LAS VEGAS -- Think that standardizing your corporate data in an enterprise data warehouse is all you need to do to create a successful business intelligence
Many companies have worked for years and spent millions of dollars to build big, monolithic data warehouses, only to find now that doing so was just the beginning of the BI development cycle. They also are learning that they may have to give up some of their centralized data controls to enable end users to take full advantage of BI tools.
“A lot of times, BI is relegated to a data warehouse with ancillary reports,” Baseline Consulting’s Jill Dyche said in a keynote speech at the BI Executive Summit, held here as part of last week’s TDWI World Conference. And indeed, when Dyche asked for a show of hands on where attendees stood with their BI programs, many indicated that their BI teams were seen primarily as report writers.
Dyche, a partner and co-founder of Baseline in Sherman Oaks, Calif., recommended that IT and business units work together to develop a structured BI strategy, first by agreeing on a set of guiding principles for their BI deployments and then by using a portfolio management approach to prioritizing the rollout of different BI applications for separate groups of users. “[That way] it’s no longer about the tool set,” she said. “It’s about BI’s ability to support corporate strategies and objectives.”
Creating a prioritized BI application portfolio also gets companies out of the practice of loading data into their data warehouses in an organizational vacuum, Dyche said. “The ideal,” she noted, “is to pair up applications with the specific data they need.”
Mark Madsen, a consultant at Third Nature Inc. in Rogue River, Ore., went even further during a pair of keynotes at the BI Executive Summit and the general conference. Madsen advocated establishing different “zones of control” over data within an organization, saying that the focus on standardizing information in data warehouses and associated data marts is “disempowering” business users and leading to the perpetuation of “the kingdom of Excel” as an alternative to corporate BI tools.
Fixed data models may cause business intelligence problems
“Data warehouse has become a bad word for a lot of people,” Madsen said. “The concept of fixed [data] models needs to change. Data doesn’t have to be identical; you have to relax the constraints.” Otherwise, he added in an interview, data warehouses risk becoming “legacy systems cranking out static reports” that don’t let organizations tap the full value of BI technologies and processes.
Jason Beard, director of global BI at John Wiley & Sons Inc., said Madsen’s advice maps to where he’d like to take the Hoboken, N.J.-based publishing company’s BI program. “We want to take our BI game up a little,” Beard said. “A lot of what we’re asked to do now is kind of routine stuff that answers questions but doesn’t go into this unstructured analysis realm.”
Generating reports out of Wiley’s data warehouse will continue to be an important part of the company’s BI strategy, Beard said, noting that many of the users in the company’s finance department just want to get data “in a repeatable and consistent way” for tracking key performance indicators. But he added that he’s looking to use BI tools to enable, say, marketing workers to do data analysis in a “more speed-of-thought” manner.
Madsen’s description of centralized data warehouses as often being too rigid “is exactly our story,” said Suresh Karu’Pakula, an IT principal and director of BI strategy and architecture at ConAgra Foods Inc. in Omaha, Neb. ConAgra has spent five years building and operating a single data warehouse, Karu’Pakula said. But now, he added, the BI team is looking to “go from being the purveyor of a complete BI stack to being the purveyor of enterprise information and to let go of some of the control.”
Data analysis needs to be faster, more iterative and more flexible than it is now, Karu’Pakula said. And in order to achieve that, he wants ConAgra to be in a position where it can provide “the right tool for the right user group,” instead of forcing them all to rely on the same BI technology.
Wayne Eckerson, director of research and services at conference sponsor TDWI, said the issue of control over data is becoming a “pain point” for mature data warehousing organizations.
“We build these things up over time and start to standardize the data and the BI platform, and then IT or the BI team starts to control stuff for very valid reasons,” Eckerson said. “But after a while, that becomes the problem. And if we’re not careful, we start to affect the ability of users to get the useful information that was the whole reason to do BI in the first place.”