Is Big Brother Pricing Your Groceries?
Surveillance Pricing and the Food System
Surveillance Pricing. It sounds ominous, doesn’t it? It sounds like Big Brother is pricing your groceries. It’s getting a lot of attention from politicians who are saying the practice should be banned. It’s worth thinking a bit about what it is and whether it would work in a grocery context.
Most of us assume that when we walk into a store or visit a website, everyone sees the same price. That assumption has never been entirely true. Airlines have long used dynamic pricing. Hotels charge more during busy periods. Students and seniors get discounts at the grocery store on specific days. Coupons target specific customers. Loyalty programs offer discounts to some shoppers but not others. In all of these cases the seller knows something about you (even if it is only which day you are there) and offers you a price based on that knowledge. Product form can also be a factor, but it doesn’t require previous knowledge of the consumer. Consumers self-identify by choosing a specific product form. In fact, tracking these sorts of purchases can provide insight to the algorithm to target other products and prices. This doesn’t just mean variety like flavours or other versions. There are multiple different versions of eggs in the grocery store: large, medium, extra-large, organic, free run, free range, solar powered (barns not chickens), brown, omega, and combinations of those as well. The differences differentiate consumers but so do pack size. On a Loblaws website today I looked at large white eggs. There were at least five versions at different prices for the exact same eggs. A dozen large white Gray Ridge eggs sell for $0.44 an egg. If you buy an eighteen pack, they are $0.41 an egg. On the other hand, if you buy a half dozen, they are $0.53. If you choose no name eggs (where the difference is not clear at all) you pay $0.33 an egg for a dozen and $0.32 an egg for a package of thirty.
Advances in technology have made it easier to get more details about you from Bluetooth connections, browsing history, previous purchases, and other things. The thinking is that electronic shelf labels will allow stores to change the prices as individual customers look at them. That may be difficult, but they could do it based on time of day or the specific day of the week. With online retailing, this sort of individual pricing is much easier because only one person is seeing the price and that price can follow the consumer directly to checkout.
The enhanced ability to price differentially due to technology raised concerns which led to a change in terminology from differential pricing or revenue management to surveillance pricing. Semantics can set the tone but let’s talk about what this really means and how likely it is.
I used to teach revenue management to hospitality students – hotels and airlines are at the forefront of this practice historically. While it is not critical, it is easier to differentially price when a product has one or more of these characteristics.
Fixed capacity – a seller cannot adjust supply up or down in the short term. A hotel only has a fixed number of rooms, and an airplane has a fixed number of seats. If demand is higher then they cannot meet it. This is why if flights are close to selling out, ticket prices rise.
Perishable products – perishability in the context of food is obvious but a hotel room is
also perishable on a given day. If you don’t sell it, you can’t carry it over like you could a pen or a cell phone. The room for that night is gone. That is why flights with many seats left as the flight date arrives, ticket prices fall.
High fixed and low variable costs – lower variable costs provide more flexibility to differentially price. The cost of one additional passenger on the flight is very low. It may just be a free drink and some peanuts. Staffing, fuel, plane maintenance are paid whether one additional passenger flies or not. Cost of goods in a grocery store varies but averages about 75% which leaves some but not a lot of flexibility for pricing.
The ability to identify and differentiate between customers or segments of customers. Seniors on senior day. Loyalty members who are regular customers. This is where the new technology comes in. It becomes easier to collect information on specific customers. The bigger challenge is keeping them apart or having them accept the differential pricing. I understand why seniors get a discount – particularly if they are getting it on a day that is less popular and the business is managing capacity. Stores don’t provide student or senior discounts on a weekend when they are the busiest. If it becomes obvious that people are being charged differently and they don’t understand it, you run the risk of losing those customers in the long term. That is an especially big deal for a grocery store that wants customers to shop there regularly.
How Does It Work?
