Uber's Algorithmic Pricing: Who Pays and Who

Eleven Business Insider employees open the Uber app at the same time, in the same room, requesting the same UberX ride from lower Manhattan to the Plaza Hotel. The highest quote comes back nearly 21% more expensive than the lowest. Nobody moved. Nobody switched neighborhoods. The algorithm just — decided.

Uber's explanation, offered via email, is that even minor GPS imprecision between nearby addresses can alter route estimates and therefore price. That's technically defensible. It does not explain a 21% spread in a controlled test. And it certainly doesn't explain what Consumer Reports found when it ran a far larger version of the same experiment: over 170 volunteers checking identical routes at nearly identical times, with some routes showing a median price gap exceeding 50% between the highest and lowest quotes.

This is the terrain Business Insider's investigation maps — and it's worth understanding carefully, because the questions it raises don't have clean answers.

From Rate Card to Black Box

Uber's original pricing was, by today's standards, almost quaint: a base charge, a fixed rate per minute, a fixed rate per mile. A digital taxi meter, essentially. Surge pricing complicated that picture by multiplying fares when demand outpaced drivers, but the underlying logic remained legible. Drivers hated the uncertainty of surges for riders but, as one driver told Business Insider plainly: "If there were no surges, I wouldn't even drive. I would quit driving Uber. The surges is when I make a lot of money."

That model — transparent formula, occasional surge, drivers keeping roughly 75-80% of the fare — carried Uber through explosive growth. It did not, however, carry Uber to profitability. By 2017, the company posted a net loss of $4.5 billion. Going public in 2019 with a nearly $9 billion accumulated deficit put the financials under a new kind of scrutiny. Under CEO Dara Khosrowshahi, the pressure to demonstrate a path to profit became existential.

The mechanism that changed the equation was upfront pricing — quietly tested since 2014, then aggressively expanded from 2022 onward. Under upfront pricing, both what a rider pays and what a driver earns are determined before the trip begins, by an algorithm, with no published formula. Uber describes the inputs as estimated trip time, distance, time of day, route, demand patterns, tolls, taxes, and fees. What the algorithm does with those inputs — how it weighs them, how it combines them, and whether it incorporates anything else — is not disclosed.

In 2023, Uber reported its first-ever annual profit: $1.1 billion.

"Algorithmic Price Discrimination"

Len Sherman, a faculty member at Columbia Business School, has a name for what he argues Uber is doing: algorithmic price discrimination. Writing in Forbes in 2023, Sherman laid out the thesis — that upfront pricing allows Uber to separately estimate the maximum any given rider might be willing to pay and the minimum any nearby driver might be willing to accept, then optimize for both simultaneously. The goal, in his framing, isn't to price the ride; it's to price you.

Uber flatly denies this. The company says it does not use personal data to personalize fares — a denial Consumer Reports, notably, does not dispute. What Consumer Reports does document, alongside Business Insider, is that prices for identical trips vary substantially across users, and neither company nor researcher can fully account for why.

Sherman's framing is worth taking seriously without overstating it. As one analyst put it in Business Insider's investigation: "It's not personal surveillance maybe so much as it is they have all this data on a bunch of people like you — people who have certain similar habits. So if we send that group of people a specific price, you're likely to respond to that in a certain way." Group-level behavioral targeting and individual surveillance are legally and practically distinct. The practical effect on your wallet may be similar.

There's also the matter of Uber's own documentation. In New York, state law requires companies to disclose when personal data is used in pricing. Scroll to the bottom of your Uber fare quote in New York, and you may find this sentence: This price was set by an algorithm using your personal data. Uber says "personal data" here refers only to location — which qualifies under New York's disclosure law. Separately, Uber patents describe machine learning systems capable of tracking phone tilt angle, typing speed, tapping accuracy, and walking speed, and of inferring from ride history whether a rider is a single working parent, along with estimated age and gender. Uber told Business Insider those patents don't establish present or past use and don't relate to pricing. Patents describe capability, not deployment. Both statements can be true.

What Drivers Are Actually Earning

The rider side of the equation gets most of the attention. The driver side is arguably more structurally significant.

