In a recent article, we analyzed how ChatGPT citation links compared to Google’s search results for flight searches. We found significant but incomplete overlap between ChatGPT’s sourced URLs and Google’s top organic search results. While major OTAs, flight aggregators, and airlines dominate in ChatGPT, it sourced domains rarely found in Google’s top 10, like Wikipedia and FlightConnections.
In this article, we cover a critical finding from that same research: a significant ‘mention-to-link mismatch’. As we’ll show, ChatGPT often discusses airlines but awards the inline citation to a different website. This isn’t a random error; it’s a pattern that disproportionately harms airlines.
Methodology Recap
Our full methodological approach is detailed in our previous article in this series. In short, we used a controlled and repeatable process to analyze ChatGPT’s foundational linking behavior. The key elements were:
- Keyword Corpus: A representative sample of 376 standardized “flights from ORIGIN to DESTINATION” prompts.
- ChatGPT’s Version: ChatGPT-5 in Fast mode.
- Session Isolation: Each prompt was run in a fresh chat with Memory disabled.
- Data Captured: We logged every sentence with inline citations, the brands mentioned, and all cited domains. The data was later processed and analyzed.
- Link Placement: In this article, we focused on inline citations, which are superscript gray circles or bubbles rendered immediately after the sentence they support. They show a numbered mini-card with all the source names, page title, domain, and a short snippet.
The Core Finding: 42% of Brand Mentions Don’t Get the Link
Our central finding is the stark disconnect between brand mentions and citation domains. In sentences that contained an inline link, 41.8% of company mentions were not linked to that brand’s domain. Instead, ChatGPT linked to a third-party source.
| Link Status | Proportion |
|---|---|
| Mention was NOT Linked | 41.8% |
| Mention was Linked | 58.2% |
In other words, when ChatGPT referred to flight-related information from specific airlines, OTAs, aggregators, and others, roughly four out of ten times it linked to websites from other companies.
Airlines Are Disproportionately Affected
This mention-to-link mismatch problem is not distributed evenly. Airlines are by far the most negatively impacted, with an Unlinked Rate of 74.6%. This is nearly nine times higher than the rate for OTAs (8.8%) and Flight Aggregators (8.6%).
| Organization Type | Total Mentions | Unlinked Rate (%) |
|---|---|---|
| Airline | 2,031 | 74.6% |
| OTA | 1,065 | 8.8% |
| Flight Aggregator | 1,059 | 8.6% |
Given that ChatGPT most frequently covered flight information from specific airlines, this high Unlinked Rate represents a massive loss of potential referral traffic from their own brand mentions.
Where Do the Misdirected Links Go?
When ChatGPT mentions an airline but links elsewhere, it most often directs users to OTAs and aggregators.
| Domain | Share of Substitute Domains (%) |
|---|---|
| kayak.com | 13.5% |
| expedia.com | 11.3% |
| cheapflights.com | 7.4% |
| flightconnections.com | 6.9% |
| skyscanner.com | 6.3% |
| en.wikipedia.org | 5.2% |
| flights.com | 4.8% |
| travelocity.com | 4.8% |
| momondo.com | 3.6% |
| flightroutes.com | 3.4% |
Kayak.com, Expedia.com, and Cheapflights.com were the top three domains to receive these “substitute” links, effectively capturing traffic from conversations about their direct competitors.
Topic Analysis Reveals When Airlines Lose the Link
To understand the context behind these mention-to-link mismatches, we analyzed the topics that ChatGPT discussed in sentences where it mentioned airlines.
