The new eCRM systems keep tabs on customer’s purchases, wherever they occur
By Peter Lucas
As recently as 12 months ago, software that recommended products based on a customer’s shopping habits at a web site was considered exciting. Imagine keeping track of a customer’s buying habits electronically and using that information to pop up product recommendations the next time the customer came to the site.
Today—old news. As with just about every other business practice related to selling on the web, retailing’s embrace of multi-channel selling is washing away the old—meaning last year’s leading-edge processes.
Now retailers want to know what their customers are buying no matter how they shop—via the web, the telephone or in person—and vendors are responding with more sophisticated software that allows a retailer’s operating system and customer service reps constant access to customer behavior across all channels.
A customer that one week purchases cooking utensils online may view fine crystal during the next visit, then spend 10 minutes questioning a call center representative for information about the manufacturer and how the price and quality compare to other brands.
But since the first generation of personalization applications lack the ability to track and store the data from off-line customer interactions, they deliver only a partial picture of the customer’s shopping preferences. That can lead to product recommendations that entirely miss the mark, especially if sales agents working off-line channels make the recommendations.
“Customers who regularly patronize a retailer want to be recognized and treated with the same level of attention regardless of which channel they use to shop,” says Gerri Spieler, research director for Gartner Group Inc. of Stamford, Conn. “Personalization applications enable retailers to gather a lot of customer data, but it is how that data is used across all touch points that creates a repeat customer. Internet retailers do not make money on one-time customers.”
To address this problem, vendors of personalization software are touting next generation applications they claim are capable of gathering and sharing data across all sales channels.
The pot of gold at the end of this rainbow is to enable representatives in any sales channel to access a customer profile, then use that data to quickly close a sale by recommending products the customer is likely to buy or to create a cross-sell or up-sell opportunity.
Closing the sale
“A high percentage of consumers contact a call center need assistance in making a purchasing decision,” explains Craig DeWolf, vice president of marketing for QuickDog Inc., San Francisco-based personalization vendor. “Making a customer profile available to call center agents makes it easier to recommend products likely to appeal to the customer that can close the sale.”
It sounds great—and if it works, it will be great. But developers of such systems have a number of obstacles to overcome, not the least of which is the unproven nature of the concept. “The Internet retail industry is a long way from pulling customer information from all sales channels into a single data warehouse,” says Peter Hunt, president and CEO of CornerHardware.com, which is working with Brightware Inc. to develop a personalization system. “That may work from an engineering standpoint, but we don’t know with certainty that it works as advertised.”
But that uncertainty isn’t stopping vendors from developing more sophisticated systems or retailers from being willing to test them.
QuickDog, for instance, has teamed with Lands’ End to provide the back-end scoring algorithms for Lands’ End Personal Shopper application, which the Dodgeville, Wis.-based retailer launched last November. Unlike the previous generation of personalization applications that generated customer profiles for use only at the retailer’s web site, the new generation creates a 360-degree profile using data gathered from each online and off-line interaction with the customer. The information is stored in a central database and sales agents working in any channel can access and update the data.
Decisions vs. rules
As a result, when an in-store sales representative or call center agent services a frequent online shopper, the representative can use an Internet-capable personal digital assistant or personal computer to quickly call up the customer’s profile.
In a reverse scenario, brick-and-mortar retailers that have built customer profiles using purchase histories and customer responses to direct mail campaigns can push pop-up windows featuring product recommendations to their core customers as they venture online. The only hitch there is that the customer must first identify herself by entering a password logged when registering to use the site.
“There is a real need for consistent personalization platforms to drive a singular personalization effort,” says Donna Boyer, director of product marketing for San Francisco-based Personify Inc. “As this happens, customer data can be used to make better marketing and merchandising decisions, rather than just automating business rules.”
In addition to QuickDog and Personify, personalization vendors include Blue Martini Software, BroadVision Inc., E.piphany Inc. and Net Perceptions Inc. Personalization applications typically start at $100,000 and can cost more than $1 million. Most customer relationship management software vendors offer a personalization module, whether it is analytics based or a search engine capable of recommending products based on click stream patterns of the current shopping session.
With the drive to integrate personalization applications across all sales channels shifting into a higher gear, several Internet retailers are undertaking their own initiatives to create better personalization applications. Only later do they turn to CRM vendors to fine-tune the package. “A lot of personalization programs look elegant, but they tend to be cookie cutter applications that lack the features we need,” declares Richard Viard, senior vice president of research and development for SmarterKids.com, which developed a personalization application. “Most of the personalization applications need to be customized so why not build an application you know will fit your needs and is expandable.”
Sales boost
Lands’ End has been one of the more aggressive Internet retailers at attempting to create a homegrown personalization application that extends across its entire enterprise.
In January, Lands’ End licensed its front-end technology to QuickDog in exchange for a minority stake in the company. QuickDog plans to target the offering to Internet retailers in such markets as home furnishings, consumer electronics and toys. QuickDog’s personalization application can be licensed for $250,000 plus installation or leased for $100,000 annually.
