STOCKING DATA

ADAM BLAIR

The most innovative and important warehouse-building boom in the supermarket industry involves not bricks and mortar but information. Early positive results with data warehousing have retailers eager to "stock the shelves" with historical and current sales data, electronic financial transactions and labor information.

Data warehouses' collections of detailed, item-level databases, combined with decision-support tools that help users organize, analyze and act on the stored information, have been most useful in category management efforts, retailers told SN. But most companies are only beginning to explore the potential of data warehouses, and anticipate they will assist with numerous other strategic and tactical applications.

"We're all understanding that data is one of our biggest assets, and if we don't take care of it and use it, we're just passing it away," said Jerry Johnson, vice president of management information systems and chief information officer at Abco Foods, Phoenix.

"We believe [item-level] data captured at the store is the most important thing" to warehouse, he added. "We'll get every bit of information we can, from direct-store-delivery, time and attendance, sales from the register, and use it to help us make better decisions for our stores."

Harding's Friendly Markets, Plainwell, Mich., has only recently started to populate its data warehouse with sales information gathered from each store's point-of-sale. The application, available to each store's manager, provides a 60-week history of item movement, according to Curt DeVries, director of information systems at Harding's.

Even this fairly basic data warehousing application, currently in 20 Harding's stores and scheduled to be available to the remaining 13 by summer, has already helped the retailer in category management decision-making, said DeVries.

"We were going to cancel carrying a private-label chip product, but the data showed it was a consistent sales performer whether it was promoted or not," said DeVries. By contrast, brand-name chips needed promotions to achieve high sales movement.

DeVries believes pressure to discontinue the private-label chip came from the DSD drivers, who did not receive as large a commission on the private-label product as on brand-name products.

"We're just beginning to use the data warehouse at store level. As managers there look at movement over time, they can make a better set," said DeVries. Data warehousing "is like a microwave -- first you use it just to heat coffee, and then you find you can't live without it."

Harry's Farmers Market, Roswell, Ga., is in a building mode with its data warehouse. The independent retailer began its data-warehousing effort last year by assembling two to three years of "detailed item-level history," said Roy Talledo, chief information officer.

After loading sales-history data into its data warehouse, a three-month process completed last September, Harry's began populating the data warehouse with labor information in January, said Talledo.

"One of the big measures retailers typically do is a sales-to-labor ratio," he said. "Now we're writing an inventory system, and we'll push some data back to the data warehouse and be able to do gross-margin analysis."

One reason Talledo has been anxious to add more databases to the data warehouse is the performance improvements he's already seen.

"Sales reports that used to take 20 hours to produce now take two seconds," he noted. "It's been a quantum leap in performance."

But greater speed and functionality are wasted, retailers warn, if the data warehouse's users don't trust the data's accuracy. Setting up a data warehouse involves painstaking comparisons to existing sources of data.

Harry's took three months "going day by day for three years' worth of data, reconciling it against accounting records and bank deposits," said Talledo. "We were building a confidence step to the end users so they would believe the numbers."

In addition, frequent interaction with a company's user community is a prerequisite to data-warehousing success.

Ukrop's Super Markets, Richmond, Va., has used its data warehouse mainly in making its category management efforts more efficient. The retailer is currently attempting to broaden the current data's usage by internal decision-makers.

"Until people start using the data, and figure out how to manipulate it and make decisions, you really haven't done anything but store data," said Konrad Spicer, manager of source data analysis at Ukrop's.

Ukrop's has established a task force that includes a "power user," according to Spicer. This power user was involved in developing the technical side of the data warehouse, and has programming and manipulation tools at his desktop personal computer.

The power user "is a direct interface between the information technology side and the user side," said Spicer. "He interfaces with category managers to make sure the [data-warehousing] tools are fitting the needs they have expressed."

Once users are satisfied as to the accuracy of the data warehouse's data and feel comfortable accessing it, there are a variety of different applications open to them. One of the first areas Harry's Farmers Market examined was market baskets. "For example, if we did a promotion, what other items were bought with the promoted one?" said Talledo.

From this fairly basic analysis, retailers can move to more sophisticated applications. As the number and type of databases in the data warehouse increases, however, information technology executives may start to call on structured data-mining tools to uncover unexpected relationships among data.

"If you have weather data and customer credit card numbers and other relationship elements, a data-mining tool is much more effective," said Talledo. "For example, with ZIP codes of customers, weather data and promotions, you can turn a data-mining tool loose and say 'Find out what's weird.' It can chew for a week and come back and say 'Here's what looks kind of strange.' "

Even for more basic analyses, however, assembling a data warehouse can help a retailer make strategic decisions.

Ukrop's category managers seek a "snapshot of where we are today," for their efforts, said Spicer. "That means assessing what every item contributes to the profitability of the category but, in addition, it means determining exactly what you want that category to contribute to the overall department.

"Is the category a loss leader? Does it contribute to volume? Is it profitable or is it designed to boost sales in another area?" Spicer added. "There are a lot of different reasons to set a category contribution a certain way."

While more advanced data-mining efforts promise future benefits, the most significant current benefit of data warehouses and decision-support tools is their ability to provide information more quickly and easily to numerous users.

People make decisions in real time," Johnson added. If I'm giving you the information two or three weeks later, you're not really thinking as well as you could have."