For Jerry Storch, vice chairman of Target Corp., Minneapolis, the advantage of sophisticated technology like data mining and knowledge discovery comes down to a very homespun, low-tech image: small-town America.
"You walk down Main Street and the merchant knows who you are," he said at the Grocery Manufacturers of America's annual executive conference at The Greenbrier in White Sulphur Springs, W.Va. "You walk into a shoe store and the merchant says, 'Oh, we've got these new pumps in that are perfect for you. You forgot your checkbook? Don't worry -- I'll send you a bill, or the next time you're in you can pay.'
"That's where we want to get to," added Storch, whose company operates 991 Target stores and 37 SuperTarget stores, as well as the Marshall Field's and Mervyn's chains. "We know you that well, so we can deal with you as an individual. We can underwrite your financial risk based upon who you are."
Technology has gotten to the point where such individualized marketing -- moving from a product/location focus to an individual focus -- can be accomplished, Storch said. "And we've invested heavily and aggressively to make that possible."
Target is not alone in this quest. A growing number of retailers are developing or deploying data-mining systems that take the raw data accumulated at the point of sale, especially loyalty card data that identifies the shopper, and shine a high-tech light on it in order to glean marketing insights about that customer, including what type of product she may be interested in or what sort of credit risk she may be.
In SN's eighth annual SN State of the Industry Report on Supermarket Technology, published in last week's issue, 29% of survey respondents said they would test or launch a data-mining program in 2002, up from 20% who said they did so last year. Data mining ranked first this year and last among programs respondents said they were testing or launching.
Companies from Stop & Shop, Quincy, Mass., to Remke Markets, Covington, Ky., are leveraging data-mined insights by mailing targeted coupons and other offers to their best and most loyal shoppers. The problem with this approach, of course, is the high cost of printing and mailing, so companies are now looking at ways to use technology at the checkout, at kiosks, on the phone and online, and to make their selective, targeted offers more cost-efficient.
Target, for example, has invested in technology from Compaq (now part of Hewlett-Packard, Palo Alto, Calif.) called Zero Latency Enterprise. ZLE, which Compaq acquired through its acquisition of Tandem Computers in 1997, is designed to enable a retailer to instantaneously access information throughout the enterprise and immediately apply that information to customer marketing, according to HP. ZLE, therefore, eliminates the "information float" or lag between the time data is collected and the time it is used.
Using ZLE, "I can know who you are and what you bought from me in the past and generate instantaneously the right offer for you at the POS," said Storch. "Most databases are repositories, but those are useless in a retail environment" because they make consumers wait for offers. "That's why we had to have zero latency."
Storch did not say to what extent Target is employing this technology, and Target declined to respond to further questions on his remarks at the GMA conference last month. But he implied that Target would use it extensively. He noted that the company currently engages in 1.8 billion contacts with its customers, or "guests," as Target calls them, each year -- 1 billion in stores, and the rest through its Web site, Target.com, its call centers and via direct mail.
"Every one of those contacts is an opportunity to apply this database to sell to guests or communicate with them," he said. "Right now, we're throwing away those opportunities. But we want to do targeted offers every time we talk to someone. The response rates go through the roof once you understand who you're talking to. The potential is vast."
Storch said Target also plans to leverage some of the highly developed personalization technology employed by Amazon, with whom Target formed a partnership last September. In addition to launching a Target store on the Amazon Web site, Target has been using Amazon's e-commerce technology services, order fulfillment and customer care services for its online properties. "Amazon is the best at personalization," said Storch.
Target also intends to more extensively use smart card technology in support of its personalization goals, Storch said. Since launching its Smart Visa card program a year ago, Target has become the second-largest issuer of smart cards in the world, behind only American Express' Blue card, according to Storch. "We should pass them by the end of this year," he said, predicting that 6 million smart Target cards would be issued by then.
Target began installing POS terminals that accept smart card chip payment this year. The chips have 64 kilobytes of memory. Storch said Target will have kiosks in every store this year that will read and write to the smart card chip. Consumers will be able to download offers to the chip off their PCs or the kiosks and redeem them at the POS. "This is true electronic couponing tied to our database without the waste of paper coupons," said Storch. "You no longer get coupons you don't need."
Storch said the data-mining capability of Target's systems will look at a shopper who buys diapers and present an offer for other kids' products. The system will also identify people who like clearance sales and inform them about such sales, so that discounts can be kept within limits. That, said Storch, will help contain markdown costs, Target's second-largest expense after labor.
On a much smaller scale, Remke Markets, which operates seven supermarkets, has used data-mining and knowledge discovery tools to market to their loyalty customers, who generate 85% of sales.
