Because you bought cross-selling, you may also like ...

Product recommendations—they help customers discover products and retailers build sales. But there is more than one method to choose from, and which is best is the subject of debate.

There’s a sweet fragrance in the air at Scentiments, and it’s not from any of the e-retailer’s myriad colognes or perfumes. It’s coming from newly implemented product recommendations technology. And this fragrance has created quite an attraction.

“These recommendations have far exceeded our expectations,” says Howard Wyner, CEO and chief of e-business. “There are astounding conversion rates in the double-digits. We scratch our heads—this is unbelievable.”

Scentiments switched on product recommendations in August. Since then, total sales are up 10% and average order value is up 40%, it reports.

“We have a pricing structure that is very competitive that aids in the ability to make a cross-sell,” Wyner says. “We’re saving customers 80% to 90% in some cases. So if they’re going in for a refill of their favorite fragrance and see that they saved $50, and then are presented with some strong recommendations, they say, ‘I’m saving so much on this bottle, I’ll roll the dice and give this other product a try.’”

Virtual salesperson

Such product recommendations act like a virtual salesperson: They step in while shoppers are considering or selecting products and suggest other products the shoppers might be interested in. The more products shoppers view and buy, the better the recommendations can become because the systems coming up with the suggestions have more data to work with.

Many e-retailers rely on vendors such as Baynote Inc., ChoiceStream Inc. and MyBuys Inc. to create and maintain product recommendations systems. Though some retailers build systems in house, they are more often delivered by vendors in a software-as-a-service model, meaning data from the retailer site flows to a system hosted by the vendor, which displays recommendations on the e-commerce site.

Vendors generally divide into two camps when it comes to how they come up with recommendations. The first method builds a profile for customers based on their actions as individuals on an e-commerce site. The second creates profiles based on the activity of all shoppers, then bases suggestions on the profile the shopper seems to fit into given her current behavior. (Some systems that focus on individual profiles can also take into account group behavior at some level.) Which of the methods is best is the subject of debate among vendors and e-retailers.

Both methods rely on anonymous cookies (see story, page 38) that give each customer a tracking identification number, allowing the recommendation system to track that consumer’s behavior. E-retailers place JavaScript tags on pages to track events and send that data to a recommendations server.

E-retailers provide product recommendations for cross-selling purposes at different points during the online shopping experience. Some e-retailers use recommendations right from the start, on the home page. A returning customer hits the site and is presented with products the e-retailer thinks she will like. More commonly, e-retailers wait, displaying recommendations on category pages, product pages, search results pages, in shopping carts and on order confirmation pages.

54% of U.S. online shoppers notice product recommendations on e-commerce sites, according to a 2007 survey by Forrester Research Inc. 34% of those shoppers say they have made purchases based on recommendations.

“Consumers are often persuaded by recommendations, as recommendations help them discover products they might not have been familiar with otherwise,” says Sucharita Mulpuru, a principal analyst at Forrester Research.

Browsing around

Bookseller Borders considers product recommendations integral to e-commerce, and they play a prominent role in the multi-channel retailer’s new web site launched in May after the retailer moved off of the e-commerce platform of Amazon.com Inc. “Recommendations are critical to the online customer experience,” says Kevin Ertell, vice president of e-business at Borders Group Inc., which uses software-as-a-service product recommendations from ChoiceStream.

While some shoppers know exactly which book or film they want and are looking to get in and out quickly, more than half prefer to browse because they don’t know what they want, Ertell says. Borders’ web strategy is to create a bricks-and-mortar bookstore experience online, letting customers “walk around the store” and browse to find things of interest. Product recommendations are key to this strategy.

“Recommendations are a great way to find what you’re interested in and then continue along a chain of recommended products that keep appearing,” Ertell says. “You get a depth and breadth of interesting items along the way during the online experience.”

Borders displays product recommendations on product pages, the shopping cart page and the order confirmation page. The greatest opportunity is in the shopping cart, Ertell says.

