Price Discrimination in Retail

Lexy: Welcome to the Data Science Ethics Podcast. My name is Lexy and I’m your host. This podcast is free and independent thanks to member contributions. You can help by signing up to support us at datascienceethics.com. For just $5 per month, you’ll get access to the members only podcast, Data Science Ethics in Pop Culture. At the $10 per month level, you will also be able to attend live chats and debates with Marie and I. Plus you’ll be helping us to deliver more and better content. Now on with the show.

Marie: Hello everybody and welcome to the Data Science Ethics Podcast. This is Marie Weber

Lexy: And Lexy Kassan

Marie: And today we are going to talk about differential pricing.

Lexy: Otherwise known as price discrimination or dynamic pricing.

Marie: So one of the articles that we’re going to talk about today is an article from Forbes. This is one that was published in 2014, but it does a really good job of explaining what the principles behind dynamic pricing are and how it’s being used by different companies. So we’ll have a link to that in the description for the show. The main thing about dynamic pricing is that it gives companies a way to try and get as much from each customer as they can. So Lexy, can you describe a little bit about how different companies look at dynamic pricing and why they might move towards it?

Lexy: Sure. This is one of my favorite things because it’s economics. In economics, this is called first degree price discrimination. Which means that each person is presented with the highest price they are willing to pay. And what it means is that you increase your profit as much as possible as a firm and still get the same amount of demand. Many companies have developed algorithms to help them identify how to better price items that a customer is interested in and give them a price that is potentially different than the next customer because of their own, what we call price elasticity, which means their willingness to pay more or less for that item.

Marie: And so for anybody that’s been in an economics class and you’ve seen the classic demand curve, usually you had to pick one spot on that curve to try and maximize the volume and profitability. But this type of model allows you to have multiple points on that line.

Lexy: Multiple equilibria. So normally you would pick a price and a firm would say we’re willing to produce a thousand units at $20 or something like that. We’ll sell them for $20, and then there’s a certain number of customers that are willing to pay $20 and hopefully it’s a thousand and therefore you sell all the units. That’s equilibrium. However, if you were able to say, well, we can sell 100 units at $50 and then another 200 units at $40 and another 300 units at $30 and so forth, we could get as many customers as are in the market to buy at potentially a higher price point than even we would have sat in that equilibrium.So you get to pick the entire curve as opposed to just one spot.

Lexy: Exactly.

Marie: That’s very attractive to companies, but typically in the past it wasn’t really feasible just because of technology didn’t allow you to do that.

Lexy: It’s also from an ethical standpoint, difficult to justify that you’re going to force people to pay different things. Now it’s one thing to say, well, yeah, but they were willing to pay it at the same time. Does that mean that they have to pay it? So that’s really what we’re going to talk about in some examples in these quick takes, both in this episode and the next.

Marie: Exactly. So in this Forbes article that I was referencing at the beginning of the episode, one of the quotes from a professor at the Harvard business school basically says historically, first degree price discrimination has been very difficult to implement mostly for logistical reasons with advances in technology and collection of big data. Then it may be that it will become easier to do. However, very quickly you start eliciting complaints about fairness. So that’s one of the reasons why we figured this would be a good article to base our discussion around because it provides some examples for us to explore and it talks about the ethics behind it. So even if you can, should you.

Lexy: Exactly. And while this article was from 2014, this type of price discrimination has been going on for quite some time initially. Some of the bigger… I want to say culprits, but it’s really proponents of first degree price discrimination were companies like Amazon that would show different prices to different consumers for the same item at the same time based on their prior browsing behavior, their prior searching behavior on the website there, prior purchasing and how often they would look at that item or other similar items. They would price differently based on essentially your level of interest and the perceived value of that item for you versus the next person. This was figured out essentially very early on. Consumers would call them out and say that they’re getting a higher price than their friend or something like that, and so while it was identified and there were fairness questions, it still persists to this day. It’s still very possible that if you go on Amazon, you’ll see a different price than someone else.

Marie: And then it can also even happen based on where you’re looking.

