What You Can Actually Learn From an AI Visibility Report
Tharro's AI Visibility Report runs your hotel through hundreds of structured queries across Perplexity, OpenAI, and Gemini — and measures what comes back. Not page rankings. The actual content of AI recommendations, and whether your hotel is in them.
The output tracks three metrics per query: mentions (your hotel was named), direct citations (AI linked to your own website), and indirect citations (AI linked to a third-party source — a Booking.com property page, a TripAdvisor listing, a travel guide). Each tells a different story about how deeply AI has embedded your property into its responses.
But the most useful part of the report isn't the numbers. It's what they reveal when you look at them in the right structure.
The Gap That Should Worry You Most: Branded vs. Non-Branded
Every report separates queries into themes. Branded queries simulate someone who already knows your hotel — searching by name, asking for comparisons, reading about it specifically. Non-branded queries simulate discovery: "best luxury beach resort in Crete", "where to stay near Heraklion for a family", "romantic hotels on the Aegean coast".
For most hotels, the gap between these two is striking — and strategically important.
In a luxury property we analysed in Crete, branded performance was strong across all three platforms. The hotel scored 391 on Perplexity in branded queries, appearing 166 times with 375 citations. When AI was asked about this hotel by name, it had a lot to say.
Then non-branded: city queries scored 31 on Perplexity, 1.5 on OpenAI, 12.6 on Gemini. Non-branded POI and region queries were effectively zero across all three platforms.
What this means strategically: AI knows this hotel but doesn't reach for it unprompted. The property exists inside the answer to "tell me about Hotel X" — and largely doesn't exist inside the answer to "where should I stay in Crete?" That's a fundamentally different kind of visibility problem than anything a review score or a rate parity tool can detect.
The actionable question: For every non-branded theme where your score is low, ask what's driving the citations you do have. Direct citations mean AI is reaching your own website — that's the strongest signal. Indirect citations show you which third-party platforms AI is drawing from when it mentions you: OTAs, review sites, travel guides. Understanding that split tells you where your presence is solid and where it's dependent on platforms you don't control.
Are Your Personas Working for You?
The report is built around traveller personas — and these aren't generic. For each hotel, Tharro generates the personas that best match that property's positioning, market, and guest profile. For the Crete property we analysed, that meant Luxury-seeking Couples, Affluent Family Vacationers, Honeymooners, Retired Luxury Travellers, and Wellness Enthusiasts. A city business hotel would get an entirely different set. Each persona generates its own query set, and you see your visibility broken down by persona, across themes, across platforms. This is where the report becomes genuinely diagnostic.
In the Crete analysis, Wellness Enthusiasts showed a branded score of 83 on Perplexity — when wellness was part of the query and the hotel was named, AI responded well. But in non-branded wellness queries, the score was 0. A competitor captured those mentions instead, consistently.
Here's the strategic question this raises: is Wellness Enthusiast a persona you're actually positioning for, or one you're assuming you should own?
If wellness is a core product pillar — dedicated programming, spa packages, a genuine content story around it — and you're not appearing in non-branded wellness queries, that's a gap worth closing. The work is building the content signals that AI can learn from: detailed property pages, editorial coverage that positions you in wellness-specific language, responses on review platforms that emphasise that dimension.
If wellness is an amenity rather than a positioning, the report is correctly reflecting your market presence — and the question becomes whether you want to invest in changing that, or whether you reallocate that energy to the personas you actually win.
The actionable principle: Use persona performance to audit your own positioning claims. Every persona where you score well branded but poorly non-branded is a persona where you have a reputation but not a discovery presence. Close that gap deliberately, not reflexively.
Competitive Intelligence You Can't Get Anywhere Else
The competitive layer of the report shows which traveller types your competitors are capturing in AI discovery — by platform, by persona, by theme.
In the Crete analysis: one competitor consistently dominated Honeymooners across all three AI platforms, appearing in comparison queries, non-branded city results, and direct competitor-branded searches. Score of 42 on Perplexity for that persona alone. Another competitor owned Affluent Family Vacationers in Perplexity's non-branded results — 37 mentions, score of 82. A third captured Wellness Enthusiasts in comparison contexts on Gemini, with 31 mentions where the subject hotel had none.
What this means strategically: If a competitor is consistently appearing alongside you in comparison queries, AI has learned to treat you as alternatives in the traveller's consideration set. That's your real competitive pairing — which may not match who you track on pricing tools. And if a competitor is winning a persona you want to own, that's not a mystery to accept. It's a researchable question: what content, positioning, or coverage has led AI to associate them with that traveller type? That answer points directly to what you need to build.
The actionable implication: Don't just look at your own scores. Look at where competitors are pulling away from you in specific persona / theme combinations. Those intersections are where you're losing the consideration of a defined traveller type in a defined search context — and that's specific enough to act on.
The Framework for Reading the Report
Start with the branded / non-branded ratio. A strong branded score with weak non-branded means you're well-known to people who've already chosen you and poorly visible to people still deciding. In a market where AI is increasingly part of early trip planning, that's an exposure worth taking seriously.
Then move to the persona layer:
- Strong branded, weak non-branded on a persona you care about → discovery gap. Build the content signals that close it.
- Weak across both branded and non-branded on a persona → decide deliberately whether this audience is worth pursuing or whether your resources belong elsewhere.
- A competitor outperforming you on a persona you want to own → trace their coverage, understand why AI has learned to associate them with that traveller type, and build a response.
Finally, use the citation breakdown. High indirect citations mean AI is finding your hotel through third-party sources — OTA listings, review platforms, destination guides. High direct citations mean AI is linking to your own website directly. The second is harder to earn but signals a stronger, more durable authority with the model.
This is also where Tharro's Opportunity feature comes in. For the gaps identified in the report — the queries where your hotel didn't appear — we go back to the AI and ask why. We collect those answers across all the missing queries, find the patterns in the reasoning, and turn them into concrete recommendations. AI is often surprisingly explicit: it will tell you it didn't recommend a property because it couldn't find sufficient content about a specific attribute, or because another hotel had stronger coverage on a particular theme. That feedback, aggregated across dozens of queries, becomes a prioritised action list rather than an open-ended research project.
The report tells you where you're invisible. The Opportunity feature tells you what the AI would need to see in order to change that.