We Analyzed AI Visibility for 20+ of the World's Largest Airlines. Here's What We Found.

Every major airline in our study is being named by AI models. Almost none of them are being linked to.

71% average mention rate. 9.9% average citation rate. 61-point average awareness gap. 20+ airlines analyzed.

We ran 1,260+ individual AI searches across ChatGPT, Claude, and Perplexity, covering 20 queries per airline across five topic categories: airline rankings, loyalty programs, premium cabins, passenger experience, and routes and hubs. What we found is consistent enough across carriers that the pattern is hard to ignore.

AI knows who you are. It just doesn't send people to your website.

That gap, between being mentioned and being cited, is the defining problem of airline GEO in 2026. And it's costing you traffic, attribution, and influence over the customer journey at exactly the moment when travelers are making high-consideration decisions.

Every airline in our study has a massive gap between brand mention rate, meaning how often AI names the airline, and URL citation rate, meaning how often AI links to the airline's own website. The industry average is 71% mention rate against 9.9% citation rate. That's a 61-point gap, on average.

The most extreme case: a leading Gulf carrier achieves a 95% mention rate and a 3% citation rate. A 92-point gap. AI models recommend that airline in nearly every relevant query. They link to that airline's website in almost none of them.

This isn't a niche problem for a handful of carriers. It holds for the largest US airlines, the European legacy carriers, the Gulf carriers, and the Asia-Pacific players. The most-recommended US domestic airline still posts only a 12% citation rate despite a 70% mention rate. The gap is structural, not incidental.

When AI models do include a citation, it's rarely the airline's own domain. The citation slots are being taken by The Points Guy, NerdWallet, Simple Flying, One Mile at a Time, Skytrax, Wikipedia, and Reddit.

Here's what that means in practice: your airline is building brand awareness through decades of marketing spend. The AI model recognizes that and mentions you by name. Then it links to a travel blogger's review of your business class seat as the authoritative source. You got the mention. Someone else got the click.

The third-party sites winning citations aren't winning on brand. They're winning on content structure. They publish authoritative, question-answering content with clean extractability. They answer exactly what the traveler is asking. Airline websites, by contrast, are built to convert, not to inform. And AI models can tell the difference.

Not a single loyalty program page earned citations in our study. Across every carrier we analyzed, covering programs that collectively represent hundreds of millions of enrolled members, the AI models did not cite a single airline's own loyalty content. Not once.

Personal finance sites and travel bloggers own this space entirely. When someone asks which frequent flyer program is best for a transatlantic trip, the AI gives them an answer and cites a NerdWallet comparison or a Points Guy breakdown. The airline's own program page, with all its official rates and benefits, doesn't make the cut.

This matters because loyalty is a high-consideration query. Someone asking that question is close to a booking decision. Losing that citation doesn't just cost you a click. It costs you influence at a critical moment in the journey.

Airlines are spending billions developing and marketing premium cabin products. The AI narrative around those products is being written by bloggers.

One carrier's flagship business class product is ranked #1 by all three AI models we tested. That product is genuinely best-in-class. The carrier's website still only achieves a 3% citation rate on queries where that product is the answer. The citations went to review sites and aviation publications that wrote about the seat.

The content gap here is specific: airlines publish marketing copy about their premium products. What AI cites is comparative, factual, and structured. Seat dimensions, Wi-Fi specs, meal service details, real customer experience data. The airlines have all of this. They're just not publishing it in a format that AI can extract and cite.

Here's where we found some signal in the noise: carriers that publish authoritative hub-specific content see higher citation rates on route-specific queries. An airline known for a particular hub tends to get cited when someone asks about flights from that hub. The route knowledge is there. The content infrastructure to surface it isn't.

Broader route queries, such as "best airlines for transatlantic flights" or "cheapest routes between major hubs," remain almost entirely unclaimed by airline domains. Every carrier we studied has gaps on these. For a query category that maps directly to booking intent, that's a significant miss.

Most airline marketing teams already know they have an AI search problem. What they often don't know is that a meaningful portion of the fix is technical and executable.

JSON-LD structured data was missing or invalid on the vast majority of airline sites we reviewed. Content extractability scores were consistently low. Canonical URL issues created ambiguity about which pages AI should treat as authoritative. These aren't brand or creative problems. They're infrastructure problems.

The good news: these are fixable. The less good news: fixing them requires someone to actually do the work. That means auditing hundreds of pages, updating structured data, rewriting content for extractability, and pushing those changes through whatever CMS and approval chain the airline runs. For most teams, that's the bottleneck.

GEO monitoring needs to be multi-model from day one. A carrier might be cited consistently on one AI platform and nearly absent on another for the same query. A carrier we studied earned strong citation rates on OpenAI for route content, while Claude and Perplexity results looked completely different.

Single-model monitoring gives you a partial picture. The strategic decisions should be based on aggregate performance across ChatGPT, Claude, and Perplexity, because those are the three platforms currently driving AI-influenced travel research at scale.

The pattern across all 20+ airlines points to the same set of fixable gaps:

Loyalty program content. Publish authoritative, FAQ-style loyalty content that answers real traveler questions. Comparison tables. Point value calculators. Redemption guides. The blogs are already writing this. Your official pages should be better.

Premium cabin specs. Publish the seat specs, Wi-Fi details, dining options, and service comparisons in structured, extractable formats. If the blogger review gets cited instead of your page, the blogger's content is more extractable than yours.

Route authority pages. Build hub-specific content that answers real route questions. AI rewards geographic specificity. Own your hub.

JSON-LD structured data. Audit and fix your structured data. This is the highest-leverage technical fix with the clearest path to improved citation rates.

Content extractability. Review your highest-priority pages for extractability. Clean page structure, clear headings, minimal JavaScript dependency on primary content, and valid canonical URLs all matter.

The visibility report tells you where the gaps are. Closing them requires execution at scale, across dozens or hundreds of pages, through your existing CMS, against real approval timelines. Most teams don't have the bandwidth to run that continuously. That's exactly the problem Gradial's GEO Agent is built to solve.

Want to know where your airline stands? Request a free GEO report at gradial.com/geo-analysis. We'll show you your mention rate, citation rate, which third-party sites are eating your traffic, and where the highest-leverage fixes are.

Gradial is the system of work for marketing. Not a visibility dashboard. An execution layer that connects to your existing CMS and content supply chain, runs GEO fixes automatically, and keeps your content current across AI models. From brief to live.