Our price or your price?

The implications of a world where prices are personalised

No queues, no store staff - the internet is an impersonal shopping experience, but this is deceptive. Online retailers can know a lot about us, and this information can be used to affect the prices we pay, in ways that may not be the same for everyone. This is not necessarily a bad thing, but with e-commerce expanding the possibilities for this are increasing and the implications are not yet clear.

Separating

One way companies may be able to increase their profits is by charging us different prices according to our willingness to pay, so-called ‘price discrimination’.[1][2] This can benefit some consumers. People who are more price sensitive could pay a lower price and buy more of a product than if everyone is charged the same price (although those who are less price sensitive may pay a higher price).[3] It may also allow companies to sell goods that they could not profitably sell at a single price.

The better companies understand how much people are prepared to pay for products, the greater their ability to set prices in this way, and richer data on customers facilitates this. For internet shopping this could include people’s past transactions, advertising click-throughs, IP addresses, computer/browser type, browsing history from the site or potentially other websites via tracking cookies. High street retailers can also collect information about customers’ store shopping habits and characteristics through loyalty card schemes (as in Tesco’s club card for example), but the scope for this is more constrained. The fact that prices at a store level have to be the same for all customers also restricts the ability to price discriminate relative to online retailing, and as people typically only see their own online transactions they may not know that price discrimination is occurring .[4]

Happening?

With the increasing potential importance of personalised pricing (i.e. where the price discrimination is based on individuals' information) in 2013 the UK body responsible for making markets work well for consumers, the Office of Fair Trading (OFT), reviewed the issue.[5] Based on submissions from businesses, and its investigations of complaints, the OFT found that businesses were not using information on individuals to set higher prices for them, although there was some targeted discounting.[6]

This finding is broadly consistent with Nesta’s 2012 Rise of the Datavores report by Hasan Bakhshi and Juan Mateos-Garcia. This surveyed 500 UK companies’ use of online analytics and found that optimising of pricing (of which price discrimination is just one possible aspect) was one of the least common areas that companies used online customer data for. 27% of companies did this, and the percentage was still only 39 per cent among companies generating more than 50 per cent of their revenues online.[7]

There could be several explanations for limited personalised pricing. One highlighted by companies in the OFT’s report is that it is seen as unpopular with consumers and is therefore risky for businesses’ reputation. There may also be a limited number of people with the data analysis skills that implementing a personalised pricing approach probably requires (Nesta is currently analysing the UK’s supply of data skills, for a discussion see this post by Juan Mateos-Garcia and Andrew Whitby). Profitability also has many dimensions and perhaps personalising prices is a more complicated area for companies to optimise (they may have one website and supply chain to analyse, but hopefully many customers). Alternatively, companies may be operating in markets where their prices are constrained by those of competitors.

While there is limited evidence that personalised pricing is happening in the UK, it seems likely that it will become increasingly technically possible, and as the supply of people with data skills grows more companies could have the capabilities to do it. Given this, what are some of things that might affect its incidence and impact?

Retailing

It is easier for firms to engage in personalised pricing if there are a limited number of other suppliers of a product e.g. maybe website A has worked out that I would in principle be prepared to pay £30 for good X, but if it is clear I could get good X (or a reasonable substitute) at 10 different other sites for £20 then that is less likely to be a sustainable business strategy for the website (or its competitors).[8]

Not only is a less competitive market generally agreed as a condition for price discrimination to occur, it is also more likely to ensure that the discrimination is harmful for consumers. Selling online removes the need for companies to have retail premises. This makes it easier for new companies to enter markets, which should make them more competitive. There are however arguments that in the longer-term online retail could become more concentrated. Organisations’ data analysis is likely to be more effective if it is based on varied and high frequency information on their customers over time. Which could imply that companies’ success becomes self-reinforcing, as increased sales leads to ever richer data on customers, making it harder for new companies to understand and enter the market. Economies of scale in bulk buying and distribution may also favour consolidation in online retailing, and the decline in high street retailing may make the retail market overall less competitive.

Switching

In the high street if I am less price sensitive I benefit from the actions of other shoppers who take their business elsewhere in response to higher prices, as it helps keep the prices I pay low. However, in a world of widespread personalised pricing online this is not necessarily the case, and shoppers that are more passive could face higher prices.

In addition to basic fair trading issues (i.e. being told you are getting the lowest price when that is not true – even ‘in-store’) there is a sense in which personalised pricing makes market prices less transparent. The effects of this are not entirely clear-cut. A lack of price transparency makes it harder for consumers to compare deals and switch between retailers. Though in more concentrated markets price transparency can actually make it easier for companies to copy each other’s prices rather than compete for customers (so called tacit-collusion – which may not benefit consumers) or even facilitate direct collusion such as price fixing (which is illegal).

