Personalized pricing: definition, real cases and the disclosure rules

Personalized pricing sets each buyer's price from data about them. How the model picks a price, and the disclosure rules now in force in the EU and New York.

Definition

Personalized pricing sets a price for an individual buyer from data about them, such as precise location, device, browsing or purchase history. It is the practical version of charging each person their maximum. In the EU and in New York, sellers must tell buyers when an algorithm personalized their price this way.

  • The decision rule is expected revenue: the price times the predicted chance that this person buys at that price.
  • New York's required notice reads: THIS PRICE WAS SET BY AN ALGORITHM USING YOUR PERSONAL DATA.

Personalized pricing means the price depends on who the buyer is, as inferred from data about them. Tinder is a documented software case. According to Dutch regulator ACM, the app used automated personalization to offer some users 50% off their first paid month based on the interest they had shown in the paid version, and in March 2024 it committed to tell users clearly about these personalized discounts (ACM).

How it works

A personalized pricing system predicts how likely each person is to buy at each price, then shows the price with the highest expected revenue for that person.

Text
expected revenue = price × P(buys at that price | what we know about this person)

Example: a model scores two visitors to a $39 plan.

VisitorAt $29At $39At $49Shown
A: new device, arrived from a coupon site10% → $2.906% → $2.343% → $1.47$29
B: company email, read the docs twice9% → $2.618% → $3.127% → $3.43$49

The inputs are whatever the seller can observe. The FTC's preliminary findings on what it calls surveillance pricing, released January 17, 2025, mention a person's precise location and browser history, and even mouse movements and products left in an online cart (FTC).

Why it matters for pricing

Disclosure is now required in two large markets. Since November 10, 2025, New York requires the notice "THIS PRICE WAS SET BY AN ALGORITHM USING YOUR PERSONAL DATA" with such prices, and each violation can cost up to $1,000 (N.Y. Gen. Bus. Law § 349-a, New York Attorney General). In the EU, ACM sums up the rule this way: a price based on automated personalization must be disclosed to consumers in advance. This is not legal advice.

Experiments draw questions too. In January 2026, New York's Attorney General asked Instacart about its price experiments and its use of consumer data, after a Groundwork Collaborative and Consumer Reports study found shoppers paying up to 23% more for the same item at the same store at the same time (New York Attorney General). If you test prices, be ready to show exactly what decided each visitor's price. In a Leverprice test, the variant comes from the test's traffic weights and stays with the visitor; with regional prices turned on, the visitor's country also sets the amount, and nothing else about them does (price testing and the law).

Common mistakes

  • Assuming buyers won't compare. Two coworkers can open the same pricing page on different laptops.
  • Using sensitive traits. ACM reports that Tinder had already dropped age-based pricing after reports suggested age discrimination.
  • Skipping the disclosure where it is required, or placing it far from the price.
  • Calling a targeted price a test. If the price depends on device, browsing or purchase history, it is personalized, whatever the dashboard calls it.

Price discrimination is the broader idea, and personalized pricing is its first degree. Dynamic pricing varies prices by moment instead of by person, willingness to pay is what the model tries to predict, and sticky assignment is how a fair price test keeps one price per visitor.

Frequently asked questions

In many places it is legal with conditions. New York requires a set disclosure when personal data feeds an algorithmic price, the EU requires advance notice of automated personalization, and pricing by sensitive traits such as age draws regulators' attention.

Is an A/B price test personalized pricing?

Not when the price is assigned at random and kept stable for each visitor, because it does not depend on who the person is. Targeting test prices by device or history turns it into personalized pricing.

What data counts as personal data here?

Data that describes or tracks a person or their device, such as precise location, browsing and purchase history. New York's § 349-a defines it broadly as any data that identifies a specific consumer or device or could reasonably be linked to one, which can include a location read from an IP address.