“It’s the same as placing products by the cash registers in the stores. It’s a chance to see something else on the way out,” Ertell explains. “Plus, we have free shipping thresholds, and product recommendations on the shopping cart page can encourage customers to buy something else to get them over a threshold.”

More likely to buy

Borders is happy with the results to date. Shoppers are more likely to buy a product going to a product page via a recommendation than they are going to a product page on their own, Ertell reports. And shopping carts that include recommended items have a higher average order value, he adds.

“Product recommendations have proven their worth,” Ertell says. “It’s not an overly significant cost to implement them, and the return on investment is definitely there.”

This is also the case at bridal accessories retailer The Knot Inc. It cost $7,000 to implement Baynote product recommendations in March, and the retailer pays an undisclosed monthly fee. In return, The Knot credits product recommendations with lifting overall sales 15%.

Recommendations appear on product pages and search results pages. The retailer is considering adding recommendations to the home page, replacing the current list of top-selling products.

“A customer can be looking at some candy, and the recommendations system will show perhaps a place card holder and ribbons to go with that candy,” explains Kristin Savilia, vice president of e-commerce. “For all other customers who bought that particular candy, they most often were looking at or buying these other products, and the recommendations system pulls to the forefront some combination we the retailer would never even have thought of.”

Staff at Scentiments were being forced to think up cross-sells. Prior to implementing product recommendations, the e-retailer regularly received questions from customers by phone and e-mail asking for fragrance suggestions based on other fragrances they liked. Now the recommendations system does the work, and it does so in a more precise and effective manner than humans can, says Wyner of Scentiments, which uses software-as-a-service recommendations from MyBuys.

“It’s about the customer experience—we wanted to set ourselves apart from other web sites and give shoppers input on other choices when choosing their own brand they like,” Wyner says. “Our customer is usually driven by their brand. They’re looking to refill their prescription, so to speak. But many see an ad in a magazine and hit the site to search, for example. The MyBuys engine mines the customer and product data and presents other products recommended just for you.”

Collective wisdom

A big question among retailers looking to buy product recommendations systems is which style to choose: one that bases recommendations primarily on the online behavior of an individual or one that bases them on the collective behavior of a retailer’s online shoppers?

The Knot selected Baynote in part because the vendor uses the group approach to recommendations. “Choosing to do cross-sells is a no-brainer, it’s a huge opportunity,” Savilia says. “But only the audience can make total sense of this since things are always changing and trendy. The audience behavior of the past couple of weeks, for instance, always keeps the cross-sells fresh. And recommendations continually increase cross-selling opportunities.”

The behavior of a community of shoppers can indicate purchase intent in ways the past behavior of an individual cannot, contends Scott Brave, chief technology officer at Baynote.

“In one client’s case, when someone clicks on a red washing machine, she is presented with other colorful washing machines, as opposed to just red washing machines or including basic white ones, too,” Brave says. “This is because the community shopping this client’s site shows that people looking at one color of washing machine intend on purchasing a washing machine in some color, not just any washing machine or a washing machine of the same brand.”

When it comes to gaining insights from the group, shoppers at TheKnot.com threw executives for a loop. Recently, when shoppers looking for a groomsmen gift clicked on a flask, they were presented with recommendations for cigar cutters—and bridesmaid gifts. At first this did not seem to make much sense. Digging into the data, Baynote made a discovery about the customers.

“The patterns of behavior of the community showed us it was the same person buying both gifts—the bride was doing the shopping,” Brave says. “That was a connection we hadn’t thought about. The intent of these shoppers was not simply to buy a groomsmen gift, it was to buy gifts for groomsmen and bridesmaids.”

Personal patterns

The other approach to making recommendations relies on the individual’s online behavior. Before a launch, a vendor using this approach may either build a profile based on historical data from an e-retailer or let the system run invisibly in the background for a period of time, gathering information to build profiles.

“We look at everything: what they look at, what they abandon, what they check out with, their response to recommendations we make, whether they take action on abandoned items in their cart, how and where they navigate, what they search, and so on,” says Paul Rosenblum, vice president of products and strategy at MyBuys. “From all of that we derive attributes about the consumer. What is their price point sensitivity, are they a sale buyer, are they a trend follower? It’s a truly one-to-one approach, a unique profile for each and every consumer.”

