Ferguson Dissented From the First FTC Personalized Pricing Study. The New Order Wants the Inputs.

“There is no fare product Delta has ever used, is testing or plans to use that targets customers with individualized prices based on personal data.” Peter Carter, Delta’s chief external affairs officer, wrote that sentence to three United States senators on 31 July 2025, in a letter answering their questions about the airline’s artificial intelligence pricing pilot with Fetcherr. On 28 October 2026, the Federal Trade Commission issued the order that puts that sentence to a test, demanding the paperwork that would prove or break it, under oath, from Delta and eight other companies.

The order, issued under Section 6(b) of the FTC Act, went to Delta, United and JetBlue; to Lyft, Uber, DoorDash and Instacart; and to Target and Walmart. Each company has 45 days after service to file a special report, sworn by a company official and produced Bates-stamped, and each must meet FTC staff within 14 days to discuss the scope. A company that refuses can be compelled in federal court. The Commission voted 2-0. The order states the study “does not have a specific law enforcement purpose.” The specifications that follow run to 23 items.

Specification 7 asks for each pricing methodology that uses personal information: every tool and input named, whether the value of each input varies by individual consumer or by category of consumer, how missing values are handled, how overrides work, and what restrictions apply in each jurisdiction. Specification 9 asks for the count and revenue, product by product and year by year, and how much of it ran through the methodology. Carter’s letter said the pricing tool works on aggregated data and that Delta shares no personal information with Fetcherr. The order asks for the inputs themselves, input by input.

Specification 20 reaches further back, to 1 January 2021, and asks for every analysis, research project and A/B test the company or a third party ran on pricing or compensation, including projects paused, suspended or terminated early, and the reasons. Results must be given, including effects on sales, revenues and profits, with the managers responsible named. A pilot in test markets, which is how Carter described the Fetcherr work, is exactly the kind of project this specification collects, with its results attached.

The definitions decide what counts. “Personal Information” in the order runs from geolocation and browsing history to income and wealth, and ends with two entries the companies’ public statements never use: information about a person’s “willingness to pay” and, for workers, “willingness to accept a Job,” plus anything inferred or derived from the rest. Grouping consumers into categories that get different prices counts as a methodology. The FTC gave the systems in scope its own name, “Super Intelligence,” defined as a machine-based system that predicts, recommends or decides. Delta president Glen Hauenstein had told investors the technology could set fares on a prediction of “the amount people are willing to pay for the premium products related to the base fares,” and Fetcherr’s chief executive, Roy Cohen, had said the system was trained using “all the data we can get our hands on,” according to the senators’ letter.

In July 2024, Ferguson concurred when the Commission voted 5-0 to open the first study, which went to eight intermediaries, the consultancies and software firms that build pricing tools, rather than to the companies setting prices. When the outgoing Democratic majority issued initial findings in January 2025, Ferguson dissented with Commissioner Melissa Holyoak, accusing it of rushing out summaries “just before President Trump takes office” and writing that “slowly dripping out information” did not serve the public interest. Days after taking office as chairman, he withdrew the request for public comment on those findings. No final report followed. The October release says the 2024 study “examined only middlemen consultancies” and the agency “learned nothing about actual pricing decisions.” The new FTC personalized pricing order carries the same matter number, P246202, and is addressed to the companies that set prices.

The senators’ letter is the other document the order answers. Ruben Gallego, Richard Blumenthal and Mark Warner wrote to Delta chief executive Ed Bastian on 21 July 2025 with eight questions. They asked what data inputs the tool uses and what it might use next, where the data comes from, which categories Delta had decided not to use, how many passengers a day saw or bought AI-set fares, what the 20 percent figure counted, which routes and competing airlines were involved, whether prices varied by access point, and what Delta had told regulators and advocates. “Prices could be dictated not by supply and demand, but by individual need,” they wrote. Carter answered with the denial, with the claim that fares are filed publicly through ATPCO several times a day, that customers need not sign in to shop and compare prices, and that the tool aggregates purchasing data, forecasts demand and adapts to market conditions. The order now asks for the inputs, the categories and the tests under oath, which is the remainder of the senators’ list in compulsory form.

Walmart and Target spoke before the order arrived. Walmart president and chief executive John Furner wrote to customers on 25 September: “We price the product, not the person.” He added that “using someone’s income, shopping history or moment of need to charge them more would violate the EDLP promise our business model is built on. We won’t do it.” The letter held the company’s AI shopping assistant, Sparky, to the same standard, and defended the digital shelf labels already in more than 2,300 stores as a way to keep the shelf price matching the checkout price. Target published a fact sheet on 15 October: “Target does not use personal information to raise prices for guests based on who they are, where they live or when they shop.” Its loyalty program, Target Circle, “can use your shopping history to help surface deals and offers that are more relevant to you and that help you save — not to raise your price.” Both statements were written as the scrutiny grew. The order’s disclosure specification, number 22, asks for the exact wording, the precise web address and the dates of every disclosure each company made about personalized pricing.

