Online, the price you see may no longer be the same for everyone. Commerce is sliding from one price to a price per person: your data, location, device, history, is used to estimate the most you will agree to pay. Born forty years ago in airline ticket pricing, this made-to-measure algorithm raises two troubling questions: is it legal discrimination, and how far can it exploit a weakness?
1 The price just for you
It begins with an asymmetry.
The shift
From one price to a price per person
In January 2025, the US Federal Trade Commission revealed the scale of "surveillance pricing": companies use your data, precise location, browsing history, demographics, down to your mouse movements and the items left in a cart, to set an individual price. One telling example: a shopper profiled as a "new parent" is shown a pricier baby thermometer first. At least 250 retailers were using such strategies.
Retailers involved
≥ 250
using profile-based targeted prices (FTC, 2025).
Data used
mouse, cart
your movements and abandoned items can adjust your price (FTC).
The asymmetry
You cannot compare a price only you can see. That is the heart of it: the seller knows your profile, you do not know theirs. The market assumed a public price; it becomes a secret between the algorithm and you.
2 The first algorithm
To understand it, go back to airline tickets.
The ancestor
When the airline invented pricing by who you are
It all begins with the 1978 deregulation of US air travel, which freed fares. To survive, American Airlines invented "yield management" under Robert Crandall, who called it "the single most important technical development in transportation management since deregulation." The principle: sell a few seats cheap and keep the rest for travelers willing to pay more. Two neighbors on the same plane pay very different prices for the same seat, depending on when they bought, their flexibility, their profile.
Deregulation
1978
the Airline Deregulation Act frees US air fares.
System's gain
$1.4 bn
in extra revenue over three years for American Airlines (Edelman prize).
Mass discrimination
The airline was the first to industrialize price discrimination. And it pulled off a feat: getting passengers to accept that their neighbor paid half as much, without scandal. It is that norm that is now moving to your screen.
3 People Express
This algorithm was first a weapon.
The cautionary tale
The weapon that brought down a rival
In the early 1980s, the low-cost carrier People Express, founded by Donald Burr, grew on unbeatable fares. In January 1985, American struck back with its "Ultimate Super Saver" fares, so low they captured People Express's customers, while keeping the rest of the plane at full fare thanks to yield management. People Express, with no equivalent system, collapsed. As told by Robert Cross, Burr summed it up: "we were profitable until American came at us with the Ultimate Super Savers."
The fare counterstrike
Jan. 1985
American's "Ultimate Super Saver" fares target People Express.
DINAMO deployed
1988
American Airlines' automated pricing system.
The nuance
People Express did not formally go bankrupt: ruined by the price war, it was bought out and absorbed into Continental in 1987. The lesson endured: to control the individualized price is to control the market.
4 Pigou's three degrees
Economics has a name for this.
The theory
Arthur Pigou's three degrees
In 1920, the economist Arthur Pigou classified price discrimination into three degrees. The third, the most common and most accepted, sets a price by visible segment: student fare, senior fare, morning showing. The second plays on quantity or version: volume discounts, subscriptions. The first, long purely theoretical, is the most powerful: charging each person the exact maximum they are willing to give. That is precisely what personalized pricing aims at.
The three degrees
①
Third degree. By observable segment: student or senior fare. Accepted, because seen as a deserved discount.
②
Second degree. By quantity or version: volume rebates, subscription plans. You choose your own price.
③
First degree. The price at each person's maximum. Long impossible for lack of information; the digital world has just made it feasible.
From the student fare to the made-to-measure price
The difference is of degree, not of nature. The student fare does not shock, because it is transparent and collective. The first-degree price disturbs, because it is individual, opaque, and aimed not at a category but at you.
5 Capturing the surplus
The stake is measured in surplus.
The mechanism
Capturing all you were willing to pay
Between the price you would have accepted and the one you pay lies a gap: the "consumer surplus." It is your good deal. Perfect discrimination, the first degree, aims to capture it entirely, by setting your price at the highest you will tolerate. Long theoretical, it becomes possible because data finally meet its conditions: to measure your willingness to pay, and to prevent resale.
Profit from your data
+12%
of profit from web browsing, versus +0.8% from demographics (Shiller).
Your surplus
−8%
fall in consumer surplus under personalized pricing (Shiller).
The maximum, not the fair price
Economist Robert Shiller's study of Netflix puts a number on it: with your browsing data, some would pay nearly double what others pay for the same service. The price no longer reflects a cost, but your willingness to pay. You are no longer sold a product at its price; you are sold your own maximum.