According to Sherman's analysis, as documented in the Business Insider investigation, upfront pricing allows Uber to find the lowest payout a nearby driver will accept — a floor, algorithmically determined, that can shift trip by trip. A 2025 study published in the ACM Digital Library under the title Not Even Nice Work If You Can Get It: A Longitudinal Study of Uber's Algorithmic Pay and Pricing, examining 258 drivers, found that upfront pricing had pushed Uber's median take from 25% to 29%, with instances exceeding 50% in the UK.

Sherman's own analysis of roughly 50,000 trips by three veteran drivers found his measure of Uber's average take rate exceeded 50% for each driver. Uber disputes that its take rates have increased, attributing the gap between rider fares and driver pay primarily to rising commercial insurance costs.

That explanation gets complicated in practice. Levi, an Uber driver and former accountant in Syracuse, New York, tracked his numbers over two years and found that Uber's listed insurance charge for the identical Ithaca-to-Syracuse airport run fluctuated between $15 and $50 with no apparent logic he could identify. Sherman analyzed Levi's data and found the insurance and operational fee fluctuations were statistically unrelated to trip time or distance. "I equate it to going to a casino," Levi told Business Insider. "There are times I'm winning big and then there are times you go into the casino and you can't win."

On a single ride Business Insider documented — $70.52 paid by the reporter, $27.31 earned by Levi before tip — Uber's service fee came to $21.29, or about 30% of the fare. Levi's share was 39%. The commercial insurance line item was $15.78. Uber also collected a cut of the wait time fee; Levi received 37 cents of the 56-cent charge. He was the one waiting.

Bill Lewis, who has driven for Uber since 2017 and accumulated nearly 40,000 rides, described the strategic reality of working under upfront pricing with characteristic precision: "You'd think as a driver, the way to make the most amount of money is by being a good driver, by being safe, reliable, having a clean car. No. The way you make money is by playing the game."

The Regulatory Gap Nobody Is Filling

This is where Business Insider's investigation, solid as it is, reaches the edge of its jurisdiction — and where the policy questions become most pointed.

The FTC issued a report in late 2024 examining surveillance pricing practices across eight major firms, warning that the use of individualized consumer data to set prices "may harm competition and consumers." The report named Mastercard, Revionics, Bloomreach, JPMorgan Chase, Accenture, McKinsey, and others operating pricing intelligence services — not Uber specifically, but the underlying behavior the agency described maps directly onto what Sherman calls algorithmic price discrimination in ride-share.

What the FTC flagged, and what makes ride-share a particularly acute version of the problem, is the combination of data depth and market dependency. The agency noted that when consumers can't easily verify how prices are set and can't readily switch providers, the conditions for extractive pricing are essentially structural. Uber and Lyft, in most U.S. markets, are the only viable options at scale. The switching cost isn't just inconvenience — it's the loss of ride history, ratings, and any accumulated loyalty pricing. That lock-in is precisely the dynamic the FTC identified as enabling surveillance pricing to function.

Yet ride-share has not, to date, triggered enforcement action on pricing opacity specifically. Part of this is jurisdictional: the FTC's 2024 report was a study, not a rule. Part of it is definitional: proving that price variation constitutes illegal discrimination requires establishing what the lawful baseline price should have been — a near-impossible standard when the company's position is that every price is the legitimate output of a complex, proprietary algorithm. You cannot litigate a black box without opening it. And no regulator has compelled Uber to open it.

New York's disclosure law is the closest thing to a crack in that opacity — a state-level mandate that forced a single sentence into the app. It is not nothing. It is also not close to sufficient.

Uber has pioneered a model that the broader gig economy has since adopted. When the mechanism by which you're priced is invisible, when drivers can't determine whether a given trip is worth accepting in the five seconds they're given to decide, and when the regulatory apparatus is still asking study questions rather than enforcement ones, the asymmetry that one Uber driver described — "making money off asymmetric information" — doesn't stay in ride-share. It scales.

Samira Barnes covers technology policy and regulation for Buzzrag.