The Pricing Fight Inside Every Airline

From the Travel Tech Podcast newsletter for episode 35, The Pricing Fight Inside Every Airline.

Everyone predicted AI chat would kill metasearch. Ask a chatbot for the cheapest fare to Europe, and why would you ever open a comparison site again? This week's guest measured what actually happens and found the opposite: when people ask AI chats flight questions, metasearch sites dominate the answers. In TravelDax's report, Google Flights appeared in 54% of relevant ChatGPT answers, with Kayak and Skyscanner close behind at around 48% each. Valentino says metasearch is doing better in AI than it does in traditional search.

Valentino Portnoy is CEO and founder of TravelDax, which tracks flight prices across metasearch for airlines and OTAs. He built a flight search with price prediction as a university project in Argentina before he knew it was called metasearch, went on to build one of Latin America's first metasearch engines, then worked inside an OTA and consulted for airlines before launching TravelDax seven years ago. Listen on Apple Podcasts, also streaming on Spotify and YouTube.

The competitor airlines actually watch

His second finding surprises people too. Roughly 75 to 80% of airline activity on TravelDax goes on tracking OTAs, not rival carriers. Part of that is price: on a metasearch results page, an airline fare sitting below five, six or seven cheaper OTA offers is effectively invisible. Part of it is presentation. Airlines check whether OTAs show their branded fares the way they intended, because a fare shown without the product behind it loses customers too.

Why two euros is a speed problem

The gaps he sees are often one, two or three euros. That doesn't mean revenue management got it wrong; it means they were very close. With every airline monitoring competitors in some form, the edge goes to whoever adjusts first, which is why Valentino would rather be wrong more often and fast than right and slow.

That is where it stalls. TravelDax had the whole loop, from data collection to price change, automated about six months after launch, yet many customers automate only around 70% of it. The last 30% runs into legacy systems he describes as 30 or 40 years old, and into organizations where revenue management, direct and OTA teams carry different targets and bonuses, and revenue management usually has the final word.

Gen AI isn't the fix for the math. When TravelDax gave language models its own data and compared their price predictions with its machine learning models, machine learning came out ahead. Where he does see AI helping is that last 30%: an agent on the analyst's own computer, doing in the legacy tool what the analyst does by hand today. The technology exists, in his view. The open question is whether airlines will trust it to change fares with no human in the loop.

A warning for anyone building connectors

We also got into Meta's Muse and its Duffel partnership, Google moving toward booking inside the chat, and why he doubts flights will ever be a priority for the big AI labs: the margins are too thin next to what they can earn elsewhere. My own view, which he agreed with: the labs can see which travel connectors get used, and as with the App Store, the best performers tend to get absorbed into the platform. Build them, but build with that in mind.

Key Takeaways

• AI is feeding metasearch, not replacing it: Metasearch sites dominate AI chat answers to flight questions, and they fare better there than in traditional search.
• OTAs are the real competitor set: Most airline monitoring targets OTA undercutting and whether branded fares are displayed as intended, not rival carriers.
• Speed is the competitive edge: When the gap is a couple of euros, the airline that adjusts first captures the sale, even at some cost to accuracy.
• Agents fit the legacy gap: Where old systems block integration, an agent repeating the analyst's manual steps is the practical route to full automation, if airlines will trust it.
• Connector builders are being watched: AI labs can see which travel connectors succeed, and the best performers risk being absorbed into the platform.

Alex Brooker, Host of the Travel Tech Podcast, Founder of Airside Labs