Study: Online pricing is secretly discriminatory
A study from Northeastern University found discriminatory pricing on e-commerce sites to be more widespread than thought. But rather than concerns about charging different prices for the same items or steering some users toward higher-price items, the researchers only found the lack of transparency a problem.
Conducted this spring, experiences on 16 popular retail and travel e-commerce sites were tracked by conducting searches on a personal laptop or smartphone while doing the identical queries at the exact same time from clean accounts devoid of cookies and search and purchase history.
Most of the experiments did not reveal evidence of price steering or price discrimination, but the price differences were significant in some of the cases.
Among the findings from the study:
- Of the 16, four general retailers and five travel sites showed some evidence of personalization, including cases where sites altered prices by hundreds of dollars.
- Cheaptickets and Orbitz offered reduced prices on hotels to "members."
- Expedia and Hotels.com steered a subset of users toward more expensive hotels.
- Home Depot and Travelocity showed different prices for those searching on mobile devices versus desktop computers.
Price discrimination is not "an inherently sinister ploy," according to the study. Senior, student and military discounts as well as coupons are technically forms of price discrimination. The problem is that the algorithm-driven online price/search differentials aren’t obvious or clearly posted.
"In the real world, there are coupons and loyalty cards, and people are fine with that," Christo Wilson, an assistant professor at Northeastern, told the Wall Street Journal. "Here, there’s a transparency problem. The algorithms change regularly, so you don’t know if other people are getting the same results."
The researchers also stressed that the discrimination may be unintentional but a side effect of the web’s personalization algorithms.
"It’s probably the sort of thing the general public isn’t conscious of," Justin Borgman, chief executive of Cambridge big data firm Hadapt, told The Boston Globe. "You could argue it’s one of darker sides of what big data can do for you."
Discussion Questions
Do you expect an eventual public outcry over the need for greater transparency around customized pricing on the web? Do you see price discrimination as one of the “darker sides” of big data?
Personalized pricing strategies that penalize shoppers estimated to have less to spend/lower lifetime value (LTV) will attract the scrutiny of regulators.
In the U.K., an even more transparent market for pricing than the U.S., the now-defunct Office of Fair Trading conducted a more comprehensive study on price personalization and concluded that most price personalization strategies in that market benefited consumers.
More sustainable and equitable strategies might include tiered value for shoppers willing or able to spend more.
Interestingly, Amazon does not personalize price in the manner described by this article. But it is among the retailers offering shoppers more explicit options for receiving a higher level of value, for example by delaying deliveries or joining membership programs.
It depends. If you are a frequent customer, whether a member or not, you should enjoy better pricing. If you are getting higher prices than someone who is not currently using the service or has purchased before, such as what cable companies do, then that would be a problem.
People are starting to realize that if you abandon your cart before checkout, many sites will throw a better price at you to get you to finalize the transaction. Two can play at this game.
I look forward to reading the test description in more detail but my initial reaction is about the same as it was toward Edward Snowden, I didn’t feel so bad about this until you told me the details. I think everyone suspects pricing differences but it’s hard to have them shoved in your face. It’s kind of like the cashier who won’t do anything when shopping in a distant supermarket on vacation without a frequent shopper card.
But there is no reason a customer must fall victim to this, because they have plenty of opportunity to survey other sources for a better price. I think if the online retailer pushes it too far they will lose the sale. I guess we’re back to “caveat emptor.”
I don’t think big data is to blame, although it might let the retailer better understand the impact.
Price discrimination is one of the darker sides of retailing. Big data just makes it easier.
Will there be a public outcry? That depends a great deal on whether or not a “class” of shoppers (i.e., a demographic cohort) is being—or appears to be—systematically discriminated against.
The more interesting question is whether these pricing strategies will make certain channels or platforms more attractive to shoppers who determine they are the best path to discounting.
One of the issues is the difficulty for individuals to know whether or not the price they are paying is different from the price anyone else pays. If groups conduct trials or regularly monitor sites, find differences and publicize them, then individual consumers will be aware of differences and may increase concern. The lack of transparency keeps individuals in the dark, and not knowing makes it possible for alarm if differences are recognized and publicized.
As we move toward personalized pricing and promotions, it will be very difficult to de-tangle discrimination from “personalized pricing.”
After all, the goal of personalized pricing and promotions is to elicit a desired response, based on who you are. So how can one’s personal characteristics not affect the offer? Is this discrimination?
I’m not a lawyer, but I believe that if a policy (such as an algorithm for offering goods or services) has the effect of treating a particular group of people in a less advantageous manner than other groups, even if there is no intent to discriminate based (for example) on an innate characteristic of that group, there are legal considerations. One example that comes to mind is the utilization of zip codes in determining insurance rates.
Consequently, regardless of the intention, this is a problem that we who engage in personalized pricing and promotions will need to consider carefully.
Price differentiation between customers is not a new concept. Supermarkets and others have had price zones for years. They certainly never issued statements that a customer on the North Shore of Chicago was paying more for an items than someone might elsewhere in the market.
With the internet, availability of data means a site might offer difference prices based on where your IP address is located. It also means you can do it by customer based on what you know about them.
