Online retailers found to shift offerings and prices
You know that friend who always scores hotel deals you never seem able to find? The reason might not be her skill at ferreting out online bargains.
Travel websites and other Internet retailers may be giving your friend better deals than you as part of a high-tech experiment called price steering that gives consumers different search results based on their buying histories, tastes — even the types of digital devices they shop on.
When browsing shoes, for instance, a connoisseur of handmade leather may get different Internet search choices than someone who typically buys knockoff brands.
And in some cases two consumers may actually get different prices for the exact same item, a practice known as price discrimination: That friend of yours may have been quoted $202 a night for a weekend stay at a Miami hotel, while you got $243 for identical accommodations.
New research to be unveiled by Northeastern University Thursday found that major travel websites and general merchandisers are testing price steering and discrimination in small doses. The techniques can make it easier for online consumers to find what exactly they want, or conversely, harder to get the best deal.
"You do a search for an item and you're shown some results, but you don't know if those are all the results," said Christo Wilson, a computer science professor at Northeastern and coauthor of the study. "Maybe there's more stuff or better prices that are being hidden from you."
The Northeastern researchers focused mostly on travel, an industry known for variable — and sometimes mysterious — pricing. But they also tested general merchandisers, finding instances where similar searches for dinner tables, for example, yielded different results: Some shoppers were shown expensive options with gilded mahogany, others got the cheap plastic.
Price steering and discrimination sit at the intersection of two of the hottest trends in high technology: personalized web browsing and big data analytics. Users of Google's search engines have undoubtedly noticed its knack for correctly guessing queries even before the typing is finished — as if computers have turned into mind readers.
For retailers, all that data generated when you visit their sites — browsing and buying patterns — can help them tailor search results to what they think you really want and — critically — are more likely to buy. Ideally, sellers can push customers toward the upper end of their spending limits because they know the price points at which they're likely to make a purchase or walk away.
Conducted this spring, the Northeastern study examined the e-commerce websites of 16 well-known outlets and found nine, including Orbitz, Travelocity, and Home Depot, that sometimes delivered different results to different customers, even those who used the same search terms. Researchers created dummy accounts to conduct tests and also provided a team of consumers with scripted searches to perform.
Importantly, the researchers concluded tailoring search results to particular types of users is not a widespread practice; discrepancies were infrequent, but in some cases the differences were significant.
"The key take-away is, this is going on," said Wilson. "Companies have a lot of personal information about you, and we do see cases where that is being used to either change prices or change the items that you see. It's difficult for you as a consumer to protect yourself because these systems are black boxes, and they change all the time."
This issue has surfaced within the travel industry before. In 2012, Orbitz acknowledged steering Mac users to higher-priced hotel rooms than PC users on the theory they have more expensive taste, though the company said the experiment was short-lived.
Responding to the Northeastern report, Chris Chiames, a spokesman for Orbitz, acknowledged the travel site displays different search results and sometimes sets different prices for identical hotel rooms. Each technique has a business rationale behind it, he said.
When users of GPS-enabled mobile devices search for lodging, Orbitz can detect their current locations and automatically steer them toward nearby hotels — something that is not possible for desktop users. That can explain some differences in listings, Chiames said.
In other cases hotels earmark small numbers of rooms for discounts and make the low prices available only to customers who use Orbitz's custom mobile shopping app.
"The cost of acquiring that customer is much lower, as they are coming directly to our app," he said. "We are then able to better reward that member and share the lower customer acquisition cost with more generous rewards."
The rewards can be generous indeed. On Wednesday afternoon, for example, The Boston Globe searched Orbitz for hotel rooms in Miami using a desktop PC running Internet Explorer, a laptop computer running Mozilla Firefox, and an iPhone using the mobile version of Safari.
Results on the two computers were the same but differed greatly from the iPhone. For example, Orbitz quoted a nightly rate for a standard room at the Clinton Hotel South Beach for this weekend that was $41 less on the mobile device. Using Orbitz's mobile apps gets an even lower price.
Another travel site, Kayak, also quoted a special rate at the Clinton Hotel for mobile shoppers — $13 less. On Priceline.com and Hotels.com, the room rates for the Clinton were the same on the computers and iPhone searches.
Several other online shopping sites singled out in the Northeastern report did not return calls for comment.
Software formulas can assemble a customer's data into a rough profile: Someone who lives in Wellesley, works at a law firm, and shops using a new i-Pad is likely to have expensive taste and a sizable budget. So the customer's search results can be modified accordingly.
Pegging some shoppers as big spenders has origins in brick-and-mortar stores. A older, well-dressed man browsing jewelry at the mall before Valentine's Day might be nudged toward a pricier display case than would a high school boy, for instance.
And while price discrimination is a loaded phrase, it is widely accepted in certain contexts — such as when senior citizens and veterans receive discounts because of age or military service.
Boosting sales may be retailers' chief motive, but there can be real value for consumers, said Justin Borgman, chief executive of Cambridge big data firm Hadapt. Borgman said he has a thing for expensive shoes and would prefer high-end products when he searches online.
But retailers aren't always transparent about their pricing strategies, Borgman noted, and that could backfire on them with consumers.
"It's probably the sort of thing the general public isn't conscious of," Borgman said. "You could argue it's one of darker sides of what big data can do for you."
Results may vary
Identical searches on popular travel site for a two-night stay in Miami this weekend sometimes yielded different prices on computers and mobile devices.