How Does AI Impact Metasearch? Solving the 1 Million Look to Book Problem

If you read the latest interview with Skyscanner CEO Bryan Batista, you’ll see an ambitious vision. He wants to move from price-sorting to an "Inspiration Engine" that builds complex, multi-modal trips. Sounds lovely but it creates a technical paradox: The more we inspire travelers to ask "Why," the more we break the legacy systems that can only answer "Where and When."

For the last twenty years, metas thrived by catching simple Google searches and steering them to Airlines and OTAs. But inspirational search can be expensive. Every time an AI suggests a complex trip, it triggers an avalanche of back-end queries.

The current infrastructure wasn't built for this. We are shifting from Visual Comparison to AI Resolution. As this scenario unfolds, the distribution economics as we know it are simply collapsing.

What is the current state of NDC adoption in 2026?

There is a lot of hype about AI agents booking travel instantly. The reality is a bit more complex. Currently, global NDC adoption (New Distribution Capability)—the Tech that allows Airlines to sell products dynamically—is still below 20-25%. Most airlines still can't handle a complex booking through an API without their internal systems melting down.

This gives metasearch most likely a 24-to-36-month transition grace period. While AI can't always book perfectly yet (although this is changing rapidly as well), it is certain that it's already disrupting how people search and plan. The window to pivot is real, but the clock is ticking.

The harsh reality of the "Inspiration Paradox"

Why 1,000,000:1 Look to Book is breaking the Air Travel Industry

As noted by Sebastien Gibergues in his last PhocusWright article, the biggest "infrastructure debt" in travel tech today is the Look-to-Book (L2B) ratio. When a human shops, they might look at 20 sites before buying a Flight or a Room night. When the same traveler uses a Metasearch platform, a single route/price query gets multiplied into thousands of L2B to Airlines, OTAs (compound effect). Now, when an AI agent shops, this number can theoretically jump to 1,000,000:1 and above as the agent will monitor the flight combinations 24/7.

I saw the early version of this during my time working at Cleartrip with a UAE based Low Cost Carrier. Our contracts had "hard caps" on search volume; if we triggered too many calls, then the Airline API was going mute.

It matter because Airline’s internal system wasn't built for a million searches per booking. From an Airline's perceptive, an AI agent almost looks like a a non-productive traffic overflow, meaning a flood of traffic that slows everything down and costs a fortune in server bills.

How Metasearch Actually Survives Today?

Airlines and Metasearch are tired of paying for unproductive searches. To stay connected, big metasearch engines have had to change their business model. They now pay IT companies like Amadeus millions of dollars a year for "precomputed" data feeds.

Instead of asking the airline for a live price every time, the metasearch engine checks a giant, pre-calculated "cache". They only call the airline at the very last moment to confirm the seat availability. In this scenario, the metasearch is optimizing its L2B ratio for ROI purposes. At the same time, it absorbs the million-query AI storm so the airline's fragile systems don't crash.

The AI Travel "Operating System"

The real threat to Skyscanner or Kayak isn't just AI Assistants; it’s the shift in how data moves. Amadeus and Sabre are now building ways to feed their giant databases directly into AI models using Model Context Protocol (MCP). If an AI agent can talk directly to the GDS (the main data source) to get a verified schedule or price, it doesn't need to visit a metasearch website. This turns the GDS players into "source of truth" of travel AI, potentially making metasearch sites less relevant. Note that GDS companies are willing to cannibalise their metasearch revenue because they would rather charge AI agents high-margin fees for "truth" than manage millions of low-value, server-crushing window shoppers.

Metasearch " I gave you 20 years of my life" >> GDS: " Yeah but she has MCP endpoints"

Obviously, as Bryan Batista says, leading metasearch players are not staying passive; they are actively re-platforming to secure their survival in an Agentic world. By building proprietary MCP-compliant endpoints, they are positioning themselves as an essential "price integrity" filter that prevents AI models from "hallucinating fares". This represents a strategic pivot from being a simple discovery site to becoming an indispensable orchestration layer for AI assistants.

Simultaneously, the industry is moving away from the fragility of old commission models toward "Resolution Fees"—charging for a solved, guaranteed itinerary rather than a simple click. Furthermore, we are seeing aggressive moves toward vertical integration, where metas seek to own the actual inventory or the liability layer to capture higher margins. This isn’t just a tech update. It’s a bold move to take back control of their own future.

The New Buyers: Why Banks are Winning

If metasearch sites need to rethink their survival, who is going to buy them? It likely won't be other travel companies. Could it Banks or Fintech players? The recent Capital One’s acquisition of Hopper’s technology sets the tone.

If an AI makes a mistake and books the wrong flight, a bank can afford to fix it or offer a "Price Guarantee", this is the Liability layer.

Banks want consumers to stay inside their app and to use their credit card points. Buying a metasearch engine is a cheap way for a bank to get millions of high-spending travelers to use their portal instead of going to Google, this is the Loyalty layer.

The 2026 M&A Landscape

We are moving away from "independent" search sites. By the end of 2026, most metasearch brands will likely fall into one of three buckets:

• Fintech Acquisition: By owning a Meta, Banks and Fintech could turn "Inspiration" into a closed-loop financial ecosystem where they can offer "Instant Refund" or "Price Protection" as a feature of the card, not a travel product.

• AI Acquisition: An AI company wouldn't buy a Meta for the brand; they would buy it for the MCP infrastructure. They need access to a clean and curated Travel data to make their AI Agents reliable.

• Social Media Acquisition: TikTok and Instagram do not want Users to leave their app to go to Google. If they acquire a Metasearch ," they could offer "One-Tap Booking" directly from a video. The Meta provides the real-time price integrity (the MCP layer) so the users can book without a "Hallucinated" price.

  
  

The face you make when the GDS are building MCP endpoints...

Final Thoughts: From Discovery to Resolution

The metasearch business model is being squeezed from both ends. On the one hand, Airlines are throttling "unproductive" traffic. On the other hand, AI agents are looking for data sources that don't require a web browser.

The next 24 months won't be a "disappearance," but a sorting period. Either they rebuild their business as the "brain" for AI assistants. Instead of earning money from website clicks, they charge for providing the accurate data needed to solve and book complex trips. Or they decide to sell their company while their data and user base are still worth a lot. By joining a bank or a major tech platform, you become the essential "plumbing" for their apps, securing a high price before independent search websites loose their relevance.