FTC asks nine companies how personal data shapes prices, job offers and worker pay

The Federal Trade Commission is demanding that nine major companies explain whether personal data influences what customers pay and what workers earn. The October 28 inquiry is broader than a question about an airline charging two passengers different fares: it also examines which gig worker receives a job and how the platform divides the customer’s payment.

Orders went to Delta, United, JetBlue, Lyft, Uber, DoorDash, Instacart, Target and Walmart. Those recipients span airlines, ride-hailing, delivery and retail. Being included does not establish that a company uses every practice under examination, let alone that it has broken the law.

The question behind the price headline

The FTC’s published special-report order separates three decisions: consumer pricing, assignment of jobs and worker compensation. It asks how personal information enters each process. Crucially, it also seeks the connection between the price a customer pays, the worker’s compensation and the revenue retained by the company.

That connection is more informative than looking at either side alone. A passenger sees a fare. A driver sees an offer. Neither necessarily sees the formula that joins them. Studying only the passenger’s receipt could miss a change in the worker’s pay; studying only the worker’s offer could miss a change in what the customer was charged.

A deliberately hypothetical example illustrates the distinction. If a customer pays $20 and a worker receives $14, the difference is $6. If the charge becomes $22 while compensation falls to $13, the difference becomes $9, a 50% increase. These are not observed prices or allegations about any recipient. The difference is also not automatically profit: other charges, costs and accounting treatment would still matter. The example shows why the two sides must be examined together.

Job allocation adds another layer. A worker could receive the same payment for an accepted job but fewer opportunities to accept one. A study that measured compensation only after completion could overlook that change. The order’s separate assignment questions make it possible to examine access to work as well as the quoted pay for work.

A study with compulsory process

The agency’s announcement identifies Section 6(b) of the FTC Act as its authority and says the vote was 2-0. That provision supports industry studies that need not have a specific law-enforcement purpose. The information demand is compulsory; a refusal can lead the FTC to seek enforcement in federal court.

Those features should be kept together. Calling the exercise merely a voluntary survey understates the companies’ obligations. Calling it a finding of unlawful discrimination or a completed enforcement case overstates what the commission has established. The immediate event is a demand for evidence about business practices.

The order seeks a sworn company report and information generally dating from January 1, 2023. Some questions about research and tests reach back to January 1, 2021. Its requests cover underlying tools, inputs and overrides, rather than asking only for a company’s preferred description of its pricing policy.

This design matters because a marketing label can be narrower than the commercial process beneath it. A company might describe a system as a discount engine, a loyalty feature or a matching tool. The useful question is which information it actually uses and which decision changes as a result. A technical description can be checked against records in a way that a broad assurance cannot.

A changing price does not prove personalization

The distinction is between a price varying because the market changed and a price varying because the seller used information about the individual. The FTC’s August 19 proposed policy statement recognizes ordinary variations tied to supply, demand and market conditions. It concentrates on circumstances in which consumers would not reasonably expect personal-data-driven differences.

Two screenshots showing different quotes are therefore a lead, not a complete demonstration. The comparison needs the same product, time, location and terms, with relevant discounts and availability accounted for. Otherwise a difference may have several explanations. Conversely, a company giving the same publicly visible base price does not by itself answer whether a personalized discount or another component changes the final transaction.

The proposed statement discusses disclosure where consumers reasonably expect nonpersonalized prices. It is a proposal, not a newly issued blanket ban. The document expressly says Congress has not authorized the commission to prohibit personalized pricing in every circumstance. Existing restrictions on unfair or deceptive conduct and a universal prohibition are different legal propositions.

The companies’ countercase belongs in the record

Reuters reported that spokespeople for Delta, JetBlue and Instacart denied engaging in personalized pricing. Those statements are important context, not conclusions the Chronicle has independently tested against internal records. Reuters said representatives of the other companies did not immediately respond to its requests; that is not evidence that they declined a request from this publication.

The strongest commercial counterargument is that data can support useful matching, loyalty discounts and more efficient use of capacity. The consumer-protection question is not answered by declaring every algorithm harmful. It is answered by identifying what the system does, whether people understand the terms and whether the claimed benefits coexist with unfair or deceptive outcomes.

For an independent driver or delivery worker, compensation and opportunity need to be assessed together. For a traveler or shopper, the relevant comparison is the full offer, not an isolated advertised number. For a smaller merchant relying on a delivery platform, a customer’s total bill may affect conversion even when the merchant has not changed its own menu prices. These are the practical reasons the study’s scope matters.

Two deadlines, neither a guessed calendar date

The published order sets a response deadline of 45 days after service and calls for contact with staff within 14 days about possession or control of the requested information. Those clocks run from service, not automatically from the news release. Without each company’s service date, assigning every recipient the same December deadline would create precision the record does not supply.

Nor should readers expect every company document to become public at the deadline. The order describes protections for confidential material. The next evidence of progress will be compliance, any contested demands and the commission’s eventual findings or subsequent legal action. Until then, the significant change is access: the regulator is asking for the machinery behind prices, work offers and pay, rather than trying to infer that machinery from the screens consumers and workers can see.