The front-end technology developed by Lands’ End uses conjoint analysis, a mathematical process that assigns values to customer preferences and weighs them against the type of merchandise for which they are shopping, such as cost vs. features and functionality.
In the case of Land’s End, customer preferences, such as style, color and fabric, are compared against apparel features in the product category for which the customer is shopping. Product recommendations are based on the weight of each customer preference vs. the attributes of the product.
Lands’ End gathers customer preferences through an optional survey that asks customers to state their likes and dislikes about as many as six outfits. It pulls together additional information from purchase history and questions asked during each shopping session, such as the customer’s objective and whether the customer is shopping for a gift or for herself.
The program then builds rules around customer preferences so as to eliminate products with features that do not appeal to her preferences. Customers can create multiple profiles, such as for active wear or business casual. Each time a customer initiates a shopping experience online, the customer selects the profile she wishes to use during that session. Each profile is available to sales agents who accept orders for the Lands’ End catalog department.
The new software delivered a significant sales boost during the Christmas shopping season. Browse-to-buy ratios for consumers using Personal Shopper were 80% better than those who did not use the application. And customers who used Personal Shopper bought 30% more merchandise than non-users, according to QuickDog.
“If a profile does not exist, the sales agent can call up the customer’s purchase history, ask a couple of questions to eliminate anomalies and then use the scoring models to identify product most likely to interest the customer,” adds DeWolf. “The idea is to make it easier to shop in a subjective environment, just as an in-store sales representative does.”
Hard work
Although other Internet retailers, such as SmarterKids.com and CornerHardware.com, have built the framework for their own personalization applications, they have been reluctant to roll the applications out across all their sales channels. The reason is that the jury is still out on whether the technology exists to effectively access customer profiles across multiple points of contact.
“Unifying customer data gathered across separate channels in a single warehouse and correctly tying that data to the right individual is an arduous process,” says David Provost, senior analyst for Waltham, Mass.-based Gomez Inc. “This concept is new thinking for online retailing and the technology behind it is still too young to tell if it is hitting the mark.”
Although no hard figures exist as to how many Internet retailers are testing the new generation of personalization software, anecdotal evidence suggests that the technology’s greenness is a major stumbling block.
SmarterKids’ Viard argues that attempting a systemwide implementation without proof that it will work as advertised can be a costly mistake.
SmarterKids has rigorously tested its personalization program using control groups to measure and improve performance since launching its site in November 1998. Initial test results revealed that the application was most effective when used as part of the site’s My Kids section where customers can shop for an individual child or a group of children, rather than in all sections. Other sections include the Family Resource Center and the Gift center.
Customers can set up a personal profile when they register at the site. The initial profile covers whether they typically shop for a child or group of children, such as a classroom, the learning style of the child or group and their educational goals. The program asks additional questions during subsequent visits to broaden the profile. A shopper must make several purchases before the site begins suggesting products.
To enhance the program’s capabilities, SmarterKids.com tapped Palo Alto, Calif.-based Responsys.com to provide backend personalization capabilities for e-mail campaigns and Quadstone, a Boston-based vendor of predictive marketing software, to supply data analysis tools.
“No two people shop the same way and trying to personalize the entire store is not easy,” says Viard. “We found it more effective to have one section of the store personalized.”
So far that business rule has paid off handsomely. The average ticket for customers who set up a personal profile is 13% higher than for those who do not. Seventy five percent of customers who set up a profile make a purchase and customers who set up a profile are 44% more likely to make a repeat purchase than those who do not.
Patience
Achieving those results has required the patience to track customer preferences and behavior over several visits before making a product recommendation. “It takes time to analyze data and recommend products based on the learning style and goals of the child,” he adds. “Customers want product suggestions they know will be of benefit, otherwise they wonder why the suggestion was made.”
CornerHardware also has developed the framework for its own personalization application without attempting to immediately extend it across its entire enterprise. The San Francisco-based Internet retailer has teamed with San Rafael-based Brightware to develop its Tool Advisor, which helps customers find the correct tool for their project based on the criteria they specify.
Rather than recommend only the most popular tool in a category, CornerHardware’s Tool Advisor will recommend other tools of the same type in that category. Pop-up windows containing information about the product category, such as an article comparing features of various products or a buyers guide, also appear.
“You just can’t assume you know what customers want and what they are thinking,” argues Hunt, CornerHardware’s president.
Because doubts exist about the ability to integrate customer data across all sales channels, some Internet retailers such as Hunt prefer not to be pioneers out of concern that potential implementation and operating problems could harm their business.
To mitigate such qualms, QuickDog is including a champion-challenger module that demonstrates the effectiveness of its application across the enterprise. Others, such as San Mateo, Calif.-based E.piphany, are promoting the ability of the application to check the warehouse to determine if a product is in stock before it is recommended. “To maximize the value of the application, online retailers need to see that it can unify data for all customer interactions across all aspects of their business,” says Mark Kanok, senior product marketing manager for E.piphany.
Until personalization vendors can unequivocally demonstrate their ability to consistently deliver on that promise, most Internet retailers are content to sit tight and wait. l
Peter Lucas is a Chicago-based freelance business writer.
|
