works now
AI names three or four hotels and stops. There is no scrolling to a second page, because there is not one. You are in the answer, or you do not exist for that search.
Here is the bit most owners have not been told.
AI has never actually stayed at your hotel.
It builds an answer out of text about you, whatever it has learned or looked up. And it will leave you out rather than guess, if the sources do not agree with each other.
Your own website
Room descriptions, the pages you have written, the facts a machine can actually pick up. Most independent sites never write these down anywhere readable.
Your OTA listings
The OTAs, meaning Booking.com, Expedia and Agoda. Usually the most complete description of your hotel that exists anywhere, which is sort of the problem.
Google profile and reviews
The panel that appears when someone searches your name, plus TripAdvisor. If your reviews keep saying "perfect for families", that is who these systems decide you suit.
What other people write
Guides, roundups, best-of lists. Around half of what gets quoted back to a guest comes from here, not from you.
One AI sent guests two kilometres to a different bar. Rather than mention the one three minutes away. Sydney boutique hotel, 40-seat wine bar, asked on three platforms. Named zero times.
This is the Sydney property's real report.
A taste of the $1,000 audit.
Yours will look different in the detail and much the same in shape.
Best-reviewed property in the comp set. Named the least. One rival appeared across all four platforms, and this table shows exactly who is winning the recommendation, and how often.
Gemini told guests they could claim a daily credit for skipping housekeeping. The hotel's own site has no record of this. It came from an out-of-date page it currently trusts more than the property itself.
Three steps. You can stop after any one.
The audit
- Where you sit against your comp set, question by question, and who is winning the answer
- Every gap costed and ranked, with the two we would fix first
- Your rates against the OTAs across six dates, and what that gap is worth over a year
- What the OTAs cost you a year, worked from your real rates
- Twenty-five questions, five platforms, every answer dated and evidenced
The build
- We make every fix ourselves, across two phases
- A written record of every change, so nothing walks out the door
- The same 25 questions re-run at day 90, question for question
- No improvement at all by day 90, and we keep working another 30 days free
Monthly report
- The same 25 questions, re-run every month
- Your rates re-checked against the OTAs
- One page, naming the single thing worth fixing next
- No passwords, no logins, nothing for you to send us
We finish the build in week five, then leave the changes to settle. At day 90 we re-run the same 25 questions, on the same five platforms, against the same comp set. If there is no improvement at all, we keep working for another 30 days at no charge.
We guarantee visibility, because we can prove it. We will never guarantee bookings. Nobody honestly can.
Who is behind this
Mag Wan
"When I ran Matchbox, I would sit with the monthly numbers and see a week go quiet for no reason I could point to. Reviews were good, nothing had changed on our end, and still some weeks the bookings just did not come. I put it down to luck for years. It was not until I was doing this kind of work for hotels with real budgets that I found the actual reason: something as boring as a machine deciding you do not exist. If I had known that while I still owned the place, I would have fixed it in an afternoon. This is the business I wish someone had built for me back then."
- Founded, operated and sold Matchbox The Concept Hostel and Habitat Hostel, Singapore
- Runs digital marketing for an international hospitality group with more than 145 properties worldwide
- Fifteen years in digital marketing, all of it in hospitality
Frequently Asked Questions
More travellers ask AI where to stay every day, and hardly any hotels have caught up yet. That gap will not stay open. These systems lean on whatever is already established, so the properties that get found now are the ones they keep recommending later. Getting in early is the advantage, not just being visible eventually.
Good, keep them. The audit is a document, not a takeover. Plenty of properties buy it purely to hand to the people who already do their web work. Every finding is written so a competent web person can act on it without us.
No. If someone tells you otherwise, ask them to walk you through why. Of the seven causes found in the Sydney audit, not one needed a rebuild, just settings, listing fields, room names, and a couple of pages that had never been written.
No. The audit is gathered entirely from outside, the same way a guest sees you. You give us your website address and your city, and there is nothing for your team to send us or chase your IT person for. During the build we ask for access to one system at a time, for that one job, and hand it back when it is done.
You can, and some owners do. What comes back is a stack of AI answers, not a plan. Asking is the easy part. Reading the answers against your comp set, working out which gap is costing you a quiet Thursday in April, and deciding what to change first, is the part that comes from having run a hotel.
The fixes stay done. What moves is everything around them, since these systems update constantly and the hotels in your comp set are working on the same thing. That is what the monthly report is for. Plenty of properties finish the build and leave it there, and that is fine too.
AI is already choosing without you.
Only three or four hotels get named per search. Every week you wait, someone in your comp set takes one of those spots. Fifteen minutes tells you whether you are in the answer, and what we would change if you are not.