Algorithmic Pricing

How It Works

Algorithmic pricing systems process large volumes of data to optimize revenue in real time. They generally fall into two camps, and the distinction between them has become the central fault line of the regulatory debate.

• Dynamic pricing adjusts prices based on market-wide signals — supply levels, competitor pricing, demand curves, seasonality, and time of day. Think of an airline raising fares as seats fill, or Uber applying a surge multiplier during a thunderstorm. The algorithm does not know who you are; it knows how many people want a ride right now.
• Personalized (or ‘surveillance’) pricing sets prices for a specific consumer by analyzing that individual’s personal data — browsing history, purchase behavior, location patterns, device type, inferred demographics, and in some cases biometric signals. The algorithm does not just know demand is high; it estimates what you are willing to pay.

The line between them is not always clean. A grocery delivery app might adjust prices for a whole city based on demand (dynamic) while simultaneously running experiments that show different consumers different prices for the same basket based on their purchase patterns (personalized). Both are algorithmic pricing. Only the second is causing legislatures to act.

Where You See It Today

Algorithmic pricing is embedded across nearly every consumer-facing industry:

• Airlines (dynamic) were among the earliest adopters. Carriers like Delta continuously adjust fares based on seat inventory, route demand, weather, and competitor pricing. Delta publicly stated in August 2025 that its algorithms rely on market factors rather than individual consumer data. On any given flight, fares can swing by hundreds of dollars within hours — not because of who is buying, but because of how many are buying.
• Ride-sharing platforms (dynamic) like Uber and Lyft apply ‘surge’ multipliers — sometimes 1.5× to 5× or higher — during storms, concerts, and bar-closing hours to pull more drivers onto the road and ration demand. Critics have noted that algorithms may also weigh signals like phone battery level as a proxy for rider desperation, pushing the model closer to the personalized end of the spectrum.
• Grocery delivery (both) has become the most contested frontier. A 2025 Consumer Reports investigation found that Instacart’s pricing algorithms showed identical shopping baskets at up to five different price points, with basket-level differences as large as 8.4% ($9.59 on one test basket) and item-level differences reaching 23%. Safeway test baskets in the Seattle area ranged from $114.34 to $123.93 for the exact same products. Whether those differences reflect city-level demand variation or individual-level data targeting is precisely what regulators are trying to determine.
• Hotels and entertainment have used demand-based dynamic pricing for decades, but the integration of personal data into these models is a newer and more controversial step.

Why It Matters

Dynamic pricing based on supply and demand is economically rational and broadly accepted — it helps allocate scarce resources like seats on a flight or drivers during a storm. But personalized pricing raises a fundamentally different question: should two people standing in the same digital aisle pay different prices because an algorithm has estimated different willingness to pay based on their personal data?

The core concerns are three-fold. First, transparency: consumers generally have no idea that the price they are seeing may be individually tailored. Second, exploitation of vulnerability: algorithms that factor in health data, family circumstances, or a lack of alternatives can charge more to people who can least afford it. Third, price discrimination: personal data can serve as a proxy for socioeconomic status, race, or protected characteristics, raising questions about whether algorithmic pricing functions as a new form of digital redlining. These are not hypothetical risks. The FTC’s 2025 surveillance pricing study documented widespread use of consumer data — including data indicating vulnerability — to set individualized prices.

The Regulatory Landscape

Regulatory attention has intensified sharply, particularly in the United States. The focus is almost entirely on personalized pricing rather than dynamic pricing.

At the federal level, the U.S. Federal Trade Commission (FTC) released staff findings in January 2025 from its surveillance pricing study, documenting widespread use of personal data to set individualized prices. In August 2026, the FTC proposed an enforcement policy statement under Section 5 of the FTC Act, signaling aggressive enforcement against deceptive or unfair personalized pricing — particularly where disclosures fail to clearly flag that a price is personalized, or where data indicating consumer vulnerability is used. The FTC and Department of Justice have also argued that using pricing algorithms to coordinate benchmark prices among competitors may constitute unlawful concerted action under federal antitrust law.

State legislatures have moved aggressively. As of 2026, over 40 bills targeting personalized pricing have been introduced across more than 24 states. The most significant enacted laws include:

• New York — The Algorithmic Pricing Disclosure Act (effective November 2025) requires retailers to display a notice that ‘THIS PRICE WAS SET BY AN ALGORITHM USING YOUR PERSONAL DATA.’ The One Fair Price Act, which proposes the broadest definition of personal data yet seen, has passed both chambers of the legislature and awaits the governor’s decision. It would ban ‘surveillance pricing’ outright, with civil penalties up to $20,000 per subsequent violation.
• Maryland — The Protection from Predatory Pricing Act (signed April 28, 2026, effective October 1, 2026) is the first state law to prohibit — rather than merely disclose — personalized pricing, though its scope is limited to the grocery sector.
• New Jersey — The Fair Price Protection Act (signed July 23, 2026) bans personalized pricing for grocery retailers and third-party delivery platforms. Penalties reach $50,000 per violation, and the law creates a private right of action.
• Connecticut — HB8002 restricts the use of revenue management and pricing algorithms in rentals and surveillance-based pricing contexts, effective October 1, 2026.

A federal surveillance-data pricing bill died in the Senate in March 2026, leaving state law as the primary regulatory mechanism for now.

Connection to AI

Modern algorithmic pricing is powered by machine learning models capable of ingesting and weighing hundreds of variables in real time. The sophistication of these systems directly fuels the regulatory concern: the more data a pricing model can consume — browsing patterns, purchase history, device signals, time-of-day behavior — the closer it gets to estimating an individual’s maximum willingness to pay. Companies often describe their systems as ‘AI-powered’ to signal technological sophistication, but the real regulatory question is not whether the algorithm is smart; it is what data it is using and whether consumers know.

Related Terms

See also Artificial Intelligence (AI) and Machine Learning.