A price on our heads

A rise in personalised pricing brings wider fairness considerations. Price discrimination can be illegal if it is based on personal characteristics in ways that violate discrimination legislation.[9] More broadly, there are distributional considerations in all this. People who are more price sensitive could well benefit from pricing being more personalised, but that won’t be everyone. A possible implication of a future of increasing price discrimination is that there will be greater benefits to us being more price conscious in our shopping.

In its report the OFT is understandably keen that there be transparency on online retailers' data collection and how that data is used (and presumably this is the view of its successor the Competition and Markets Authority (CMA), which this month replaced it).[10] There are, however, as the report also notes, issues on whether consumers engage with their own data protection, and questions as to how understandable sophisticated price discrimination will ever be for most people. Making pricing in concentrated markets more transparent can sometimes also facilitate anti-competitive behaviour like price-fixing, which can harm consumers. If personalised pricing is hard to detect, and detrimental for consumers overall, then, as occurs with price fixing, there might be an argument for legislation to ensure its strict prohibition. Economics does not currently provide strong support for this view though.[11]

Online personalised pricing may anyway not become widespread. Perhaps it will be too unpopular, or prevented by price comparison technology or competition between companies. We should not however forget that prices are just one of several ways that companies influence what we buy. Advertising, joint deals, discounts and contract terms can all affect whether we buy things, and be informed by the data our actions generate. A data-driven world, one where who we are matters more than ever, is potentially also one where there is both more, and less, choice.

 

 


[1] If the price differences reflect different costs to the supplier of providing the product then this is not price discrimination, for example with financial products (e.g. mortgages, insurance, payday loans) the same product is often sold to different people for different prices, as the cost of the product for the seller depends on the riskiness of the buyer. The pricing of, and access to, these products has been data driven for a while, a tendency which is increasing. As an example, see this discussion by Joe Deville of the role of data at Wonga.

[2] It is also not the same as an auction. Auctions are also typically trying to get participants to bid their maximum valuation, but where auctions differ is that the purchase of the good is typically rivalrous e.g. if someone buys the Picasso, or spectrum license, then that deprives the other bidders of the chance to obtain it. However, it is interesting that with the rise of large-scale auction techniques (such as in Google’s advertising auctions) auction techniques are increasingly being used in place of more traditional pricing.

[3] Theoretically, the profit maximising approach would be to charge everyone the exact price that maximises the amount that they spend on a good, so-called perfect price discrimination. This is in practice very difficult to implement, so cruder forms of price discrimination are used, although technology may allow this to become increasingly refined.

[4] Although loyalty card schemes do allow for personalised discounts, which are, in a sense, a form of price discrimination. Prices for the same product in-store can also be varied over time, and the introduction of digital price displays may give high-street retailers greater flexibility in this area in future.

[5] OFT (2013), ‘Personalised pricing: Increasing transparency to improve trust in the market’ and ‘The economics of online personalised pricing’. Both of which are available from here.

[6] OFT (2013), ‘Personalised pricing’, p2.

[7] Bakhshi, H. and Mateos-Garcia, J. (2012), ‘Rise of the Datavores How UK businesses use online data’, Nesta, pp22-23, Figure 7.

[8] One of the reasons for this is that there will be many consumers buying the good at the cheaper price who then have an incentive to undercut the discriminating firm by still reselling their purchase at a profit.

[9] The extent to which pricing algorithms might use this kind of information is something that is hard to know, but it is not inconceivable that it could be occur. There are currently wider concerns that the growing use of data could be leading to discrimination or segregation, see http://blogs.hbr.org/2014/01/big-datas-dangerous-new-era-of-discrimination/

[10] The CMA replaces the OFT and the Competition Commission (CC). The competition regime in the UK having traditionally been divided between the two organisations. The OFT also had a consumer protection remit which the new organisation is taking on.

[11] As shown in the OFT’s review of the economics literature, OFT (2013) ‘The economics of online personalised pricing’ even with a duopoly (a very concentrated market with only two companies competing) there are economic models where the introduction of price discrimination, as opposed to each firm setting a single price, can lead to intensified competition that benefits consumers. This occurs in models where each firm can set a lower price to target people who have a particular preference for the other firm.

 

Author

John Davies

John Davies

John Davies

Principal Data Scientist, Data Analytics Practice

John was a data scientist focusing on the digital and creative economy. He was interested in the interface of economics, digital technology and data.

View profile