The two most common topics were “Layover & Connections” (775 mentions) and “Fare Pricing” (749 mentions). Both topics showed extremely high Unlinked Rates, at 73.5% and 65.8% respectively. This means that in the two contexts where users are most likely to see an airline mentioned, roughly two-thirds of those mentions link to a different website. And the pattern continues across the rest of the top five topics.
| Topic | Brand Mentions | Unlinked Rate (%) |
|---|---|---|
| Layover & Connections | 775 | 73.5% |
| Fare Pricing | 749 | 65.8% |
| Flight Frequency & Scheduling | 269 | 56.9% |
| Specific Flight Dates | 244 | 51.6% |
| Flight Duration & Distance | 106 | 69.8% |
These findings show that across the five most common topics in which airlines are mentioned, the best-case scenario is a 50/50 chance of getting the link, with an Unlinked Rate as high as nearly 74%.
Six Reasons Airlines Don’t Get the Click
Why does ChatGPT disproportionately link away from airlines, even when discussing their own routes and fares?
While our next article will provide a detailed root-cause analysis, our preliminary investigation points to several potential factors contributing to the high Unlinked Rate for airlines:
- Non-Existent Pages: The specific origin-destination pages that ChatGPT expects to source from an airline may not exist at all.
- Low Topical Coverage: Key topics sourced by ChatGPT, like detailed flight information (e.g., schedules, connections) or booking insights (e.g., cheapest days to fly), may be absent on the airline’s pages.
- Incomplete Fare Coverage: Pages may lack distinct, indexable prices for different itinerary types. For example, if a page doesn’t show fares for one-way, round-trip, direct, and connecting flights, ChatGPT will likely prefer an aggregator that presents this data more comprehensively.
- Missing Fares: Some airline flight pages may not display any price information, forcing the model to cite a source that does.
- Crawler Accessibility Issues: Airlines may be inadvertently blocking or limiting access for ChatGPT’s bots (ChatGPT-User, OAI-SearchBot), preventing the model from successfully parsing their content.
- Inherent Citation Bias: Beyond technical or content gaps on airline websites, ChatGPT may also display a tendency toward citing OTAs and aggregators. We are still investigating why these domains are more “attractive” citation targets for ChatGPT, even when equivalent information is available and accessible directly on airline pages.
Conclusion: Mentions Aren’t Enough
While ChatGPT overwhelmingly mentions airlines in most flight-search responses, it often does not provide a link to their websites, even for fare prices that airlines publish directly on their flight pages.
The data indicates the mismatch is not random but stems from a mix of airline website limitations and ChatGPT’s citation behavior. Non-existent pages, missing or limited fare coverage, incomplete flight information, and crawler accessibility issues may prompt the model to rely on third-party sources. Additionally, ChatGPT shows a systemic bias toward OTAs and aggregators, amplifying the loss of referral traffic for airlines.
In practice, this means high-intent users who read about an airline in ChatGPT are more often funneled to OTAs and aggregators. Unless airlines adapt, they risk losing a growing share of referral traffic from this channel.
Frequently Asked Questions
ChatGPT often mentions airline brands in flight-related answers but links to third-party sites like OTAs or aggregators instead of the airline’s own website. This mismatch results in lost referral traffic for airlines.
Airlines have a 74.6% Unlinked Rate, nearly nine times higher than OTAs and aggregators. This is due to missing fare data, limited page coverage, and crawler accessibility issues on airline websites.
When ChatGPT links to OTAs instead of airlines, it can potentially divert high-intent traffic away from the airline’s domain. This undermines SEO efforts and reduces direct bookings, impacting revenue and brand visibility.
The most affected topics are “Layover & Connections” and “Fare Pricing”, where over two-thirds of airline mentions link to other domains. These are key decision-making moments for travelers.
Airlines can optimize their site structure, ensure fare data is indexable, improve topical coverage, and allow OpenAI’s crawlers access for AI bots like ChatGPT-User and OAI-SearchBot. These steps increase the likelihood of being cited directly.
airTRFX creates scalable, SEO-optimized pages for every airline route and fare type. By ensuring comprehensive coverage of flight combinations, fares, and page content, airTRFX helps airlines reclaim lost traffic and improve citation rates in AI-generated content.
