A new white paper Speed of Data: The New Airline AI Arms Race looks at how Delta Air Lines is using AI in pricing, and how other airlines are likely to respond.

Delta Air Lines has courted controversy by using AI to price its tickets – 1% of fares back in the fall, 3% of fares at the time of its earnings call in July, and with a plan to hit 20% by year-end.

But they say they’re completely misunderstood. They aren’t doing ‘personalized pricing’ which is scary, they’re just automating. So what are they actually doing, and why is this so confusing?

Here’s What Delta Is Doing Today

Delta partners with tech firm Fetcherr to link a transformer‑based model directly into fare filing and price setting so prices update near-continuously. Fetcherr says this delivers a 10% increase in revenue. Delta calls the tech a “super analyst.”

Fetcherr’s tech predicts the next number (price) instead of the next word. A GPT‑style transformer ingests inventory, historical fares, competitor prices, events, weather, and all the information that should influence decisions to generate an optimal price and push it live in seconds. That’s “top analyst‑quality decisions at machine speed.”

Airlines have had sophisticated models for years, but humans gated the last mile. What Delta and Fetcherr are currently doing is removing the human bottleneck, so frequency of repricing becomes the competitive edge.

Why The Backlash?

The paper claims that this tech has been conflated with personalized “dynamic pricing,” but that the model does not use customer‑specific data – it’s classic RM data, just faster. That’s no doubt true today.

So when Delta was asked by Senators to address their plan to use information about individual customers to figure out how much they’ll pay, Delta responded that they aren’t doing this at all. They’re really just using computers to help their analysts become more efficient.

Where could we all have gotten the idea that Delta “will have a price that’s available on that flight, on that time to you, the individual”? From Delta’s President. Glen Hauenstein shared at their November 2024 Investor Day that they aren’t there yet but what they are working towards is offering specific prices on specific flights to specific customers. This is called “offer management.” (Emphasis mine.)

[T]his is again a full reengineering of how we price and how we will be pricing in the future.

And if you think about it today, there are two disciplines. There is pricing, which sets the price points and then there’s revenue management, which controls access into the inventory of those price points.

And over time, we think this is going to get melded together, that it’s going to be really just offer management. That we will have a price that’s available on that flight, on that time to you, the individual. Not a machine that’s doing an accept reject and a static price grid.

There’s a fear that someone might have to travel to a funeral, or pay any price to visit a dying loved one, and Delta will know that. The thing that constrains their ability to charge a premium, though, even if they did know a given customer’s actual demand is competition. That’s what makes competition so important, and why governments should stop blocking competition and subsidizing incumbent carriers.

Nonetheless, Delta isn’t doing this today, so Delta prefers to speak publicly about what they are currently doing – while telling investors that they are going to do something much more in order to convince them they will earn more in the future and thus warrant a higher stock price.

Where Is AI Pricing Going?

Delta is very clear that they see the future as using AI to target individualized prices at specific consumers. But they aren’t there yet.

We’re going to see a short‑term arms race in the industry. Other airlines have to deploy comparable automation or accept a pricing‑speed disadvantage. The hardest part of this is data plumbing. Delta has already spent about a decade unifying data. And other airlines are likely far behind. That gives Delta a temporal advantage.

Several airlines use Fetcherr including Azul, Virgin Atlantic (49% owned by Delta), Westjet (15% owned by Delta) and Viva Aerobus. Azul was actually first to acknowledge it publicly in 2022.
Fetcherr isn’t the only one in this space. JetBlue partnerd with FLYR. Virgin Atlantic did, too and reported a 10% increase in seat fees priced dynamically. flydubai is using PROS and Lufthansa announced a partnership with them as well.

Ultimately we’re going to see a lot of incremental progress before we get to anything like the kind of ‘offer management’ that Glen Hauenstein talks about. First they’re going to have to address system underperformance on off‑peak days (discounting too much), potential “flash‑crash”‑style price spirals when algorithms react to each other, and what kind of human supervision is required and even possible when fully turning decisionmaking over to the machine.

The Tool Will Decide When To Offer The Lowest Prices – Not Which Individuals To Gouge The Most

What’s mostly being misunderstood is that AI is likely to lead to greater confidence offering lower prices to specific customers, rather than raising prices on individuals (who at least currently can buy tickets without logging in at Delta.com, but at Expedia, or buy from another airline).

The way airlines generate a lot of their incremental revenue isn’t just figuring out whom they can charge more to, it’s to figure out how to charge less to new customers without offering those same low prices to current customers who are buying tickets anyway at higher prices.

• Airlines want to charge high fares to price-insensitive business travelers
  
• And they want to charge low fares to attract price-sensitive leisure travelers, to the extent they have empty seats to fill (since those fares are almost all incremental profit)
  
• But they don’t want the low fare offered to a leisure traveler to be available to that business customer who would have bought at a higher price.

Personalized pricing has been an airline goal for the past decade and a half, and one of the ideas behind ‘new distribution capability’ (“NDC”). But the trick is simply to figure out whom to offer the lowest fares to! And AI can help them offer more low fares to more people without cannibalizing high fare revenue (as well as avoid offering those low fares and selling too many seats that customers will buy later at higher prices).

Airlines long used Saturday stays and 14- or 21- day advance purchase requirements to segment business travelers (who buy last minute and don’t stay away the weekend as much). This is now accomplished with basic economy (meant to offer low fares to price-sensitive leisure travelers) and even offering the lowest fares only when more than one passenger is on a reservation (because business travelers usually book solo). AI can move towards less blunt instruments to accomplish this goal.

I’ve been trying to understand why American Airlines CEO Robert Isom blasted Delta for a “breach of trust” during his airline’s earnings call. Maybe it was just ‘sure we underperform financially but at least we do it ethically’ or maybe he wanted the story to be something other than his airline’s performance. But it also may be offering an early excuse for why they’ll lag Delta in rolling out continuous pricing – using ethics as a substitute for technical capability.

But make no mistake, this is spreading – even to American. American’s Vice Chair and Chief Strategy Officer Steve Johnson mentioned upside from “the opportuntiies to deploy AI in revenue management” at the Morgan Stanley Laguna Conference on Thursday.