Drivers and couriers get a second set of specifications. Specification 11 covers offer assignment: which workers are offered which jobs, in what order, and which jobs are hidden from which workers. Specification 14 covers compensation methodology, including the extent to which each pricing methodology affects what a worker is paid, and the share of the price the consumer pays that the company retains as revenue, in detail sufficient to determine the split. “Compensation” is defined to include surge, peak and boost multipliers, quests and streaks, guarantees, tips, deductions and fees. The release gives the worker example in the same terms as the shopper test: compensation set from a driver’s “phone usage or acceptance history.”

Oxford researchers gave that question numbers in June 2025, from 258 London Uber drivers’ own records covering more than 1.5 million trips between 2016 and 2024, obtained through data access requests. After dynamic pricing arrived in London in early 2023, inflation-adjusted hourly pay fell from £22.20 to £19.06, the median take rate rose from 25 percent to 29 percent, and some trips carried a cut above 50 percent. “The higher the value of the trip, the more of a cut Uber takes. So the more the customer pays, the less the driver actually earns per minute,” lead author Reuben Binns said. Uber said it did not recognize the figures. A rider posting on Instagram the day after the order, under the name sooziwho, described the same pattern from the other side of the app: “Uber does that also. Three of us in the same place pull up a ride. They’re all vastly different prices. We go with the cheapest one obviously.” Three phones, three prices, proves variation. It does not prove the mechanism. The distinction is the whole subject of the order. Consumer Reports drew it the same way in June 2026, when 174 volunteers in 18 states found a median 42.4 percent spread between the lowest and highest prices for the same rides ordered within minutes. The organization said it was not disputing the companies’ claim that base fares are not personalized, and asked whether market forces alone explained the gaps. Uber and Lyft have said they use personal data for discounts and promotions, not for base fares.

In December 2025, a study by the Groundwork Collaborative, Consumer Reports and More Perfect Union sent 437 shoppers to the same stores at the same times in four cities and found prices varied on about 75 percent of items, with up to five prices for a single item and an average gap of 13 percent between the lowest and highest. Instacart said the tests were randomised and not tied to browsing history, ended them on 22 December 2025, and said shoppers in the same store at the same time see the same prices. The FTC issued a civil investigative demand on Eversight, the pricing software firm behind the tests. The separate $60 million settlement Instacart reached with the FTC that month concerned deceptive marketing of free delivery and membership, not pricing, and the two should not be confused.

The people posting about the order are asking a narrower question than the coverage. Nathan Hull, posting on Instagram on 30 October, wrote: “Your browsing history is not a coupon. The FTC just ordered Walmart, Target, Uber and six more companies to hand over data on whether the price you see depends on who you are. Ever check a price on your phone and swear it changed?” On Threads, stormycloudinc put it as a question to readers: “If your browsing habits changed a price, would you want to be told?” That is the disclosure question, and it is the one the FTC’s draft enforcement policy statement, published for comment in August, was written around. Chairman Ferguson said then that a listed price is expected to be the same for everyone, and that a business which fails to tell consumers how personal data sets a price may violate the FTC Act. Law firms summarizing the draft listed its examples: a delivery service charging more when data suggest the customer is less able to leave home, a grocery chain charging more for milk when household data show children, a hotel raising rates for funeral travel, a rideshare charging more when data show no competing apps installed. The draft proposes no ban and creates no new penalties; it says undisclosed personalisation may be deceptive. The comment period, extended in September, closed on 25 September.

Four states have already written their own answers, and the order’s loyalty probe runs into them. New York’s disclosure law has been in effect since July 2025, and its broader One Fair Price Act passed both houses in June 2026 and awaits the governor. Maryland’s ban on predatory pricing by food retailers and delivery providers took effect on 1 October 2026. Connecticut’s restrictions take effect on 1 July 2027 and New Jersey’s on 1 August 2027, including a one-year pause on new electronic shelf labels. Several of the bills preserve loyalty and rewards programmes, which is the line the FTC is now probing: loyalty participation is named in the release as one of the factors that may set a price. The Transportation Department, meanwhile, closed its own review of the ten biggest airlines on 4 September 2026 with no violations found and no penalties, while adding that airlines should avoid setting prices on individual data. Fifty-four days later, three of those airlines were ordered to produce their pricing records under oath. JetBlue has said it does not use personal information or web-browsing history to set individual prices. United, Lyft and DoorDash had not, in the coverage surveyed, answered the orders on the record.

The reports must be sworn by a company official and produced Bates-stamped, with individual customer records and the contents of customer communications excluded. The order sets a meeting with FTC staff within 14 days of service and gives each company 45 days to file.