Personalized pricing can benefit the consumers who need it most

On Aug. 4, the Senate Judiciary Subcommittee on Crime and Counterterrorism convened a hearing with a pointed title: “ Your Data, Their Profit: The Consumer Cost of AI Surveillance Pricing ”

The title of the hearing, in which I testified, captured a real concern. AI-enabled pricing can sound less like innovation than extraction: companies collect data, infer vulnerability, raise prices and call the result efficiency.

That fear deserves to be taken seriously. Personalized pricing has acquired an ominous new name: surveillance pricing. The phrase is powerful, but it can obscure the central paradox: Different prices can look unfair, but one price for everyone can be even less fair.

A uniform price feels neutral because everyone sees the same number. In practice, it can exclude people who would buy at a lower price while preserving access mainly for wealthier or less price-sensitive consumers. The visible unfairness of price variation must be weighed against the invisible unfairness of exclusion.

We have seen this problem before. In the early 1980s, airlines were developing what became known as yield management . Passengers on the same flight could pay different fares depending on timing, restrictions, flexibility and demand. Consumers disliked the complexity, and many still do. Two people sitting side by side may have paid dramatically different prices for the same trip.

Yet if regulators had banned yield management because it looked unfair, air travel might have remained a premium service only for the rich, the famous and the expense-account traveler. Instead, airlines used variable pricing to fill seats that would otherwise have gone empty, offer lower fares to more flexible travelers and preserve capacity for travelers who valued convenience more.

American Airlines, the innovator, later estimated major gains from its system. But as all airlines adopted yield management, they did not prosper collectively. According to Airlines for America, cumulative net profits for the U.S. airline industry from 1979 to 2024 totaled $28 billion over 45 years. After inflation, that figure is essentially zero or slightly negative. Consumers, however, gained broader access to air travel.

That history matters, because price variation often looks unfair before its full effects unfold and are understood. A firm that must charge one price faces a margin-volume trade-off. Set the price high, and price-sensitive consumers are priced out. Set it low, and the firm sacrifices revenue from consumers who would have paid more.

Personalized pricing can make mutually beneficial transactions possible. The consumer who would not buy at the uniform price gets a lower offer, the firm gains a sale and consumer access expands.

This does not mean every form of personalized pricing is defensible. Pricing built on deception, illegal discrimination, privacy abuse, exploitation of sensitive data or monopoly power deserves scrutiny and, where appropriate, prohibition.

AI can make these risks more serious. A company might infer that a consumer is desperate, confused, isolated, sick or unlikely to comparison shop, then raise the price in ways the consumer cannot see or challenge. It might use protected characteristics directly or through proxies. These practices are not harmless innovation.

The hard task is to distinguish those harms from access-expanding and efficiency-enhancing personalization. Student discounts, off-peak fares, targeted coupons for a price-sensitive shopper, lower subscription tiers and flexible travel fare all involve price variation and consumer data. Consumers may accept them because they preserve agency and help people qualify for a better price. A hidden surcharge imposed because an algorithm concludes that a loyal or vulnerable customer will not leave is different. The economics may be related, but the legitimacy is not.

The right question, then, is not whether AI-enabled price variation should be condemned as a category. It is when price variation deceives, discriminates, invades privacy, exploits market power or denies consumers a meaningful chance to understand and respond. Regulators should target those harms without blocking pricing innovations that could expand access and benefit consumers.

A practical framework would ask five questions.

Was the data collected legally, transparently and within reasonable consumer expectations? Are consumers being deceived about the price, the basis for the price or the availability of alternatives? Are protected classes being harmed directly or through proxies? Is the firm using market power to prevent consumers from switching or rivals from competing? Does the pricing practice expand access, improve matching or intensify competition, or does it merely extract more from consumers who cannot protect themselves?

If the answers point to deception, discrimination, privacy abuse or coercive market power, intervention is warranted. If not, policymakers should be careful not to outlaw price variation merely because it is unfamiliar or uncomfortable. Personalized pricing needs rules, transparency, restraint and accountability. But it also needs room to evolve.

The airline analogy is useful because it shows what can be lost when visible unfairness dominates the debate. In trying to protect consumers from abusive pricing practices, Congress should not overlook the invisible unfairness of uniform pricing: products priced too high for marginal consumers, innovation chilled by regulatory fear and markets less able to serve heterogeneous demand.

If we had banned yield management in the early 1980s, we might still think of flying as a luxury. We should not make the same mistake with AI-enabled personalized pricing.

Z. John Zhang is a professor of Marketing at the University of Pennsylvania’s Wharton School.