Hubert here. Apologies if this is a dupe but earlier email got bounced. Can big airlines really achieve meaningful increases in unit revenue by algorithmically figuring out how much each individual carrier is willing to pay for a ticket?
Multiple articles quoted Delta executives telling investors that they could achieve big gains this year using “artificial intelligence” based systems to “get inside the head of consumers.” https://viewfromthewing.com/several-airlines-now-quietly-let-ai-set-ticket-prices-surprisingly-thats-great-news-for-your-wallet/ was recently posted here. Few articles have really focused on what a radical change this would represent and many have swallowed the industry PR claim that this is just like the “price discrimination” they’ve been doing for decades. I can’t think of another consumer industry that has studied and applied “price discrimination” more aggressively than airlines. But shifting from “prices vary depending on things like time before departure but everyone making a specific request at a specific tie will see the same fare based on fare rules that are totally transparent” to “we can show every user a different price based on logic no one will be allowed to see” would be a more radical change than anything since deregulation. A rational (but cynical) explanation is that Delta figured out that any press release using the words “artificial intelligence” would quickly boost their share price. Or perhaps Delta management actually believes all the AI hype. We need explanations on how these radical new systems would actually work from outsiders who don’t have a vested interest. The “price discrimination” at the heart of traditional airline revenue management drove major legitimate efficiency gains. It increased revenue from peak-demand yields while also offering consumers lower off-peak/advance purchase fares, driving big improvements in capacity utilization. Will AI drive the next big airline pricing breakthrough or is this mostly hot air? Will gains come from true efficiency gains or just from the ability to raise fares? Anyone seeking to provide answers will need to address a variety of issues, including: 1. A major portion of airline consumers and agents—who are exactly the “less price-sensitive” travelers being targeted here—are extraordinarily savvy about how airline pricing works, are capable of figuring out major pricing changes and would be react angrily if it seemed they were getting screwed by AI driven changes 2. There are lots of independent sources of airline prices that savvy users could check to see in the airlines are trying to gouge them, along with multiple ways to bypass prices linked to their identity (check prices without signing in, call the 800 number, use VPNs, etc). Recent attempts (AA for example) to force corporate travelers to only use airline controlled channels have failed. 3. There are quite a few bloggers (like the one cited here) who would see this is a major story if evidence began mounting of big price differentials 4. Is Delta using AI to eliminate a few low-level pricing clerks or to rethink its overall pricing strategy? Many current LLM use cases focus on the ability to automate repetitive junior staff work. LLM seem to be able to quickly summarize past case precedents faster than a paralegal could. But they are not coming up with novel strategies for winning difficult, complex cases. Airlines had automated all the decision rules for matching/not matching competitor fare changes based on market conditions decades ago so the potential to cut pricing department jobs seems limited. 5. As for pricing strategy, the big airlines have been trying to figure out what pricing rules best maximize total revenue for decades and have gotten pretty good at it. What’s the chance that a recently assembled large-language model suddenly finds a breakthrough algorithm everyone had missed for decades? If the AI approach succeeds it would suggest that all the airline executives who were trying to “get inside the head of consumers” were all overpaid idiots and were easily outperformed by a black box that made changes that no one can explain. 6. Airlines have been trying to figure out “customer price elasticity” forever. Simple concept but virtually impossible to accurately measure across millions of customers booking very diverse types of flights. Airlines long ago figured out how to merge FF customer data with basic external data (credit card spending, did they own a home, etc). Not much evidence it drove big unit revenue gains and presumably reached the point of diminished returns. 7. “Algorithmic Pricing” is better termed “Surveillance Pricing”—as with the big tech monopolies to capture reams of private data about individuals I order to slightly better tailor online ads that pop up in real time. Airline tickets are not the kind of impulse purchase that the vast majority of Google/Facebook type ads are targeted at. What specific (historically) private data about their customers that Delta never had access to do they think would allow them to transform their pricing practices? 8. Has Delta explained what it is trying to maximize here? Traditional revenue management--short-term revenue given a fixed schedule? Medium term airline profits, including the staffing/capacity cuts after fares rise? Long-term revenue extraction from elasticity-defined categories of customers accounting for loyalty losses? Accurately measuring elasticity differences among millions of customers is difficult enough; integrating the operational/capacity cost components that drive profitability is much harder and seems even less appropriate for LLMs. 9. While this kind of pricing was a major FTC target under the previous administration, the current administration quickly killed those initiatives and apparently sees this type of consumer exploitation as an important driver of the economic growth they have promised. This facilitates the kind of “AI” driven abuses seen in heathcare where black boxes deny much higher rates of treatment/insurance claims without any possibility of appeal or accountability. But do you think frequent flyers will start to raise hell if they think the airlines are suddenly trying to raise prices well above what supply/demand conditions would allow, or would they say, yes this will cost