“Driving Up Revenue”

Delta’s new AI-powered pricing scheme is the latest development in a long line of novel pricing strategies pioneered by the airline industry, many of which eventually spread to other sectors.

Airlines operate on a complex “hub and spoke” network system, sharing infrastructure such as terminals and air traffic control, which requires some coordination and information exchange between airlines. Before the industry was deregulated in the 1970s, airfares were controlled by federal regulators, but in the half-century since, airlines have relied on private consultants and novel strategies to maximize profits amid tight margins.

Airline executives “talk about [their new pricing methods] on their earnings calls, Harvard Business School case studies are written about how successful they are in driving up revenue, and other companies follow suit,” said Owens.

According to airline industry expert Bill McGee, senior fellow at the American Economic Liberties Project, one example is “dynamic pricing, which devolved further into surge pricing and surveillance pricing.”

“It goes back a lot further than a lot of people realize,” he said.

Today dynamic pricing is typically associated with platforms like rideshare app Uber, which uses algorithms to jack up prices during peak “surge” hours for customers. But the pricing scheme was first pioneered by airlines when tickets for a flight would start filling up quickly. For the remaining seats, the airlines would raise the final price, meaning that passengers would pay dramatically different prices for the same flight, a practice sometimes known as “price discrimination.”

Airlines’ pricing regime has been shown to squeeze consumers. A study from Forbes found that just before a holiday weekend, dynamic pricing can increase ticket prices by five times as much as what those flights usually cost for popular routes.

That ubiquitous practice was an outgrowth of the airlines’ revenue-management business in the 1980s, in which RealPage’s principal scientist, Jeffrey Roper, got his start with Alaska Airlines. These specialized divisions were dedicated to devising new methods to “maximize revenue.” That mostly entailed studying and identifying the maximum price point that consumers would pay for different flights and would later lead to the introduction of junk fees on items like baggage and seating.

With the rise of the internet, data collection through online travel sites like Expedia became integral to airline price setting. According to a 2016 investigation from McGee at Consumer Reports, the resale travel sites use surveillance “cookies” to track online users’ activity across the web, gathering personal information such as location or even users’ browser history. Just as Big Tech platforms use consumers’ data to serve targeted advertisements, air travel sites use the information to serve targeted pricing.

Long before AI pricing, Owens explained, comparison shopping had already become much more difficult because of another airline-pioneered tactic: drip pricing. Airlines impose additional fees at the end of the checkout process that are hidden from the initial price, such as paying for seats or baggage — making it difficult to comparison shop for the best price.

The development of strategies like dynamic, personalized, and drip pricing has laid the groundwork for today’s surveillance pricing, which uses personal and behavioral data to wring as much money from each individual passenger as possible.

“It makes it harder to afford travel,” said Owens. “And it also makes budgeting really difficult for families” if they can’t roughly estimate what air travel will cost.

At a business conference last September, Lina Khan, former Democratic chair of the Federal Trade Commission, explained what AI-assisted pricing could look like for the average traveler.

She offered an example in which a traveler may be charged more by the airlines’ price-setting algorithm if the airline has information that the traveler has to be at a certain location by a set date, like a funeral or family emergency. Democratic senators echoed that example in a recent letter sent to Delta executives raising concerns about its AI pricing.

“Given just how much intimate and personal information that digital companies are collecting on us, there’s increasingly the possibility of each of us being charged a different price based on what firms know about us,” said Khan at the event.

Now, Khan’s forecast may be coming true with Delta’s partnership with Fetcherr, which has previously promoted its ability to enact “hyperpersonalized” pricing.

“Hardly Any Safeguards in Place”

Fetcherr, the company behind Delta’s AI business, is an Israeli startup founded in 2019 with close ties to the Israeli government; its current chief operating officer was a senior-level executive in the office of Israeli prime minister Benjamin Netanyahu until 2023.

The company’s three founders hail from the e-commerce industry, a sector where sophisticated pricing strategies have proliferated in recent years. The executives brought what they learned from the industry and applied it to airlines, comparatively a “legacy industry,” as CEO Roy Cohen explained in one 2023 YouTube video.

Fetcherr promises clients that its AI pricing system increases revenue by 4 to 6 percent, according to one 2023 slide deck. One airline pricing executive with Virgin Atlantic interviewed by the company on its podcast declined to give specific revenue increases from using Fetcherr, but told listeners, “We’re quite happy with the amount of benefit.”

In a blog post now stripped from the company’s website, as Thrifty Traveler reported, the company boasted that it could “find the optimal price for each customer,” an indication that it was using its technology to enable personalized pricing at a granular level.

In an interview on Fetcherr’s YouTube channel, the director of revenue management at Azul Airlines, one of Brazil’s largest international airlines, described such personalization in greater detail. “We are trying to make it as tailor-made as we can for each customer,” he said. “The more granular we can make this, the better offer we have.”