Lexy: Yes. One of the examples that we were talking about were based around where you are physically located at the time of your browsing. This was one that Staples actually got called out on and that was back in 2012 when there was a case in which two people who were maybe a few miles apart, we’re shown different prices based on where their computer was located, essentially. Their IP address. What happened was they determined that Staples was dynamically pricing based on the location of their competitors, so Office Max, Office Depot, other major office supply companies. Where there was more competition, they priced lower to try to capitalize on the consumer demand in that market and be able to steal competition away.

Lexy: The issue with this was that their competitors tended to be located in more affluent areas and so the presence of a competitor acted as a proxy variable to affluence of the area. Meaning the more wealth there, the higher the average income in that area, the more likely the presence of a competitor, the lower the price. And so what happened over time is that you were seeing higher income areas seeing a lower price and lower income areas seeing a higher price. Essentially penalizing less fortunate people

Marie: just because they didn’t have as much competition. So therefore they weren’t dynamically producing as many discounts for those areas.

Lexy: Exactly. Staples justified this by saying that – this was all online – in a normal brick and mortar location, you would do the same thing. If you knew that there were competitors, you’re trying to undercut the competition. And so you would in those locations try to make the price lower so that people were more likely to come to your store then go to a competitor.

Lexy: But again, is it fair to do that programmatically? Algorithmically? Simply based on the presence of a competitor when at the end of the day that really causes it to be a penalty to those who are poor because they don’t have those competitive presences in the market?

Marie: Exactly.

Lexy: So what’s interesting though is that there are reasons to know where somebody is located when they come to website.

Marie: Oh, absolutely.

Lexy: Apart from pricing, what would you do with somebody’s location on our website?

Marie: So one of the key things that you can do with location data is you can potentially customize an experience. So think of it this way. If somebody is coming to a site and you know that they’re in Florida, you could show them products that are more interesting to them like swimwear, whereas if you know they’re from the, you might show them winter jackets, so it helps you customize the user experience. The idea is that if you know certain things about your customer, you can personalize the experience and we know that personalization of somebody’s experience on a website can provide them with a better user experience and help them find the information they’re looking for that much faster, which is something we’re always trying to accomplish.

Lexy: Sure, and especially with location. I think about some of the buy online, pickup in store or checking local inventory to see if something maybe is in stock and they’re a local location. So if you go onto a website and it’s automatically detected your ZIP code or a ZIP code kind of near you, that might be an opportunity where they’re using your IP address, your physical location to try to customize your experience and give you a better shopping experience on the website

Marie: or let you know if there is a physical location near you versus just only being able to use their online shopping capabilities. Yep. There are definitely ways that companies can use this information to provide a better user experience should they use that though to determine the pricing. That’s where the question comes in and that’s actually how this Forbes article kind of summed up is that many firms consider targeting coupons to be the most effective way to implement differential pricing rather than risk during up resentment by using surcharges. So Lexy, you do a lot of work in this area. Can you talk a little bit about the benefit of maybe coupons? Yeah, the, the pros and cons of coupons over just this type of dynamic pricing.

Lexy: It’s interesting. It’s really more of a psychology, like a consumer behavior question at that point. So behavioral economics is one of my passions and this plays very nicely into it. What happens is that people don’t necessarily want to think that they’re getting charged more than somebody else. They feel that’s unfair, but there all right if somebody else is paying full price when they get a discount because some, there’s something special about them or their behavior.

Lexy: I So as an example, getting a 20 percent off discount from let’s say bed bath and beyond because they do this all the time. If you get a 20 percent off discount because you got a coupon in the mail and you remember to bring it into the store, you’ve done something special. You deserve that. Versus saying, well, the price when I go to pick that item up off the shelf is going to be different. When I go pick up that item versus Marie goes and picks up that item off the same shelf. The base price is the same to both of us. I just happen to have brought a coupon

Marie: and I think that also seems fair because that other person could have gotten the discount if they had joined the membership program, if they had gotten the discount in the mail, if they were a loyal shopper at that place, so they were part of the loyalty program. There are certain things where it still seems accessible even if they didn’t get a chance to use it on that particular purchase. Sure.

Lexy: Or even if they just asked, when they get to the checkout and they say, hey, that person over there has a coupon, can I get that same deal? And they say, yeah, okay, coupon code coupon code. Can you apply that for me? Whatever it might be. I mean a lot of people do that.

