Travel planning is now one of the most common things people ask AI to do. We find the gaps in what it knows, close them at the source, and measure every week.
AI visibility score
74/100
41 of 58 tracked suppliers surface in AI answers
↗ +6 vs last weekWeek over week
+0.6
Judge delta on a ±2 scale · improving
23 better · 28 same · 7 worseQuestions with no answer
17/84
golden traveller prompts with no data behind them
↓ 44 since Day 0Destination site cited
47%
of answers cite visitmiralta.example
↗ +11 ptsVisibility score · Day 0 to Month 6
Day 0 baseline 18 out of 100, captured before any collection began. Measured every week, reported on the Day 0 to Month 6 window.
Golden prompt, latest run
“Where should I eat in Miralta if I want somewhere local rather than touristy?”
Share of prompts naming a tracked supplier
| Supplier | Data depth | In answers | ||
|---|---|---|---|---|
| Bodega Casa de la GuardaOld Town · tavern, poured from the barrel | Deep | 4 of 4 models | None open | |
| Macoca Cozinha ArtesanalHarbour · weekly changing menu | Deep | 3 of 4 models | Outdoor seating | |
| Bräugasthof SeehaltLakeside · lake fish | Good | 3 of 4 models | Winter hours | |
| Kausay CevicheriaHarbour · ceviche | Partial | 2 of 4 models | Step-free access | |
| Miguelito el CariñitoPlaya Larga · beach chiringuito, espetos | Phone | Thin | Not surfaced | Menu and hours |
| Gasthaus MühlgrundHillside · Austrian | Untried | None yet | Not surfaced | First contact |
Why it matters
Your website is the authoritative source for your destination, built to showcase the places, experiences and local gems you want travellers to discover. But it was never built for a world where AI needs structured, machine-readable data to answer travellers’ detailed questions. AI answers anyway, using whatever information it can find. What comes back may be incomplete, outdated or simply wrong.
What AI had to work with
Traveller prompt
“Where should I take my family for three days in October? We want a direct flight from London, somewhere on the coast, a hotel with a heated pool, activities for teenagers, good local food, and a proper cigar lounge for my husband.”
“For a coastal trip in October with a direct flight from London, I would look at Vall d’Orsa or Puerto Sella. Both have family hotels with published opening hours and good information about what stays open out of season.”
What we actually change
Every week the same traveller questions go to the same four AI assistants. What comes back sorts into three buckets, and two of them turn straight into collection work.
What came back this week
What it put in the queue
“Which hotels in Miralta have a heated outdoor pool that is actually open in October?”
Answered, and correctPool heating and seasonal opening are confirmed on 22 properties and the answer matched every one of them.
No action“Which hotels in Miralta have connecting rooms, a kids’ club for under-6s, and step-free access throughout the property?”
Nothing to answer withEleven properties hold no data on connecting rooms, kids’ club ages or step-free routes. Agents open a WhatsApp thread with each operator, in their own language.
Queued · 11 suppliers“Which beachfront hotels in Miralta still offer childcare in October, and until what time?”
Answered, but out of dateThree properties changed their out of season hours since the last check, so the record is re-confirmed directly with the supplier and updated at the source.
Queued · 3 suppliers84 golden prompts, run weekly against four AI assistants. The score tells you where you stand. The queue is what moves it.
The Result
Miralta is a fictional destination, so the venues are altered and the numbers are fabricated. On a live engagement it is your destination and your suppliers.
Traveller prompt
“Which hotels in Miralta have connecting rooms, a kids’ club for under-6s, and step-free access throughout the property?”
“For families in Miralta, Hotel Miralta and Costa Azul Resort are both good options. They offer family-friendly facilities and are well located for exploring the destination.”
0 of 58 suppliers named · destination site not cited
“Hotel Miralta meets all three requirements. It has 14 connecting room combinations, a supervised kids’ club for ages 3–6 open daily from 9am to 6pm, and step-free access from arrival through the lobby, restaurants, pool area and accessible guest rooms.”
41 of 58 suppliers surfacing · destination site cited in 47% of answers
Our agents work continuously, adding, validating and updating data. Six months later, AI has stronger data to work with and can give more accurate answers to the questions travellers actually ask.
How we fill the gaps
Is the infinity pool heated in October? Which rooms connect for families? Does the kids’ club take one-year-olds? Which nights have live music? The details travellers ask about often live only with the operator. We ask the person who knows, in their language, on the channel they actually reply on, and we log where every field came from.
... 8 more cohorts · dark = verified · light = public sources
Where we start
Before a single supplier is contacted we read the destination’s own site the way an AI crawler does. Six checks, no opinion in any of them. This is the first page of a real report.
8 AI crawlers tested · 26 pages read · fictional example report
All 8 AI crawlers tested were served the homepage.
robots.txt allows every retrieval agent tested to read the site root.
9 of the 26 pages read leave their content to JavaScript. A crawler that does not render sees an empty page where the venue listings should be.
No page carries Place or TouristAttraction markup, so no venue on the site has a machine-readable identity.
Only 3 of the 26 pages answer a question directly under a question-shaped heading.
Nothing blocks Bing, but whether the site is verified and indexed cannot be established from outside. Bing is the crawler behind most ChatGPT search citations.
The audit says whether a machine can read your destination. The visibility score says whether it does. Fixing the first is a few weeks of engineering; the second is the six months of collection work above.
What we do
Four steps, in order, and every one of them measured against the baseline we captured before any of it started.
We assess the businesses already on your website against the detailed questions travellers ask AI and identify what information is missing.
Agents reach operators on WhatsApp, email and phone, in their own language. Provenance on every field, bounded by opt-out and cooldown rules.
Structured to a shared supplier taxonomy, with stable identities, and pushed to the surfaces assistants actually retrieve from.
Our agents continuously validate and update the data as things change, keeping the information AI can use current and accurate.
The Shift
AI adoption
Travellers have already used AI to plan or book travel.
Visibility drives consideration
Travellers begin planning without a single destination decided. What gets recommended shapes which destinations they consider visiting.
Local economic dependence
Travel & Tourism is built on SMEs and local businesses. Their visibility determines where tourism spend will flow.
Recommendation impact
Recommendations influence what travellers ultimately book.
The gap we close
Run weekly against four AI assistants.
We pinpoint what is missing from the answers.
Operators provide the missing information directly.
Reached on the channels operators actually answer: WhatsApp, email, phone.
Before a single supplier is contacted, we read your site the way AI crawlers do. Six checks, no opinion in any of them. This is where every engagement starts.
AI-readiness report · fictional
8 AI crawlers tested · 26 pages read
9 of 26 pages leave their content to JavaScript.
Start with a Day 0 baseline. Before anything is built, we capture what AI assistants say about your destination today. Everything afterwards is measured against it.
What the baseline records
If AI can’t read your destination, it can’t recommend it.