Is this right? I am not sure but I really don’t expect a public outcry over it.
Since we’ve had this model in place with the airlines and a few other companies for some time now without riots or demonstrations, I doubt you’ll hear much more than the type of complaints we hear about them: grumbling. And don’t forget about Starbucks. Go to London and see what they charge for a grande blonde roast vs. say, Columbus Ohio (about double, FYI).
Frankly, IMO, it’s just smart business.
There is already considerable frustration at the seemingly-random pricing found in travel. Will it convert to outcry? At some point.
Saddest for retailers is that the impact of this irrationality delivered to customers will be a trickle—increasing rejection of that retailer’s product. But it will be a long time before the trend is strong enough to detect in the data.
Why do they call this price discrimination? It is simply pricing. If the customer sees the price and likes it, they buy. If they don’t like it they keep looking. Since when is it incumbent on the retailer to have pricing transparency? If anything, the internet has given the price advantage to the consumer. They can keep looking and looking and looking until they are satisfied, if they want.
Personally, I find myself much more satisfied that I am getting a good price online than in a store. I certainly have many more choices.
If big data analytics are used to identify loyal shoppers and then used to reward those shoppers, I believe that’s a great thing. If there is a more “sinister” utilization of this data, then that’s something to deal with directly.
Customized pricing online is a great tool for testing pricing elasticity of demand in real-time, and when used for the right reasons (i.e., to reward frequent buyers, members, etc.) the consumer benefits from the technology.
Sure, there can be a shadow side to customized pricing, such as when higher prices are shown to the same consumer set, but solely based on prior search history, zip code, etc. It really is up to the customer to determine if the price shown is of sufficient value to make the purchase.
Consumers have clearly shown that they will accept and even seek out retailers with dynamic pricing programs. Safeway “Just for You,” CVS “My Weekly Ad,” and Target “Cartwheel” are all examples of dynamic pricing programs. My favorite example, is the San Francisco parking meters that charge for parking based on current demand. All of these dynamic pricing models work, because they are transparent to the shopper.
What doesn’t work is price obfuscation, and secret pricing programs based on undisclosed criteria. Information is too easily shared on the internet for brands to think this can still work. Services like CamelCamelCamel, Decide.com (recently purchased by eBay), etc… make obfuscation a non-starter.
Amazon is changing 2.5M prices a day and is growing faster than almost any retailer. Shoppers see those price changes in the form of clear notifications when an item in their cart changes.
Clearly, dynamic pricing can be a competitive advantage when done in a transparent way.
Discriminatory pricing is an unfortunate description which conjures up images of pricing based on unacceptable factors. However pricing based on offering individuals incentives that are tailored to their purchase decision criteria has been around in the form of coupons for some time.
Retailers know that purchase decision criteria vary from one individual to another. By offering a single price to all customers the retailer is running the risk of pricing too high for some customers (disappointing some customers and reducing the volume they can sell) and of pricing too low for others (reducing the brand equity of the product and delivering lower financial performance for the retailer). Personalized pricing improves the overall efficiency of the market by ensuring that the right volume of goods is sold at the right price for customers and the retailer—it’s micro-economics in practice. If the retailer is charging a price that is too high and not delivering acceptable value across the entire value proposition then customers will go elsewhere.
Having said all of that, retailers do need to be careful about how they deliver personalized pricing and how they manage communications around it. No one likes to feel that they have been discriminated against but I think consumers understand the concept of rewarding loyal customers by providing them with personalized incentives. Technology is now allowing companies the flexibility to deliver personalized pricing that enables them to be much more customer-centric and retailers need to embrace the concept of personalized pricing, but they always need to do so with a wary eye on the potential backlash that could result if they manage it inappropriately.
Public will outcry over everything—as soon as they are bored with their current cry. Memberships with service providers—like hotels or airlines—demand a better price. People with a purchase history at the retailer’s site should be given an edge. Want a better price? Fill the cart and leave it.
Dynamic pricing is not going away, and customers by and large will not be too bothered by it. For some types of products, negotiation is a part of buying: big ticket items (cars, appliances, furniture). Prices paid change by customer ability to negotiate or ability to pay.
Also, as another commenter pointed out—airlines have priced to different customers for a long time.
As much as retailers have to fight the good fight on margin, you can expect them to use any method that might raise profits.
Algorithms and big analytics are smart business if they improve the probability of the potential buyer buying. I realized that this was happening to me while I was playing the pricing game with a couple of popular “blind” hotel travel websites. The built-in algorithms knew my sensitive price points after having used their services many times. So be it. This is ultimately supply and demand, the consumer makes the ultimate decision. Where retailers negotiate, who knows what each other has paid? Are loan rates discriminatory because actuarial science determines risk?
Any time business prudently uses the laws of probability to predict an outcome, the term discrimination is misplaced.
I do not believe that there will be a public outcry for pricing transparency. The cases are too sporadic and it is not clear which consumers are benefiting and which consumers are losing. In some cases, the same consumers may be benefiting and losing, just in different situations.
This is not price discrimination; rather it is customization of pricing based on consumer situation most of all. Think variable pricing for theatre tickets. This is “much ado about nothing,” I believe.