Two years ago, Best Buy CEO Brad Anderson caused a huge controversy in the marketing industry when he proclaimed in a Wall Street Journal article that he wanted to get rid of up to 20 percent of Best Buy's customers. Realizing that some of his customers cut into profits significantly by insisting that Best Buy honor its pricing guarantees, by actually redeeming rebates and by purchasing a higher percentage of loss-leading products, Anderson made waves when he referred to these deal-seekers as "devils."
As those waves reverberated throughout the marketing trades, it seemed that marketers of all stripes cleanly separated into two camps. One agreed with Anderson, wondering why a retailer wouldn't want to bucket its less profitable customers and treat them accordingly. The other camp thought it unfathomable that a company that claimed to be "customer centric" could have such a negative attitude toward such large segments of its customer base.
Gentlemen, we have the technology
Behavioral targeting technology allows for the type of segmentation that might have helped Anderson figure out which 20 percent of his customers were the devils. When many of us think of behavioral targeting, we think of it as something that helps us hone in on the sweet spot of our target market.
But what if the same technology is applied to figure out who the devils are, such that we can avoid targeting them with ads altogether?
It appears that some companies in the space are getting there. [X+1] CEO Toby Gabriner says that if behavioral targeting principles are combined with technologies like progressive optimization and advanced audience profiling engines, you wind up with something he calls "audience screening."
Audience screening allows advertisers to determine if a prospect is likely to respond to an ad. If so, the prospect sees the ad. If not, he doesn't. According to Gabriner, the decision is made in near-real time.
"This process happens in fractions of a second and can be used to better qualify the audience, allowing advertisers to reach only those folks most likely to convert as well as the publisher to generate a higher cost basis for their inventory," he said. "It's a higher reward, lower risk model which is becoming more effective as time moves forward."
Where might this be heading?
As advertisers, we've been segmenting online ad buys by behavior for years, trying to identify our best customers. It seems only logical that we apply the same technology to help save money on our media buys by skipping the folks we're reasonably certain won't respond.
But it begs the question: If we know a lot about our least profitable (or even unprofitable) customers, how long before we use that information to develop profiles that help us skip over the deal-seekers and tire kickers? Is Anderson's concept of devil customers alive and well two years after it created such a stir in the industry?
Conceivably, an online electronics retailer could profile its customers in a number of different ways to identify the devils. Logfile, site analytics and customer purchase history data could easily provide information on which customers repeatedly redeem coupons and promotional codes featured in deal-seeker newsletters like GotApex? or Fatwallet.com. We could also index profitability against purchase history for every customer in a retail database, or look to external resources to see where the devils might be coming from. A few relatively simple algorithms could easily limit exposure to unprofitable customers.
Cut off or cooperate?
To me, though, that approach seems lazy. I'd rather spend time looking at what separates unprofitable customers from profitable ones and seeing if the retailer can do anything to generate some profit from its unprofitable segment. Customers are people, and it's highly unlikely their goal is to hack the system for the sole purpose of screwing the retailer. Odds are they're just looking for a good deal.
A good retailer should be able to figure out a way for both parties to be happy.