me a lot of money but this is how American capitalism was always designed to work so I won’t complain? It is important to remember that the central economic/consumer welfare issue isn’t the specific “pricing algorithm” that might be used but the exercise of anti-competitive market power. The deliberate, systematic elimination of meaningful competition meant that airlines are free to raise prices with impunity and to pursue practices that their most important customers (would) totally hate. If real competition existed would any airline be paying “AI” consultants to figure out how a black box algorithm might be able to charge their best customers higher fares? I think the efforts to shift from traditional “price discrimination” to a much more radical system based on estimates of the price elasticity of individual customers is the industry’s biggest issue at this point. I don’t see much evidence that knowledgeable industry observers understand how big this potential change could be or that the financial analysts or the journalists following the industry are prepared to dig into these questions. Hope I’m wrong. While I am reluctant to reopen Uber issues with this group, they provide some useful context to the airline pricing questions. Uber lost $32 billion over 12 years, and those huge losses were not reversed after the change in senior management in 2018. The biggest single factor driving Uber’s profit recovery was abandoning the previous practice of offering the same fare to any customer and offering the same payment to any driver making comparable requests (e.g. pick up locations, distance, time of day) and replacing with algorithmic formulas that estimated the highest fare/lowest payments they would accept. [1] This illustrates why Delta management might think that telling investors that it was introducing radical price discrimination based on offering different prices based on the “price elasticity” of each individual customer might juice the stock price. But it ignores the fact that Delta has none of the structural advantages that allow Uber to maximize exploitive discrimination. Uber rides are last minute purchases and riders have no ability to compare prices. There are no independent Google/Kayak/Expedia sources of taxi pricing information. Delta frequent flyers understand airline pricing and would quickly figure out if changes were unfavorable. Uber users have no real idea how Uber pricing works, and Uber users have passively accepted huge post-pandemic price increases thanks to the billions Uber (and Lyft) used to drive independent competition out of the market. Taxis are not airlines but in both cases the real issue isn’t “innovative algorithms”, it is the ability to exercise anti-competitive market power without any fear of market discipline. It also illustrates a dilemma Delta might face if its introduction of radical price discrimination had any actual potential to bear fruit. It would want to publicize sudden unit revenue gains (and how it achieved them) in order to boost its stock price. But publiczing these gains would royally piss off the high-yield frequent flyers Delta has been working so hard to keep happy. Uber recognized that everything it was doing to increase profits was nakedly exploitive and would be attacked if widely understood. So it has adamantly refused to offer investors any meaningful explanation of what was one of the biggest profit turnarounds in history, and actually cut back on its already very limited SEC reporting. Companies can’t make major changes that would produce a furious backlash from customers, drivers, journalists and politicians unless market competition has been eliminated. [1] Here are a couple recent articles about external studies of Uber’s radical new algorithmic approaches: Simon Goodley, Rough ride: how Uber quietly took more of your fare with its algorithm change, The Guardian 19 June 2025; Simon Goodley, Second study finds Uber used opaque algorithm to dramatically boost profits, The Guardian 25 June 2025 The other very important driver of Uber’s ability to achieve profitability after years of multi-billion dollar losses was its ability to crush all efforts to provide any rudimentary labor law protections for its drivers, including efforts approved by California legislator and Supreme Court. For a fuller explanation of Uber’s financial turnaround see Hubert Horan: Can Uber Ever Deliver? Part Thirty-Five: What Drove Uber’s Recent $8 Billion P&L Improvement?, Naked Capitalism, 25 Feb 2025
-------------------------------------------------------------------------- Revised: 20250507 You are receiving The Mifnet because you requested to join this list. The Mifnet is largely a labor of love, however the infrastructure isn't exactly cost-free. If you'd care to make a small contribution to the effort, please know that it would be greatly appreciated: https://wardell.us/url/mifbit All posts sent to the list should abide by these policies: 1) List members acknowledge that participation in Mifnet is a privilege--not a right. 2) Posts are always off the record, absent specific permission from the author. 3) The tone of discussions is collegial. 4) Posts are expected to be in reasonably good taste. 5) We discuss ideas and not personalities, and we don't speak ill of other Mifnet members. * The Mifnet WEB SITE is: https://www.mifnet.com/ * To UNSUBSCRIBE from this list at any time please visit: https://lists.mifnet.com/ OR: SEND THIS MESSAGE via email: mifnet-requ...@lists.mifnet.com?subject=leave * Send Mifnet mailing list POSTS/SUBMISSIONS to: mifnet@lists.mifnet.com * You may reach the person managing The Mifnet at: mifnet-ow...@lists.mifnet.com * Please consider the DIGEST version of The Mifnet, which consolidates all list traffic into 1-3 messages daily. See instructions at: https://lists.mifnet.com/ * Manage your personal Mifnet SUBSCRIPTION at: https://lists.mifnet.com/ * For a list of all available Mifnet commands, SEND THIS MESSAGE via email: mifnet-requ...@lists.mifnet.com?subject=help * View The Mifnet LIST POLICIES and PRIVACY POLICY at: https://mifnet.com/index.php/policies * View instructions for Mifnet DELIVERY PROBLEMS at: https://mifnet.com/index.php/delivery-problems * View The Mifnet LIST ARCHIVE at: https://lists.mifnet.com/hyperkitty/list/mifnet@lists.mifnet.com/
