A Personalized Price Now Needs an Explanation

Key takeaway

The Federal Trade Commission's proposed enforcement policy does not ban personalized pricing. It changes the operating burden. A seller that uses personal data to decide what one consumer should pay may need to disclose that the price is personalized, explain the basis for the decision and identify the types of data used.

That turns a pricing experiment into a customer-facing data claim. Retailers will need more than a model and a conversion lift. They will need to know which inputs shaped the offer, what the interface told the customer and whether the explanation remains accurate as vendors, features and targeting rules change.

What’s changing

On August 19, the FTC voted 2-0 to seek public comment on a draft enforcement policy statement addressing personalized pricing: the use of personal data to set a price according to what a business thinks an individual is willing to pay.

The proposal draws an important boundary. The FTC says Congress has not authorized it to prohibit personalized pricing in every circumstance. Instead, the agency argues that undisclosed personalization may be unfair or deceptive under Section 5 of the FTC Act when consumers would reasonably expect a common price.

In that setting, the draft says a business should clearly and conspicuously disclose three things: that the price is personalized, the basis for the personalization and the types of data used. The FTC announcement says comments will be accepted for 30 days after Federal Register publication.

This is not a final rule, and the legal theory may change after comment or face challenges in enforcement. But it is specific enough to create an immediate governance question for retailers, marketplaces, travel platforms and subscription businesses: can the company reconstruct why this customer saw this price?

The proposal distinguishes individual personalization from familiar market-wide variation. A ride may cost more in a busy neighborhood; taxes and market conditions may differ by region; credit and insurance prices necessarily reflect individual risk. The sharper concern is a product that customers ordinarily expect to have the same price at the same place and time, but whose offer changes because of browsing behavior, purchase history, inferred income or data acquired from another firm.

Why it matters

Personalized pricing is usually managed as a revenue optimization problem. Data teams estimate willingness to pay. Merchandising sets constraints. A vendor may supply a score or audience. The commerce system chooses an offer. Success is measured through conversion, revenue per visitor and margin.

The FTC's approach adds a second system: an explanation trail. A disclosure that says only “prices may vary” may not satisfy a policy that asks for the basis and data types. Yet a specific disclosure cannot be written reliably if the pricing team does not know what a third-party model used or if the feature set changes without the interface changing with it.

That makes vendor opacity an operating risk. A retailer may receive a recommended price without receiving a usable account of whether the recommendation relied on the customer's location, device, purchase history, loyalty status, inferred urgency or behavior across other sites. The retailer still owns the price shown to the customer.

It also complicates the word “discount.” The FTC gives the example of a customer who thinks a special price reflects loyalty, when the offer actually reflects an estimate based on disposable income or shopping behavior elsewhere. A higher personalized price dressed as a benefit is more dangerous than an openly differentiated offer because it changes the customer's comparison behavior.

The practical dividing line is not simply dynamic versus static. It is whether a price changes because the market changed, an eligible benefit applied, or the business inferred something about this particular person's willingness to pay. Those mechanisms can coexist in the same pricing stack, so the explanation must identify which one produced the final offer.

What operators should do

Start with an inventory of every system that can alter the price before checkout: promotion engines, loyalty rules, logged-in offers, coupon services, demand models, experimentation platforms, travel repricing, sales-assist tools and vendor-supplied optimization. Record whether each decision operates by market, segment or individual.

For individual decisions, create a durable event record linking the displayed price to the model version, relevant input categories, eligibility rules, source system, disclosure version and time. Do not retain unnecessary sensitive data merely to build the log; record enough provenance to explain and audit the decision.

Require vendors to describe the input categories they use and notify the retailer when those categories change. A contractual promise that a tool is “AI-powered” or “privacy compliant” does not answer whether a displayed explanation is true.

Separate benefit eligibility from willingness-to-pay estimation. A loyalty price based on a published membership rule is easier to explain than a price chosen because a model predicts the member will not comparison shop. Both may be individualized, but they do different economic and legal work.

Finally, test the disclosure in the buying journey. It should appear where the price can influence the decision, remain readable on small screens and give the customer enough information to understand the nature of the personalization. Legal text buried in a privacy policy cannot repair a misleading offer at the shelf.

Bottom line

The FTC is not proposing to freeze prices. It is proposing that individual price differences carry an explanation when consumers would otherwise expect a common offer.

For operators, that makes price provenance part of the product. A personalized price that cannot be reconstructed, explained and matched to the disclosure is no longer just a sophisticated experiment. It is an unsupported claim made at the moment of sale.