Does generative AI make airlines invisible?

You might think airlines are the obvious source for flight questions. AI assistants seem to disagree and surface information from other players instead.

Bain & Company and Profound analysed, for ATPCO, 12,000 prompts to AI assistants (ChatGPT, Google AI Overview, Google AI Mode, Gemini, Perplexity and Copilot) and 64,000 URLs cited by those tools in August 2026.

The study indicates that only about 6% of those results lead to an airline’s own site. The report adds that online travel agencies are cited roughly three times more often than airlines.

Metasearch engines are the most frequently linked among outbound sources. One site alone accounts for 16.3% of citations, though the report does not say whether this is Skyscanner, Google Flights, Kayak or another comparator.

The research highlights a paradox: the primary source of flight information, held by the actor with the most accurate data and the most complete offer, is referenced significantly less by LLMs than third‑party channels.

Several hypotheses help explain this relative invisibility of airlines.

OTAs and metasearch have structured, aggregated data in unified formats that LLMs can consume more easily. Far from being displaced by LLMs, these intermediaries often make their work easier.

By contrast, airline data structures, long shaped for GDS and rigid, criteria‑based search, do not readily lend themselves to conversational queries.

Many intermediaries have also moved quickly to serve this new discovery channel, Expedia and Booking.com have developed a module for ChatGPT. In May 2025, Skyscanner announced a partnership with OpenAI.

This LLM preference for second‑hand information has consequences. For airlines, the near‑mandatory presence of an intermediary squeezes margins via commissions and weakens control of the customer relationship.

For travellers, the report flags risks of inaccuracy or bias, such as poor product descriptions due to enforced uniformity, for example, a premium economy cabin labelled as business class, or the omission of some results.

AI agents are not only a relay for travellers who know exactly what they want.

Natural‑language discovery opens a broad, inspiration‑led field that helps explain OTA and metasearch investment.

LLMs need actors that can understand vague, open‑ended requests, and those actors in turn need airlines’ precise, structured data.

Should airlines adapt to the possibilities of this new distribution channel? “Airlines do not own every part of the inspiration journey; that was true before AI as well. But we believe LLMs create an opportunity for airline products to be a more meaningful part of that journey.” said Kristine Jenkins, corporate strategy at ATPCO.

“For example, a traveler looking for a relaxing trip might consider destinations where a business-class experience is within budget, or where nonstop service, seat comfort, or other product attributes better match their preferences.”

“Previously, a traveler might have needed to narrow down a destination first to find information on those options.”

“With LLMs, that information can be part of the discovery process - provided that airline products are accurately findable, understandable, and comparable.”

ATPCO therefore urges airlines to invest to strengthen their presence across these channels.

First, airlines should provide structured, standardised, machine‑readable data.

They also need a semantic interpretation layer so natural‑language queries such as “widest seat”, “lounge access included” or “non‑stop flight with workspace” map cleanly to the structured data.

Finally, ATPCO encourages airlines to equip themselves with tools to monitor brand presence on these channels, measure visibility and ensure the accuracy of information relayed.