The executive and the Fetcherr representative framed the technology as a win for consumers, as is typical in the airline industry, which has often framed new pricing strategies as novel ways to offer consumers perks and discounts.

“In the end, all the customers are benefiting,” the Azul Airlines executive said.

But as McGee emphasized, this is a common talking point deployed to conceal the actual impacts of personalized pricing, which aim to ensure that each consumer pays as much as possible for a ticket, thereby improving airline revenues.

“The airlines talk about this as if you’re getting discounts,” McGee said. “You’re not getting discounts. . . . You can’t have a discount if you don’t have a baseline fare.”

In Delta’s recent letter to lawmakers, the company denied employing any personalized pricing, presumably including Fetcherr’s offerings, claiming it had “zero tolerance” for predatory or discriminatory pricing.

Instead, a Delta executive wrote, the company’s AI pricing tech was merely “a decision-support tool,” providing “informed insights for our analysts, who oversee and fine-tune the recommendations to ensure they are consistent with our business strategy.”

But experts say that it will be difficult for Delta to ensure there won’t be discriminatory pricing because the AI model used by Fetcherr isn’t transparent.

Testimony from Fetcherr’s other airline customers indicates that the company’s platform, which delivers price recommendations straight to consumers with little oversight, is something of a black box. The tech ingests data on competitor behavior, booking trends, web searches, current events, and constantly fluctuates airfare prices accordingly — doing so without human input.

“We have hardly any safeguards in place,” the Virgin Atlantic executive said in the YouTube interview, praising the technology’s efficacy. “There’s no human putting a step between the AI making a recommendation and it going into the market.”

And it’s not just Fetcherr. Other pricing consultants also advertise AI-powered airfare and airline pricing strategies, including companies like PROS Holdings, a pricing optimization company; Flyr, a start-up backed by venture capitalist Peter Thiel; and Sabre, a Texas-based travel company that claims to bring “modern retailing” to airlines.

In one 2022 blog post, Flyr advertised personalized pricing capabilities similar to those that Fetcherr has said it offers.

“Airlines need to sell the right product to the right customer at the right time, in the right channel, and at the right price,” one Flyr executive wrote, claiming that the company provided its clients the “optimal product bundle for each traveler, adjusting in real-time without the airline analyst having to wait for the next filing.”

One essential ingredient in surveillance pricing is market power. With the industry dominated by four major airlines, and even fewer airlines operating many routes, there isn’t a real competitive check holding companies back from raising prices.

Another potential risk of price-gouging is that if airlines raise prices too much, they might see demand plummet as consumers pull back on spending. But Fetcherr claims in company materials that its algorithm takes this risk into consideration.

In one case study shared by the company, its algorithm recommended a series of price increases for Virgin Atlantic, some as high as 17 percent. In one example, Fetcherr predicted that a 3 percent increase in airfare “is expected to result in a 0.6 [percent] decline in demand, ultimately resulting in a 2.3 [percent] boost in revenue.”

“It generates money every day,” Fetcherr’s CEO said in an interview posted on the company’s YouTube page. “That is the purpose of the system, to generate money for its customers.”

Collusion in the Sky

Algorithmic collusion is another potential threat. While Delta is currently Fetcherr’s first and only partnership with one of the “big four” airlines, policy experts warn that other major airlines could also sign up for the consultant’s services, creating the conditions for coordinated price hikes similar to what RealPage has been accused of doing in the housing market.

This wouldn’t be the first time that the airline industry has been embroiled in a price-fixing conspiracy.

In the 1990s, the most powerful airlines turned to a company called the Airline Tariff Publishing Company (ATPCO), which functioned as a quasi-independent central clearinghouse for airlines to share and publish information about airfares.

While this centralized hub served practical purposes for travel agents to scour for the best prices, it set off alarm bells for federal regulators as a would-be vehicle for collusion.

In 1992, the Department of Justice filed an antitrust lawsuit accusing ATPCO and its airline partners of facilitating a collusion scheme by sharing competitive, sensitive information and raising air travel tickets.

By getting instantaneous digital access to every price change from their competitors, airlines could adjust accordingly to find the maximum amount they could charge customers without losing their business.

The case, however, never went to trial as the Justice Department settled, instead imposing a modest consent decree prohibiting certain information-sharing practices. ATPCO still functions as the main clearinghouse for the industry, processing eighteen million ticket fare changes every day.

Fetcherr relies on the pricing information disseminated by ATPCO to train its generative AI and optimize pricing recommendations for clients.

“Working with partners like Fetcherr demonstrates our commitment to assisting the industry in driving new innovation, improved decision making, and optimizing commercial performance,” ATPCO’s sales head said in a 2022 statement announcing the partnership.