Dynamic Pricing, Personalised Pricing

When I was a kid in the 1970s, supermarkets were still a fairly new phenomenon for German consumers. Like McDonald’s, Bonanza and popcorn, they were considered to be “American”. Suspiciously shiny and efficient, just like Google and Apple today.

Supermarkets, in those days, were not what they are today. I sometimes wonder if, as a boy, I could have imagined do-it-yourself checkout points without a till girl (yes, they were all girls those days). And even less I would have believed that, one day, the checkout boy would not take cash from me but offer me cash instead: “Would you like to draw some cash from your debit card?”

Nostalgic Offline Pricing

Prices, in those days, were anything but dynamic. There were heavy restrictions on sales and bargains. Retailers could not just lower prices as they liked, and consumer discounts were illegal until the 1990s. The logic of these restrictions was simple:

Consumers were to be protected from hasty decisions. They were to look at the quality of goods on offer and not just at the price.

Current Offline Pricing Scenario

Prices today are much more dynamic than 40 years ago. Retailers change prices frequently and attempt to make offers that the consumer simply cannot resist. “Buy one, get one free.”, “25 pc off”, loyalty cards and vouchers are all examples of prices becoming dynamic.

Prices in German supermarkets are quite dynamic today but still man-made.

E-Commerce Pricing Scenario

In e-commerce, the story is totally different. In a recent study, Amazon was found to have changed prices more than 1.000.000 times within 24 hours (Valentine’s Day). The price ranges were remarkable: For a specific camera, the price was found to have changed, on average, 3.6 times an hour ranging between 700 EUR and up to nearly 1.700 EUR within 36 hours.

Dynamic pricing on Amazon is obviously not man-made. It is driven by big data, by algorithms. And chances are that these algorithms will soon cross the thresholds of supermarkets on a big scale.

Prices for hotels and airline tickets have been dynamic for a long time. They change, at least, daily. All changes are based on forecasts of demand. The higher the demand that is forecast, the higher is the price. The key factor of the forecasts is the up-to-date record of actual bookings. If today, against all odds, planes from Barcelona to Berlin are empty for the time of the summer fashion week, these prices will instantly drop. Just as instantly, they will double when, all of a sudden, hundreds of tickets are sold for that week.

The Rise of Pricing Dynamics

Dynamic pricing is spreading everywhere: Prices are becoming dynamic not just on Amazon but also for concert, cinema and sports tickets. The same is true for in-game items in computer games and, of course, for Uber. Uber’s success story is the success story of dynamic pricing. Generally speaking, dynamic prices can lead to increased margins. At the same time, the customer can make substantial savings by affording flexibility. A typical Win-Win: Which one of us has not found himself taking the cheap early morning flight instead of the expensive one at lunchtime because the price difference was enormous?

Dynamics & Personalisation

It is only one little step from dynamic to personalised pricing. When pricing becomes personal, the forecast becomes personal: What am I, Niko Härting, prepared to pay for a Jetblue flight from New York JFK to San Diego on a Tuesday afternoon?

In order to come up with a decent forecast, the algorithm needs to be fed with customer data, especially with data on the customer’s “track record”: What flights has the customer booked in the past? What prices has he payed? In which instances has he decided not to book when seeing the price?

Visions for Personalised Pricing

Both, amongst tech enthusiasts and tech sceptics you find visions of personalisation that go far beyond the customer “track record”. Give the airline more information and let big data do the job: Algorithms might “discover” that users like me (Niko Härting) who are active both on Twitter and Facebook but not so much on Linkedin, will typically pay significantly more for flights when booking on a Sunday. When actually booking a flight on a Sunday, such a user will, therefore, be offered a price that is significantly higher than average.

There is little evidence that prices are personalised in such a way on a big scale today. The few examples that can be found do not require much snooping. When you buy electronics and use an iPhone, you might be charged more than a customer using a Dell laptop. And sure, prices may not be the same if you order goods from Germany or from France. __________ Anzeige: __________

Big Data, Personalisation and Data Protection Law

Of course, personalisation could move faster if retailers had more personal information about their customers. Data protection law does, however, limit the ability to collect such information significantly. The customer “track record” is, no doubt, PII. Therefore, such data must be deleted according to European law as soon as they are not necessary any more for the transactions recorded. If the retailer wishes to keep an extensive record, he needs “informed consent”. He needs to inform customers, in detail, what the purpose of storing the record is, and the customer has to agree.

Informed Consent as Prerequisite

Obtaining a customer’s consent may be difficult but can be done. Even more challenging is obtaining a window shopper’s consent. If you have a hotel and know I (Niko Härting) have visited your website plenty of times without ever making a booking, you may want to use this information to make him a really good offer upon his next visit of the website. This does, however, require consent. Such consent cannot be obtained via hidden smallprint that nobody ever takes notice of.

• Tame Personalisation in German Supermarkets

Back to German supermarkets. Quite suprisingly, there is an element of personalisation in an innocent device that few customers take notice of. At many supermarket checkouts, small machines are used to print out vouchers. And the vouchers are personalised, though only fed with the data of the actual purchase. If you buy baby oil, you get a discount on nappies. If you buy vegan cheese spread, there is a tofu “two for one”.

• Scary Personalisation in the US

Many of us are scared by the story of the US supermarket chain Target that allegedly rewarded a pregnant minor with a voucher for nappies before the even girl knew that she was pregnant. The story seems too fantastic to be true, and I have my doubts about its accuracy. However, the requirement of informed consent is likely to prevent something similar from happening in Europe. Full-scale personalisation of online and offline shopping is incompatible with European data protection laws.

Blending Privacy, Transparency and Regulation in Free Markets

However, the trend to dynamic prices is likely to continue. And when the price is just dynamic but not “personal”, the focus is not so much on privacy and data protection but on transparency and consumer protection.

When the market is free, it should be the rule (and not the exception!) that prices are not fixed but responsive to the rules of supply and demand. When the pricing is left to algorithms, there is, however, a need of transparency and of control. Nobody will want pricing to be based on forecasts on a customer’s gender, religion, race, political persuasion or sexual orientation. In order to prevent discrimination and manipulation, we need some degree of public oversight over the algorithms. This is, however, a broader issue that is not specific for cases of dynamic pricing.

For a West Berliner in the 1970s, it was easy to experience static prices. You just had cross the border to the East. In East Berlin, prices were fixed by the government and shelves were notoriously empty.