6 From the plane to your screen
The loop closes today.
The present
Surveillance pricing, heir to yield management
The same logic is spreading through commerce, and even returning to airlines augmented by artificial intelligence. In 2025, Delta announced it would hand AI, via the firm Fetcherr, up to 20% of its domestic fares. Senators warned of a price pushed to each customer's "pain point"; Delta denied targeting individuals by their personal data. Amazon had tested variable prices as early as 2000; a 2025 investigation estimated Instacart could cost some shoppers up to $1,200 more a year, before it backed off.
Delta and AI
→ 20%
of domestic flights priced by AI targeted for end-2025 (Fetcherr).
Measured surcharge
$1,200/yr
estimated for some Instacart customers (Consumer Reports, 2025).
Dynamic is not personalized
One distinction is essential. A dynamic price varies with supply and demand, but stays the same for everyone at a given moment, like a peak-hour surge. A personalized price differs by who you are, at the same moment. The first is debatable; the second is worrying.
7 Is it legal?
Then comes the first question.
The law
Legal, under conditions, and in a grey zone
As it stands, personalized pricing is not banned, but framed. In Europe, the GDPR governs the use of your data, and the "Omnibus" directive has required, since May 2022, that you be informed when a price has been personalized by an algorithm. In the United States, the old Robinson-Patman Act protects only businesses from one another, not the consumer; but New York has required, since November 2025, the notice "this price was set by an algorithm using your personal data," and California has been investigating since January 2026.
Mandatory disclosure (EU)
May 2022
the Omnibus directive requires flagging a personalized price.
Required notice (New York)
Nov. 2025
"price set by an algorithm using your personal data."
The line not to cross
One limit stays firm on both sides of the Atlantic: discriminating on a protected characteristic, origin, sex, religion, is illegal, even when a "neutral" algorithm reaches it indirectly through proxies. For the rest, the law mainly demands transparency, and runs behind the technology.
8 Abuse of weakness?
There remains the moral question, the heaviest.
The exploitation
When the price targets your distress
Charging more to those with the least choice has an old name. It is the "pain point" senators feared: charging more to the traveler forced by a bereavement. It is, more everyday, the urgency signal: as early as 2016, an Uber executive admitted that one of the best signs a customer would accept a surcharge was the low battery on their phone, proof that they are in a hurry and have no way out; the company has since denied using it. A purchase late at night, or at the very last minute, betrays the same thing: the less choice you have, the more you pay. It is also the "poverty premium," the surcharge the poor pay, nearly £490 a year in the United Kingdom, on energy, credit, insurance. It is the loyalty penalty, four billion pounds a year across the Channel. In each case, the price no longer follows a cost: it hugs a vulnerability.
The "poverty premium"
≈ £490/yr
surcharge paid by a poor household in the UK (University of Bristol).
Loyalty penalty
£4 bn/yr
lost by loyal customers across five markets (CMA, 2018).
Abuse of weakness, really?
In French law, abuse of weakness (article 223-15-2 of the Criminal Code) punishes the exploitation of a known vulnerability. But let us be rigorous: it has never been applied to an algorithmic price, and tying it there remains an open, unsettled question. The parallel is moral before it is legal. As economist Daniel Kahneman noted, setting a price on a cost seems fair; setting it on the buyer's distress seems, by contrast, unfair.
9 The price you don't see
What remains is what to do with it.
The shadow side
Between targeted discount and exploitation
Not everything is dark. Personalization can also lower prices for the least well-off: by some studies, more than half of consumers would gain, through targeted discounts. The student fare is a discrimination, but a progressive one. And competition, where it exists, limits what a seller can extract. The real problem, researchers say, is not that prices vary, but that they vary in secret.
The compass
①
For the consumer. Demand a non-personalized base price and the right to compare; switch devices, clear your cookies, refuse tracking where you can.
②
For the citizen. The question is not "can the price vary," but "how far can a vulnerability become a price." Transparency is the first safeguard.
③
Clear-sightedness. Tell the discount that widens access from the extraction that targets your maximum. The first serves the market; the second turns it against you.
The price you don't see
For a century, the posted price protected the buyer: it was public, therefore comparable. The personalized price breaks that pact. The real question is not whether prices move, they always have, but whether your weakness, on a day of distress or ignorance, will become the price you pay.
Reference: abbreviations & acronyms used (FTC, GDPR, £bn, $).