ULTRA MART EMPLOYS INTERNETWORKING PROTOCOL FOR DATA MINING

DEENA AMATO-McCOY

MENOMONEE FALLS, Wis. -- Ultra Mart here began using internetworking protocols last week that allow the retailer to analyze valuable point-of-sale data in hourly increments vs. waiting for daily reports.

A new wide area network allowed Ultra Mart to begin testing Transmission Control Protocol/Internet Protocol in its Neenah, Wis., store Oct. 6, said Mark Makofski, desktop computing manager for Ultra Mart.

The installation of the wide area network -- dedicated digital telephone lines that will funnel individual store data directly to the retailer's data warehouse within minutes -- was completed earlier this month.

The retailer also plans to use TCP/IP to examine customer purchase information gathered via its frequent-shopper program. Ultra Mart plans to probe buying patterns of its frequent shoppers, captured through a customer relationship marketing application that it began testing in its Neenah store Oct. 1.

"By switching to TCP/IP, we will be able to read data files live off our front-end systems in order to summarize our sales in hourly increments rather than reading daily reports," said Makofski.

Ultra Mart uses its data-mining tool to query information stored on its data warehouse, which resides at the retailer's headquarters. Makofski said the data-mining tool analyzes sales and item movement based on information collected at its seven stores' POS systems.

"The data-mining tool is used to summarize many individual transactions," he explained. "The server at our headquarters uses file-transfer protocol to pull transaction files off each store's in-house processor.

"The downside to using FTP is that you are connected to a flat file -- nothing is live and all information is stored. We need to wait until stores close, pull files in their entirety, then do the report mining," he added. "By using TCP/IP, all information is live so we can mine from items taken directly from store processors each hour."

Ultra Mart ultimately plans to bring all seven stores on-line with TCP/IP, with the second store going live by the end of the first quarter of 1998, said Makofski.

Like the data-mining information, the new relationship marketing software will allow information collected at the POS to be transmitted to Ultra Mart's data warehouse in hourly increments.

"We have been tracking which customers are purchasing what products, and in which categories, through our frequent-shopper program," he explained.

"We will query the data to analyze what discounts are being redeemed in which categories," said Makofski. "In turn, we can do better targeting and promoting of customer loyalty, rather than just promoting discounts."

By using the data-mining tool, Ultra Mart will also be able to improve its financial reporting based on sales and movement information.

The retailer is already focusing on the next area it will enhance through its data-mining tool. "Our next project is to work on cashier productivity," Makofski explained. "This will allow us to query information that will let us investigate scanning vs. idle time cashiers spend at registers. It will also let us examine cashier accuracy, and track voids and overrides." The retailer will begin testing this during the first quarter of 1998.

The data-mining tool, DataMiner, is provided by MGV-America, Raleigh, N.C. The customer-relationship marketing software is provided by S2 Systems, Dallas.