Gutting the StoreBy Brian P. Watson | Posted 2007-05-14 Email Print
Expanding its product lines helped Dick's Sporting Goods grow from a mid-market player to one of the country's premier athletic retailers. But first it had to become more nimble in the ways it managed and tracked its offerings.GUTTING THE STORE
Business intelligence software wasn't around at the retailer's humble beginnings, when Dick Stack, at the spry young age of 18, left his job at an upstate New York Army/Navy store in 1948 to open his own bait-and-tackle shop. According to company lore, Stack's former boss scoffed when the young employee suggested expanding the store's product line to include fishing supplies, telling the teen he'd "never make a good merchant."
Turns out the boss underestimated the entrepreneur. Nearly 60 years later, Stack's company now publicly traded and led by his son, Edward offers everything from cricket bats to yoga mats across the U.S. The chain now boasts 268 stores in 34 states, with $3.11 billion in revenue in the 2006 fiscal year.
But as Dick's grew, it needed a game plan for analyzing how it inventoried and sold its various product lines.
Back in 2003, the company had some basic reporting tools included in a merchandising system from STS, a specialty vendor acquired in December 2000 by the U.K.-based NSB Group, a competitor in the retail software market.
While acknowledging that the STS system is not primarily a reporting system, Schroer says it took too long to create reports. Gathering data to run a report required several steps, he says. To start, an employee had to go to a computer connected to a Microsoft Access database, where financial reports and statistics were stored. Once there, the employee had to define the specific data he wanted for instance, how batting gloves were selling in the stores.
Then the system sent that data through a Unix server to the employee's personal computer, where it had to be downloaded into an Access database or Microsoft Excel spreadsheet and checked for accuracy. If the employee needed another data set, he had to start all over again.
The entire process, on average, took up to an hour, Schroer says. But for more complex analysis for instance, incorporating separate data sets, like how those batting gloves were selling in off-season months, or what regions sold more that could stretch several times over. Beyond that, though, the quality of the data in those reports was questionable. Since there was no warehouse storing all the company's statistics, there was no telling whether the data sources used to create a report were accurate, Mewherter says.
Dick's had deployed some business intelligence tools, including Cognos Business Intelligence Series 7, as well as Crystal Reports (a product of Business Objects) and Brio (now owned by Hyperion, which was bought in April by Oracle). But users never embraced the software.
Mewherter says it was an internal issue namely, the absence of a training program and not technological blips that forced out Cognos, Crystal Reports and Brio. "We had tools that were being poorly used because we hadn't trained people well," he says. That led employees to think the Cognos tools didn't work.
Part of the problem was that users weren't included in mapping out how Dick's would use the software to mine data and create reports, Schroer says. The dismissal by users meant going back to the drawing board. "Instead of taking what we had with Cognos, which in their mind was a failed package, we decided to go through the whole process again," he says.
That effort began with the data warehouse. Under Mewherter, Dick's built the warehouse on an Oracle 8i database, with customized capabilities to extract data, transform it to meet business requirements and load it into the warehouse. (They've since upgraded to 9i and were moving to 10g at press time.)