Chicago—Computer reseller CDW Corp. began transitioning from a data reporting environment to one of business intelligence (BI) about two years ago. But it wasn’t until about 12 months ago that the push to leverage data in order to drive a competitive advantage began in earnest, said Dan Verdeyen, director of application development for the Vernon Hills, IL-based multichannel merchant. He discussed that push in a Monday morning session at Retail Systems 2006 here.
The transition from simply reporting data to using the information to initiate improvements throughout the company was “difficult,” Verdeyen admitted. One of the greatest challenges involved creating the team. His department had started with 10 employees and currently has 14—but 11 of those staffers joined the team only within the past year. Employees who’d been adept at generating and analyzing reports weren’t always able to develop the skills necessary to take the data and determine how best to act upon them—something Verdeyen had been reluctant to acknowledge.
Because the purpose of BI is to use data to improve processes and performance, “you need to build in a culture of change,” Verdeyen said. That as much as anything is key to success.
“Any system can work,” said Verdeyen, who noted that CDW uses an open-data access and an open architecture in which all systems are integrated into its transactional system, “if you start with, ‘What is our business objective? What data do we need?'”
Among the initiatives implemented as part of the BI model was what Verdeyen called “beacon tracking.” CDW wanted to optimize its online ad expenditures—no insignificant matter given that CDW.com has roughly 100,000 unique visitors a day, is responsible for shipments of 30,000 products each day, and generated $1.8 billion in revenue last year. One way it did this was to assess the revenue generated by the top left corner of its site. If sales of the product featured in that prime position are tracking below average after one week, the company moves quickly to change it.
Another initiative was to track “at risk” customers—those whose expenditures with CDW have been declining over a period of time, for instance. The company used its homegrown CRM application to single out and flag such customers, so that the sales and marketing teams could tailor specific contact strategies to retain them.
Verdeyen offered several suggestions for those looking to move their companies to a BI model:
• “Get to a known state and stay there.” In other works, “make sure that the core of your data is good and continually check it and audit it,” he said. It’s very common and very easy for systems to become unwieldy and absorb bad data if they aren’t consistently monitored.
• Define terms across the organization. How marketing defines a “customer” may not be the same way that IT defines a “customer.” Before proceeding with a project, be sure that all the departments are defining even the most basic terms the same way.
• Try to keep data in 12-month to 15-month chunks. Verdeyen is not an advocate of using customer information that’s older than that. Incorporating data that extends back two years or more makes a project less manageable, he said. What’s more, the return on older data is “questionable.”
• “Stay focused on as short of a chunk [of data] that can bring as much value to the organization as possible.”