Remke's data-mining tool, Allegiance, from Triversity, Toronto, has been used to offer inducements from the chain and its suppliers to change shopping behavior, said Pat Iasillo, Remke's director of consumer relationship marketing. For example, low- or non-deli department shoppers have been offered $1 off a deli purchase of $5 or more. These offers commonly redeem at 5% to 10%, he said.
In the past year, Remke has tested a knowledge discovery tool, Promo Coach, also from Triversity. While data mining extracts useful knowledge from large collections of data, knowledge discovery tools go one step further and glean hidden patterns within large data sets, according to Peter Wolf, vice president, business development, for Triversity.
Last year, Remke conducted a series of tests that compared the impact of promotions based on data mining with the impact of promotions based on knowledge discovery. The chain found that the latter generated an average coupon-redemption rate of 14.2%, compared with 4.5% for the former. Sales for knowledge discovery promotions were 22.5% higher than those for data-mining promotions.
In one category, gel snacks, Remke used data mining to promote a particular brand to non-users of the brand who were purchasers of a different brand of gel snack; with knowledge discovery, Remke promoted to shoppers who didn't buy any gel snacks. Despite expectations, during a five-week promotion period the redemption rate for the knowledge discovery group was 43.13%, compared with a rate of 5.55% for the data-mining group.
What Iasillo especially likes about the impact of knowledge discovery is that it expands a category, rather than just switch shoppers from one brand to another with no jump in category sales. "It's awesome, spectacular," he said.
Data mining "doesn't necessarily get good results," Iasillo said. But knowledge discovery, by comparing the behavior of shoppers who don't shop a category with those who do, can pick out non-shoppers with a high potential to be persuaded to buy from the category, he said.
Iasillo is also working on ways to lower the cost of reaching shoppers with targeted offers. Currently using a monthly newsletter mailed to shoppers, he is developing alternative delivery vehicles that include the POS, kiosks and the Web. "That's where you'll really see the returns," he said.
He is focusing initially on updating his POS system, which he said would be done in "a year or less." Besides cost savings, the advantage of the new system will be the higher frequency of communications at the checkout with shoppers who come into the store one to three times per week. "When you send people something every four to six weeks, it's hard to get them turned on to the program," he said. "But if I can keep reminding them with offers, then we'll really be rocking and rolling."
Another company that markets data-mining software purporting to boost market-basket size rather than simply switch brands is Sagarmatha, based in Kfar Malal, Israel, with a U.S. office in Marietta, Ga. The company has boosted the average shopping basket of loyalty card shoppers at 94-store drug store chain Super-Pharm Israel, Herzliya, Israel, by 17% over the past year, said Will Phillips, president of U.S. marketing for Sagarmatha. Sagarmatha has also worked with Israeli grocery store chain Super Sol, and plans a pilot at a U.S. supermarket chain in September, Phillips said.
Sagarmatha's automated system, called Personal Promotion Builder, uses sophisticated technology whose origins are in biotechnology (see story, below) to detect shopping voids. These are categories that shoppers aren't purchasing in a store despite evidence that they should be doing so, which could mean they are getting the products elsewhere, explained Phillips.
Super-Pharm sends out around 12 personalized offers to loyalty shoppers each month, some designed as "loyalty builders" aimed at current purchases, and others designed as "sales builders" that shoppers aren't currently buying, said Phillips. The system, offered on an application service provider (ASP) basis, can also target shoppers at the POS, at a kiosk or via the Web.
Marv Imus, president of one-store Paw Paw Shopping Center, Paw Paw, Mich., uses a data-mining tool to target recent and frequent purchasers of a category for additional offers, getting a redemption rate of between 3% and 5%. "I'm more for brand enhancement than switching based on price," Imus said. "Wal-Mart's got that [price] game."
Imus has a robust Microsoft SQL database with seven years of history on shoppers in his loyalty program, who now represent 80% of his clientele. He said he is starting to group his shoppers by "life stages" -- young married couples vs. those with kids in college, for example -- so that he can make offers based on the particular life stage a shopper has reached.
Paw Paw is currently in a beta test of a new knowledge discovery tool from AdPilot, New York, that uses a "predictive response algorithm." Imus said the system is "50% better" at predicting which consumers will respond to an offer. "It gives us the ability to manage our promotional funding better," he added, noting that the system can determine the rate of return for different price points. Paw Paw never had this ability in the past, he said. "We're real excited about it."
The AdPilot system, which Imus said will be in "a more workable version" by year's end, can suggest pricing for different strategies, such as building traffic or increasing profitability. It provides information "we can really use" rather than "hoping we are doing it correctly," he said. He expects to use it as a shelf management tool as well.