In an online shopping context this works much easier. Two shoppers on the same website. One has visit several competing websites which may mean they are price sensitive. The second shopper is a regular customer and does not appear to shop around. If the algorithm decides that the first may leave if they don’t like the price and the second is likely to buy as they have in the past, they might get different prices. The retailers see themselves as being strategic and increasing revenue. Two sales instead of one. If customer two finds out they paid more they may feel that they have been unfairly treated. It might mean they don’t shop there anymore. That is the risk the retailer faces.
Advances in artificial intelligence and data analytics make this increasingly feasible. But grocery stores are a more challenging case. Consumer Report had 400 people all order the same basket through Instacart at the same time across the US. They found wide variation even within individual markets. Instacart didn’t admit to wrongdoing but did say that they were “thoughtfully pricing certain items with high or low price sensitivity” which led to “improved price perceptions.” They also promised to stop doing it.
In a physical store context, it is more difficult. Coupons allow for price differentiation. Offers can come across an app for loyalty members. It is more difficult to change prices in the aisle for individual customers. There is likely some ability to track using cell phones and collect some information. The fear is that digital price labels can be changed for individual customers. It would take some real coordination to make that work as the price would have to be carried forward to the cash as well. You would also have to be confident that nobody else saw the price or noticed it changing.
What is the Case for Individualized Pricing?
There are several arguments proponents make for differential pricing. The first is that it’s been happening for a long time and technology is allowing companies to fine tune their offerings to consumers. There are discount days for students and seniors. In that case people must self-identify. The added benefit to retailers is they can use the price offering to balance demand by encouraging students and seniors to come on a less busy day. That benefits those who shop on busier days. There are coupons which allow not only trial of a new product but access for more price sensitive customers. There are other strategies too: loyalty offerings, weekly specials, flyers, price matching for those that shop around, and price reductions on products close to their best before date. The argument is that price sensitive or low-income consumers will have access to products that they might otherwise not have.
Airlines are a good example of this. Differential pricing allows leisure travelers (who book early and/or stay over Saturday night) to access cheaper fares. Business travelers want more flexibility and often have less notice, so they often pay more. Without that pricing structure, average prices might increase and total flights might actually decrease.
What is the Case Against Differential Pricing?
There are three primary reasons people argue against it.
The first objection is fairness. The argument is that people who buy the same product under the same conditions should pay the same price. If and when consumers discover another customer paid less simply because of an algorithm, they may view it as fundamentally unfair. A fundamental tenet of revenue management is the ability to separate different segments (even if the segment is one person) and/or to be able to tell them why the other customer got a better deal. People are generally fine with special offers to seniors, students, or loyalty members. They will be less understanding if the reasons are opaquer. This is a risk for companies, especially those like grocers who want customers who shop regularly. This is a long-term relationship that could be compromised with a perception of unfairness. Because we shop so frequently, we also have much clearer expectations of grocery price levels.
The second concern is privacy. Setting individual prices requires data. The more data you have the better the algorithm will be at setting prices. For many consumers it may be more upsetting to know the amount of information the company has about them than the fact that they are using it to set an individual price. This goes beyond websites tracking their purchases to location data, browsing behaviour, or other personal characteristics being used in the algorithm.
Finally, even when consumers understand that there is a lot of personal information, the lack of transparency about the algorithm is unsettling. How are prices set? How does their price compare to other’s – is it higher or lower. Trust depends on transparency and this is moving in the opposite direction and again highlights the risk for companies who have to walk the line without crossing it and losing customers.
What Does All of That Mean for Food Retailing
Food retailing poses both an interesting opportunity and a specific challenge. Retailers already use loyalty programs extensively. These programs collect significant amounts of information about purchases. Retailers will know what products consumers buy, how often and when they shop, how sensitive they are to promotions, among other things. That allows targeted offers and customized coupons rather than individualized shelf prices. There are suggestions that digital shelf labels will make it easier to individualize prices but that comes with some challenges. If you are flashing prices on the labels there are two potential challenges:
It is difficult to ensure that a customer who is offered a lower price is the only one who sees the lower price on the label.
