The US Federal Trade Commission (FTC) is facing a growing question over the use of consumer data in retail pricing: could supermarkets and their technology providers eventually determine how much an individual shopper is able or willing to spend, and then use that information to influence the price offered to them?
The concern is not simply about prices changing. Supermarkets have long adjusted prices according to promotions, competition, supply, demand and stock levels. The more controversial development is the ability of sophisticated pricing systems to analyse information about an individual customer and build a picture of their purchasing behaviour, including what they buy, how often they shop, how much they normally spend and how responsive they are to discounts.
That creates a very different pricing proposition. If a supermarket’s system knows that one shopper normally spends $200 on a weekly shop while another spends $80, could it calculate that the first customer is more likely to accept a higher price? Could the system then adjust an offer, discount or product price according to that customer’s estimated willingness to pay?
The FTC has been examining what it calls surveillance pricing, including the use of detailed consumer information and technology to determine or influence prices and promotions. Its work has highlighted technology capable of using personal and behavioural information to estimate consumers’ price sensitivity and potentially tailor prices, discounts or offers accordingly.
For grocery retail, the issue is particularly sensitive because supermarkets already possess enormous amounts of purchasing information through loyalty schemes and digital shopping platforms. A customer’s basket can reveal far more than the products they want. It can show shopping frequency, preferred brands, response to promotions and approximate spending patterns. Combined with other data, that information can potentially create a remarkably detailed picture of a shopper’s commercial behaviour.
There is an important difference between giving a customer a personalised discount and charging that customer more because the retailer believes they can afford it. Consumers generally welcome a targeted offer that reduces the cost of their weekly shop. The reaction could be very different if the same technology were used to identify customers who are less price-sensitive and therefore considered suitable for a higher price.
This raises a fundamental question for regulators and retailers: should two shoppers standing in the same supermarket, buying exactly the same product at the same time, be offered different prices because an algorithm believes one of them can afford to pay more?
The technology could become even more powerful as supermarkets expand the use of artificial intelligence, electronic shelf labels, personalised apps and digital loyalty programmes. These systems can make pricing considerably more flexible, allowing retailers to change offers quickly and target promotions with increasing precision.
The challenge for the FTC is therefore likely to be where legitimate personalised retail ends and unfair personalised pricing begins. Dynamic pricing itself is not necessarily unlawful, and retailers can have legitimate reasons for changing prices. The greater concern arises when consumers are unaware that personal information is being used to determine the commercial terms presented to them.
For consumers, transparency may become the most important issue. A shopper should arguably know whether the price displayed on a screen or in an app is the same price being offered to everyone, or whether it has been influenced by information the retailer has collected about that particular individual.
As supermarkets become increasingly data-driven, the traditional question of “What does this product cost?” could eventually become “What does this product cost for me?” That is a much bigger question for the FTC, retailers and consumers alike, because once technology can estimate what a shopper is prepared to pay, the boundary between personalised service and personalised exploitation becomes considerably harder to define.