MyBuys technology also includes a group-oriented algorithm, but it uses it primarily for shoppers new to a site, for whom the system lacks historical information.

“We have found that when you treat people as individuals, they respond better than when you treat them with a groupthink mentality,” says Lisa Joy Rosner, vice president of marketing. “If the name of the game in recommendations is relevancy, you don’t get more relevant than serving the individual exactly what they want rather than taking a guess with what the group might like.”

Then again, why not use both? ChoiceStream, which runs individual- and group-oriented algorithms, believes the best results come when recommendations are based on an individual’s behavior. However, group recommendations can come in handy, says Toffer Winslow, executive vice president of sales and marketing.

“In many cases there is an additional layer of sophistication to be had by filtering the wisdom of the crowd through the prism of a shopper’s unique preferences,” he says. “Let’s say the wisdom of the crowd says a particular pair of brown pants happens to be highly correlated with an orange shirt. But when I look at the shirt, a more sophisticated recommendation engine would know I have never bought or even looked at anything that is orange, but that I do happen to buy yellow things. So it may then recommend that same shirt to me but in yellow.”

There is no one-size-fits-all solution to product recommendations, Winslow adds, which is why ChoiceStream has built a platform that can accommodate different approaches.

While it may one day employ the individual profile method, Borders opted for the group approach and uses the ChoiceStream system to offer recommendations based on the cumulative shopping patterns of millions of shoppers, Winslow says.

“The algorithm is running customer segments to match a customer up with other similar people to give us a wider set of data to draw from,” says Ertell of Borders. “There’s just not enough data for any individual person. As much as we all think we’re individuals, in reality there are tens of thousands of people who are really close to us. By tying people together as cohorts, we’re able to leverage a lot wider data set to provide better recommendations.”

bill@verticalwebmedia.com

When the cookie jar is empty

Product recommendations rely on cookies to tie what a shopper does online with a unique identification number issued by the recommendations system. The system stores data anonymously by the ID number so it can generate individual recommendations or feed a larger set of data to generate group-oriented recommendations.

Many Internet users recognize the benefits of cookies, a recent study by JupiterResearch reveals. 25% of Internet users appreciate the convenience of not having to log in on return visits, 20% value the storing of information to reduce typing on forms, and 20% appreciate the storing of items in shopping carts. What’s more, 14% value the benefit of recommendations.

However, 55% of consumers delete cookies on a regular basis, ranging from daily cleansing to quarterly cleanup, the JupiterResearch study says. An additional 24% delete cookies sporadically, an overall increase in cookie deletion from years past. The most common method for deleting cookies is through browser settings, using default or predefined privacy settings to remove tracking cookies, the study says.

This would seem to present an obstacle for product recommendations vendors trying to build the profiles they use to make recommendations. Not so, the vendors say. They have built in workarounds.

“If your cookie is gone, we don’t care,” says Scott Brave, chief technology officer at Baynote Inc., a product recommendations vendor that provides recommendations based on group behavior. “If you delete your cookie and come back three months later, we’ll still be able to judge current shopping intent by your current online behavior and you’re still getting what the wisdom of the community says is the best recommendation, regardless of whether the site recognizes you.”

When it comes to making recommendations based on individuals’ online behavior, recommendations vendors using this method have included ways to keep an eye out for returning customers.

“When we see a new customer and give them a cookie, but in fact they are a returning customer, we can detect that we do know the customer but under a different ID,” says Paul Rosenblum, vice president of products and strategy at MyBuys Inc., a vendor of site personalization and product recommendations technology. “They will sign up for more alerts or log in to make a purchase—as soon as we get their e-mail address, we connect the old cookie and the new cookie and merge the two profiles together.”

The same goes for site personalization and recommendations vendor ChoiceStream Inc. “A shopper who deleted a cookie would be a new user until they log in to the site,” says Toffer Winslow, executive vice president of sales and marketing. “Our system then sees the old cookie ID number and links that to the previous cookie and merges the records.”