There has to be coordination so that the price on the shelf is consistent with what is charged at the checkout. Both staffed and self-checkout scan the bar code to determine price. There would need to be tech that identifies the individual again and adjusts to register the correct price.
I expect tech can make the connection, but a crowded aisle may make it difficult to show individual prices to individuals. Inconsistency in prices could cause problems with customers. Loyalty apps can achieve the pricing connection without showing the price to other consumers and they are scanned at checkout to easily validate the price. It’s just not clear they can easily do it for other customers.
It’s also worth talking about grocery strategy. There are two priority objectives; share and basket. Share is how many people come into the store and basket is how much they buy when they are in the store. They already use loss leaders – products priced cheaper -to draw people into the store. These are products that are staples and are in most shoppers’ baskets. They are the ones that are the first level of price comparison. The other products are those that build the basket. They are considered to be less price sensitive and generate higher margin. That does not mean, however, that grocers can simply increase these prices at will. Grocery apps have lowered the cost of price comparison. Many stores offer price matching if you find a product cheaper somewhere else. While that might sound like an out for stores who are risking higher prices through surveillance pricing, if it happens too often, the customer might just switch. Stores still regularly publish flyers which sets a price ceiling. My guess is that if there is any individualized pricing it would be downward to build basket rather than upward to try to extract more margin and risk the whole basket. I might be naïve but there are more moving parts and significant risk for grocers to take this too far.
Do We Need Regulation?
It is worth remembering the Instacart got caught doing this and have stopped or at least modified their behaviour. It is easier to execute this sort of differential pricing in an online environment. The purchase frequency does create risk of noticing pricing differences, but they may not occur if the characteristics of the customer are identified early. The customers who use Instacart might also be less price sensitive than those that shop in store. Journalists and others will watch and hold companies to account. It is not perfect but can play a role.
There has also been considerable development on tech to track prices and hold retailers accountable. There are web sites that track the differences in airfares over time and advise on the optimal time to buy a ticket. Hotel aggregation sites advertise that they can find the best price for any given hotel. They also get a margin from the hotel when they book a room. This reduces the incentive to price differently on different platforms. These are painful and costly lessons that companies have learned both financially and reputationally. I expect the existing shopping apps will incorporate adjustments for price variation. This might set a price ceiling for surveillance pricing limiting pricing actions to discounts. There is still a question of whether it is acceptable to differentially price based on personal information gleaned from a variety of sources rather than just the information they use today.
I am not arguing that there is no incentive or possibility for surveillance pricing in other markets. Products that are purchased less frequently in contexts where price is not as obvious to other consumers are ripe for algorithmic pricing. I am arguing that the likelihood is low of significant price differences in food, particularly of significant price premia. We have to ask ourselves if any price differences are discounts to expand the basket or if there is a real possibility that grocers will take the risk and charge some customers more in an effort to grow margin and, if so, what the criteria would be.
Regulation can also work in unintended ways. Will we preclude offering special deals to loyalty members based on previous purchases? Will we preclude bulk discounts where you get a discount if you buy two or more? Will we preclude seniors’ discounts? Do we risk increasing the average price even if some people won’t pay more? What about grocery chains who have discount brands? Proponents of regulation will say my examples are ridiculous, but we need to be clear on what our objectives are and how we differentiate between acceptable and unacceptable practices. The real question is whether the issue to be regulated is differential prices or how the targets are identified. That distinction matters and will affect how it can be enforced.
In the end, I believe that regulation would be a solution looking for a problem. There are real technological checks and balances and also external parties that will be vigilant. Instacart got caught. This issue is getting a lot of attention. People pay more attention to food prices than to other prices. If the focus is on food prices this is unnecessary. If the focus is broader, we need to be very clear about what and why were are regulating to make it clear and enforceable without unintended impacts on consumers.
Keywords: surveillance pricing, revenue management, food, food prices, technology, grocery
Comments