Lexy: What’s interesting about this example I think is that we talk about dynamic pricing as something that can be implemented online very discreetly on an individual basis. That lends itself because it’s a one to one communication at scale. Meaning each interaction is really one to one, one customer to the brand at a time, which is something that’s fairly new. Amazon started doing it long ago, but many other companies are now able to personalize to each individual that comes to their website.

Marie: Yep.

Lexy: That is not the same in store. Meaning when I go into a retail location, I have to bring something physically with me or to tell them something about me at the time of checkout for them to know that it’s me in general. They can’t show a different price at the store to Marie and I. I just have to have something else that differentiates me. In the online world, we could each be shown a different price from the start.

Marie: Absolutely.

Lexy: And so it sets a different anchor, essentially the anchoring bias of the perceived value of the item differently for each of us and

Marie: even the perceived value of the brand at that point.

Lexy: Yeah. There was one article we were looking at from McKinsey that was actually very recent. I think it was 2017. They were talking about retailers trying to use dynamic pricing to optimize their profitability and to optimize the pricing model for their items as to how much they bring them in for how much they set them on discount, where they sit in terms of profitability based on whether or not consumers have an expectation of the price of that item. So as an example, if I know that a coffeemaker that I’m interested in is a $20 item and I go onto a website and see that for $18, I might think that website has lower price items because I think that price is low. I may think, oh, their other prices must be low as well.

Marie: Yep.

Lexy: It sets that anchoring bias. And we’ve talked about anchoring bias in some prior episodes. It sets that anchoring bias that this is a low price website. I can expect I’m getting the lowest price if I go and shop there.

Marie: Or even if it’s not a low price website, it’s a place where I can get a good deal. And so I want to come back here because I know I can get good deals here.

Lexy: Exactly, so if are some specific key value items, meaning that there are items that people have a concept of what they would pay, those are things you may want to price lower so that you can bring customers back because they feel they’re getting a good deal.

Marie: Absolutely. One of the things to take away from this episode is that the price you see could be different from the price that other people see and then also for the practitioners out there that are listening, it’s to consider the fairness of the types of algorithms that you’re putting in place for your shoppers and consumers.

Lexy: Yeah. The other thing that has come up a number of times in the articles we’ve read leading up to this is the need for transparency. To say, as a provider of a price to a consumer, we are using information. Or we are pricing this based on your behaviors or what have you. Because chances are consumers will find out and it’s important that you be transparent to tell consumers that you’re using that data. If one is found out, the different prices came up for different people to explain why and how that happened.

Lexy: In the McKinsey study, it was actually quite interesting. I talked about the fact that if a category or pricing manager did not understand the algorithm that was being used and the inputs to the price they were getting, they didn’t feel comfortable using it. Transparency is not just to the consumer. It’s also to the people who are making business decisions. So as a data scientist, making sure that your internal customers, your category and pricing managers are comfortable with the algorithm, what it’s doing, how it’s providing advice so that they’re comfortable with using that advice.

Marie: Absolutely, and going back to the consumer side, the Forbes article that we started this conversation with, some companies suffered negative publicity when differential pricing is documented such as in a 2012 academic study on 200 online stores. So yeah, that just goes back to the idea that not only do the internal resources at a company need to understand the algorithm, but consumers also want to understand the algorithm because it can be negative press. If this comes out and you weren’t transparent and you were doing differential pricing in a way that customers feel like they were being taken advantage of, yeah,

Lexy: It seems as though customers are more sensitive to dynamic pricing in certain industries. So in today’s episode we were talking about retail. Next episode we’re going to talk about some other industries in which it seems dynamic pricing is simply accepted. That’s

Marie: Very true. Thank you for joining us for this episode of the Data Science Ethics podcast and we’ll see you next time. Thanks so much.

We hope you’ve enjoyed listening to this episode of the Data Science Ethics podcast. If you have, please like and subscribe via your favorite podcast App. Also, please consider supporting us for just $5 per month. You can help us deliver more and better content.

Join in the conversation at datascienceethics.com, or on Facebook and Twitter at @DSEthics where we’re discussing model behavior. See you next time.

This podcast is copyright Alexis Kassan. All rights reserved. Music for this podcast is by DJ Shahmoney. Find him on Soundcloud or YouTube as DJShahMoneyBeatz.