Still, cheap storage and servers don’t guarantee good information. Indeed, bad data continues to bedevil many corporate data miners, asserts Michael Schrage, co-director of the E-markets initiative at the MIT Media Lab, in Cambridge, Massachusetts. “It’s an article of faith, almost to the point of a joke,” he says, “that every organization I’ve dealt with that wants to data mine looks at the quality of data and realizes it’s not sufficient to do it.”
To clean up the problem, some companies have turned to what’s known as ETL (extract, transform, and load) software. Sold by such vendors as Ascential Software, Ab Initio, and Informatica, ETL programs scour data before it’s routed into warehouses. Some companies use custom programs: Winston-Salem, North Carolina-based Krispy Kreme Doughnuts Inc., for example, uses homegrown data validation and exception applications that act as business-rule black boxes. The programs help keep funky data out of its financial-data warehouse, explains CIO Frank Hood.
Other businesses are attempting to reduce errors by reducing the number of inputs that feed such corporate reports as the general ledger. Earl Shanks, who has been CFO at NCR for two years, says the company used to deal with about 1,200 customized reports generated by the finance and administrative organization. A standardization project has reduced that number to just over 100, he says.
Consolidation projects could prove crucial in meeting regulatory requirements. Obviously, the fewer systems in place, the less data integration required. In addition, the deep-sixing of some programs should reduce the amount of chatter finance managers have to deal with. Case in point: Siegel recalls that before Pfizer deployed its financial-data warehouse, the company’s financial managers had to access 14 distinct systems. “A financial manager does not have the time to be an expert in 14 different systems,” he remarks.
Not surprisingly, some ERP vendors are flogging instance consolidation—that is, the adoption of a single version of a program—as the simplest way to comply with the new regulatory requirements. Since consolidating existing software tends to generate workflow efficiencies, it’s equally unsurprising that customers seem to be listening. A recent survey conducted by AMR Research revealed that 65 percent of publicly traded companies are strongly considering instance consolidation to help them deal with Sarbox (see “Six Degrees of Automation” at the end of the article).
Instance consolidation comes with its own difficulties, though. Cost tops the list. According to AMR, the price tag for an instance consolidation works out to about $10 million per $1 billion in annual revenues. What’s more, instance-consolidation projects can take anywhere from 12 months to two years to complete and often require a full reimplementation of the system.
Can You Demonstrate That?
For the moment, most finance executives aren’t worrying about instance consolidation. They’d be content instead to document exactly how numbers get rolled up into the general ledger.
At Hudson United, Nagel-voort was brought in to head up the bank’s compliance efforts and internal controls. He says the company’s executives are comfortable with the output of the controllers group. But, he adds, “if the CFO and CEO have to put their names on the dotted line…”