Personalised pricing can make the figure on a shopping page depend partly on information about the person looking at it, rather than only on the product and the shop’s ordinary price list. US consumer regulators have examined technology that uses location, browsing habits and shopping behaviour to vary offers or display prices. That is a reason to ask better questions when comparing prices online, not proof that every retailer is secretly charging every customer differently.
Most shoppers already understand changing prices. Airline tickets, accommodation and seasonal offers can rise or fall because of availability and demand. The more unfamiliar possibility is that the same seller may know something about you and use that information to decide which offer you see. A lower discount for one person can matter just as much as a higher sticker price for another.
Personalised pricing is not just a change in the sale price
A Federal Trade Commission study published in January 2025 described how pricing intermediaries can use data such as precise location, browsing behaviour, items left in an online basket and shopping history. The regulator called attention to systems that help businesses tailor prices, promotions or product presentation to the individual. The findings concerned practices and capabilities observed through its investigation, not a measurement of how often every named retailer charged two particular people differently.
There are several distinct ways this can work. A site could change the price of an identical item, offer a different discount, place more expensive products near the top or decide which promotion a shopper qualifies for. These outcomes are not interchangeable. A regulator investigating tailored offers does not automatically establish that a specific consumer paid an inflated amount for an unchanged product.
The underlying data may be held by the retailer or by a service provider working with it. A company might compare activity during a browsing session with past transactions, broad location or other permitted signals. The analytical methods can include algorithms that estimate how likely someone is to buy. Not every rule-based pricing system is modern generative AI, although machine learning can be used in the same commercial area.
Why your abandoned basket could matter
Imagine placing an expensive appliance in a basket, leaving the website and returning later. A straightforward explanation for a changed price could be that a sale has ended or the seller updated its standard prices. A more personalised explanation could be that the system has learned something about your interest. Without access to the seller’s decision process, a shopper usually cannot tell which explanation is right from the displayed price alone.
That uncertainty makes the subject interesting. We tend to read a price as a public fact about an item. But an offer can also be a message aimed at a specific buyer. When the same retailer operates multiple websites, apps and loyalty systems, the ability to distinguish a general promotion from individual targeting becomes more important.
The FTC’s staff used hypothetical examples to illustrate the practices its study uncovered, and it did not publish each participating firm’s commercially sensitive records. That limitation matters. A vivid illustration, such as a person being shown a more expensive option after browsing for baby products, should not be reported as a proven transaction at a named shop unless there is separate evidence.
Regulators are asking how much disclosure shoppers deserve
In August 2026 the FTC sought public comment on a draft enforcement policy concerning personalised pricing. It described personal-data-driven prices based on what a company believes someone is willing to spend. The agency warned that misleading people about such practices may fall within laws it already enforces. A proposal for an enforcement statement is not the same as a universal ban or a final court ruling.
The regulator acknowledged limits to its legal authority. That is a useful corrective to simplistic headlines that announce personalised prices are illegal everywhere. Whether a practice breaches a particular rule depends on the jurisdiction, the data used, what consumers were told and other facts. This article describes US regulatory evidence, not a determination of the legal position in every European country.
A consumer should also recognise that transparency is a separate issue from the commercial fairness of the price. A disclosed personalised discount may still be confusing or unwelcome. Conversely, changing a public price because stock is low may be perfectly ordinary and not personalisation at all. The dispute centres on what data changes the offer and whether people understand the process.
What can a shopper actually compare?
The most useful first check is the total payable amount, including delivery charges, optional extras and conditions. A very low headline price can be less attractive once fees appear. If an item is sold elsewhere, compare like-for-like models, sizes, guarantees and sellers rather than trusting a personalised recommendation to have found the best value.
Prices may also differ across legitimate channels without proving unfair targeting. A retailer’s app might have a members-only offer. A physical branch may run a different promotion. A price-comparison service can lag behind a seller’s live checkout. Those explanations should be eliminated before claiming a particular algorithm singled someone out.
One practical experiment is to compare an item while signed in and signed out, or to look from a different browser. A difference is a clue to investigate, not proof of discrimination. Location, inventory, shipping destination, taxes, eligibility and timing can also explain a variation. Screenshotting the displayed offer and noting its terms helps if a charge later seems inconsistent with what was advertised.
LiveAIWire has explored the way automated pricing can affect Amazon sellers. That is a related but different mechanism: a seller’s algorithm can react to competitors or market conditions without using a particular buyer’s personal profile. Both systems complicate the old assumption that prices are set once, manually and for everyone.
Where AI shopping assistants enter the picture
People increasingly ask an assistant to narrow a catalogue or propose a suitable product. That can save time, but it does not settle whether the underlying offers are tailored, sponsored or genuinely best for the buyer. LiveAIWire has examined how AI-generated product recommendations can steer a shopping decision. The sensible response is to check the underlying retailer and transaction terms rather than treating a conversational summary as a price guarantee.
An assistant might also act on behalf of the shopper, moving from recommendation to basket preparation or purchase. At that stage, both permission to spend and the source of the price deserve attention. The earlier LiveAIWire discussion of trust in AI shopping agents is relevant because the person remains responsible for deciding whether an offer is acceptable.
There is a further reason not to make comparisons solely through an automated assistant. The assistant may be excellent at describing products while lacking complete knowledge of today’s prices, delivery terms or a loyalty discount that appears only during checkout. Readers can use it to prepare questions and identify alternatives, then verify final quotes with sellers themselves. An independent search is especially worthwhile when an offer seems inexplicably different from a price a friend saw.
The effect on household budgets may appear small on a single purchase but can be hard to notice over time. That observation is a reason to retain receipts and examine recurring charges, not a claim about how much any individual household is being overcharged. If a personalised offer is challenged, the useful evidence is the actual displayed price, the purchase conditions and the information the seller provided about tailoring. The central question is disclosure and fairness, not whether software was involved.
The everyday significance of a data-shaped price
The distinctive issue is not whether computers can change numbers on a screen. They have long done that. It is whether a seller can use observations about one individual to decide what that person is likely to tolerate or buy, while leaving the mechanism difficult to inspect. For shoppers, understanding the difference makes comparisons more deliberate. For retailers, it puts a premium on clear explanations and truthful representations.
Even a sceptical shopper should distinguish an estimated willingness to pay from a confirmed individual surcharge. The evidence needed for the second is much more specific. The FTC’s evidence establishes that these capabilities and practices deserve scrutiny. It does not establish a secret surcharge on every online purchase. A useful shopping habit is therefore neither blind trust nor automatic suspicion. Treat the first personalised offer as a starting point, compare independent alternatives and make the purchase decision on the real terms that you can verify.
About the Author
Stuart Kerr is Technology Correspondent at LiveAIWire, covering artificial intelligence, cybersecurity and the social impact of emerging technology. LiveAIWire is an independent, human-led technology publication using AI-assisted research, editorial production and original AI-assisted editorial illustrations under his direction.