Can you use personalized recommendations to automatically create the “perfect salesman” for your website? Yesterday’s guest David Selinger from richrelevance thinks you can. Check out this interesting take on customer targeting.

Futuristic visions of personalized shopping and their place in the everyday consumer experience have been a topic of discussion for some time. In 2002, even a Tom Cruise flick, Minority Report, portrayed the concept in an arguably plausible future (and just a touch creepy!)—Tom cruising through a Gap store where an ad welcomes him back and asks if he enjoyed the shirts he had bought previously.

So how close are we to this futuristic vision of the shopping experience? And more importantly can personalization also be consumer-friendly?

The answer is definitively yes, as exemplified by the evolution of the online shopping experience—rudimentary interfaces have evolved into full-fledged personalized experiences as seen on Amazon, which recommends items to users based on their shopping behavior. With interactive tools at hand and an increasingly savvy consumer base, the online world has been shown to be the best proving ground for highly personalized product recommendations.

So what are the benefits of these targeted recommendations? Until recently, most online retailers have required users to explicitly express what they want—restricting them to direct keyword searches or navigation through the layers of categories and subcategories of products. Unfortunately, this does not serve to elucidate the expanse of inventory available as these methods are user-action driven: the consumer asks for X and hopes to get X or something similar. Personalized recommendations confront this issue directly by bringing online the general appeal of the in-store shopping experience—the natural navigation process by which customers may browse from one clothing rack or department aisle to another. These recommendations not only act as a window into various product categories, they also accommodate each customer’s preferences as demonstrated by their behavior.

Despite this fantastic vision, personalized recommendations of the caliber seen on Amazon’s have been incorporated into only a few sites due to the high cost of development. Fortunately, several companies (including my own) offer SaaS technology so that any retail site can personalize their shopper’s experience—creating an interaction that one might describe as having a perfect salesperson laying out relevant products a shopper is most likely to be interested in.

Let’s further refine this definition of the “perfect salesperson”. The perfect salesperson would put the shopper in charge, enable him to naturally and effortlessly move from one product to another, from one category to another and most importantly, to choose where he wants to go next, but providing relevant input at just the right time. For example, let’s revisit the personalized shopping vision proposed by Minority Report. Tom Cruise strolls into the men’s pants section where he is shown not just one product recommendation but three. On his right, a screen suggests “Better Together” showing particular shirts that go with the jeans he’s just picked up. Simultaneously he’s shown a recommendation for “Top Selling Jeans,” in case the pair he currently has in hand don’t quite do it for him. Thirdly, he’s shown “People Who Browsed the Jeans Section Also Browsed these Shoes” While looking at one product, he’s been given multiple perspectives on the product—comparables, related products and so on.

Rather than assuming that Tom will react to one type of recommendation, this “perfect salesperson” (or salesperson-computer-screen) has taken into consideration Tom’s individuality as a shopper—offering him multiple types of recommendations and enabling him to determine what best appeals to him, right at that moment. Today, it’s the shirt. He’s in and out with a great pair of jeans and a sweet matching shirt. Tomorrow, when he comes back, though, it may be the shoes he pairs with his jeans. Ultimately, there is no single “best” set of recommendations since no two customers shop alike. But, the best salesperson knows to communicate the context of the product recommendations.

This scenario is captured by the technology developed by my company, richrelevance: “ensemble learning”. Ensemble learning is the only technology that not only actively suggests items that reflect each consumer’s interests and behaviors, but uses more than 15+ distinct recommendation types. The approach sharply contrasts the traditional method of adding to every page a box that says—“We recommend these items.” As illustrated above, richrelevance embraces the individuality of a shopper, which describes our job as finding numerous relevant types of recommendations, and letting the shopper decide which recommendation is best for him. Further, ensemble learning is 100% transparent to the shopper, displaying recommendations using clear, crisp messages, empowering the shopper to choose how he wants to shop. To take this a step further, ensemble learning also constantly evaluates its own performance, testing different approaches and checking what works. Based on this closed-feedback loop, the system optimizes in real time choosing the most effective recommendation types for each stage in this shopper’s experience. The result is constant learning and optimization.

So while we probably won’t be seeing the world of the Minority Report in everyday life anytime soon (and probably don’t want to!), we’re now experiencing the benefits of personalization online—an increasingly intelligent and most welcome tool. Customers can trust product recommendations from a system that is obviously learning and approaching them as an individual, offering complete relevance and complete transparency.

Tags: cross-selling, customer targeting, eCommerce, personalized recommendations, up-selling, Upselling