You search for a flight on your phone. The price looks high, so you leave the page. A friend searches for the same flight at the same time and sees a lower price. Curious, you refresh the page. The price shifts again. Experiences like this are becoming increasingly common online, and they are one reason policymakers are beginning to ask whether regulators should be able to investigate how pricing algorithms actually work.
Most people describe this as dynamic pricing, where prices are adjusted in response to real-time demand fluctuations, such as charging higher fares during peak hours. Sometimes that explanation is accurate. But another emerging practice deserves closer attention: algorithmic personalised pricing.
Unlike dynamic pricing, which typically adjusts the price in response to overall demand conditions and shows the same price to all customers at a given moment, personalised pricing works differently. The price shown to a particular customer can be shaped by analysis of personal data ranging from demographic indicators (such as age group or education level) to dynamic behavioural signals, such as browsing history, past online transactions, and online activity on social media. Most of this information is gathered directly from users or purchased from third-party data vendors.
Algorithms can construct detailed profiles of individual consumers in order to predict individualised willingness to pay. The objective is not simply to respond to market conditions, but to refine the price for the individual, based on how likely a consumer is to accept a higher price rather than search elsewhere. In economic terms, the ambition is straightforward: charge each buyer as close as possible to the maximum price they are prepared to accept.
In competitive markets, this strategy should be difficult to sustain. If one firm charges significantly more than its rivals, consumers can switch to a cheaper alternative. Competition disciplines pricing. But that discipline weakens when switching becomes difficult. Consider something as ordinary as buying hair colour. The market appears competitive, with many brands and retailers. Yet once a consumer finds a shade that works, switching becomes risky. Trying a different product involves trial and error, wasted time, and risk of a bad result. These small frictions make switching less attractive. Over time, many repeat buyers simply stop searching for alternatives. Not because they prefer the higher price, but because restarting the search process is costly.
Economists call these frictions switching costs. Even small obstacles, such as time constraints, uncertainty, or effort, can reduce consumers’ willingness to switch suppliers. When switching becomes difficult, demand becomes less sensitive to price increases. A similar logic appears in other settings. A traveller who has already booked a rental car may be offered the option to skip a long queue at the airport for an additional fee. The customer is already committed to the transaction. Time becomes a switching cost, and the “choice” is priced as an escape. These examples illustrate an important point: markets can look competitive while still leaving consumers effectively locked in.
Artificial intelligence is making this mechanism more powerful. Once a pricing system can reliably predict switching behaviour, the firm no longer needs to test the market price. It can test your price. AI pricing systems are shifting markets from discovering a market price to discovering your price.
For regulators, this raises a difficult question. UK consumer law and competition law address different types of problems. Consumer protection rules focus on issues such as misleading practices or hidden information. Competition law addresses a different concern: whether firms with significant market power can exploit consumers because competitive pressure is too weak to constrain pricing.
As pricing becomes more data-driven and individualised, the boundary between consumer protection and competition policy begins to blur. This raises a broader political question: if artificial intelligence allows companies to personalise prices at scale, should regulators treat algorithmic pricing as the next frontier of consumer protection in the digital economy?
This debate is no longer theoretical. The UK government has recently signalled that it is considering whether the Competition and Markets Authority (CMA) should receive stronger powers to investigate algorithms across both its competition and consumer protection functions. If pricing decisions are increasingly made by automated systems, regulators may need the ability to examine how those systems behave in practice. Traditional competition investigations rely on tools such as price comparisons, cost data and internal documents. But personalised pricing complicates these benchmarks. When each consumer may see a slightly different price, the overall price level can appear stable even while some individuals consistently pay more than others.
The challenge is therefore not simply legal theory. Competition law already prohibits dominant firms from imposing unfair prices. The real difficulty lies in evidence. To determine whether personalised pricing crosses the line into exploitation, regulators would need to understand how pricing systems respond to different consumers, for example whether prices systematically increase when a buyer is unlikely to switch.
In practice, that behaviour often sits inside complex algorithms that neither consumers nor regulators can easily observe. Granting the CMA targeted powers to test and investigate algorithmic systems would not expand the scope of competition law. It would simply allow existing rules to be applied in markets where pricing decisions are increasingly automated.
If algorithms are increasingly determining what each of us pays, regulators will need the tools to see how those algorithms work.
Otherwise, the discipline of competition risks quietly giving way to something else: prices tailored not to the market, but to each consumer’s vulnerability.
Many consumers have experienced price changes online for flights or hotels, but few understand why. AI increasingly allows firms to predict how likely an individual customer is to accept a higher price and adjust prices accordingly. In competitive markets, consumers can switch suppliers, but small frictions such as time constraints or simple convenience can weaken that discipline. As a result, markets that appear competitive may still leave some consumers effectively locked in. This raises an important question: if algorithms can identify consumers who are unlikely to switch, should regulators be able to investigate how these pricing systems operate?
