Taking Notice of Data Quality: As DQ Discipline Goes Enterprise-Wide, Even the "C Suite" Is Getting Involved

There was a time when data quality was thought of as an information technology or clerical problem--if it was thought of at all. Yet, to the IT guy, a particular data element is defined not as a unit of meaning, but a cluster of fields that are either filled or not. To the teller, data is the stuff of a transaction and meant to be a speedy exercise, because customers are waiting. And your self-service customer who lets his fingers do the walking? He probably has his own agenda and the bank's data quality isn't it. Meanwhile, to the rest of the organization, for whom data refers to living, breathing customers and actual products, the sheer volume of the records in question can be terrifying. How much of it is accurate? Where might the mistakes be coming from? How can they be fixed? And can we still sell to and service customers in the process ? While no one would publicly admit to accepting a database riddled with errors (the kind that come from typos or poorly executed data extractions), the reality is errors exist and tend to be worked around. "Traditionally, inaccuracies or messiness around records were thought of as a cost of doing business and the problem was brushed aside," says Ashok Vermuri, senior vice-president, banking and capital markets unit North America, Infosys Technologies, Bangalore, India. Infosys counts data quality among its business process and IT capabilities. The problem has probably always been underestimated, Vermuri explains. Where a typical institution may have assessed its inaccuracies in databases or warehouses in the range of 5%, the actual figure is probably closer to 15%. You can call me Ray, or you can ... Most of us have gotten a telemarketing call where it's clear that the person is reading from the wrong script. Frank Dravis, vice-president of information quality for First Logic, La Crosse, Wisc., gets it all--Dravies, Davis, Dave--so he can empathize. "Data quality absolutely impacts sales and marketing initiatives," he says. "You can annoy and offend people if you aren't in a position to acknowledge the relationship." Yet mistakes most often occur not from stupidity but from scale. Consider the sheer number of records involved and information's changing nature (so that today's accuracy is tomorrow's error). Consider, too, the reality that as banks grow by merging they have ample legacy technology to blend. Add to those factors, the outsourcing of projects, reorganization of departments, changing personnel, and the reorganization of IT itself over the lifecycle of a typical company, and you have messy data in the making. These errors can range from calling a woman Mr., to misspelling a name, to using an incorrect or outdated business title, to bungling a Social Security number or something equally as problematic. But now the issue of information management is getting a second look and, increasingly, senior-level executives (CEOs, CFOs, CIOs--the "C suite") are becoming involved in the project work designed to clean up the company's informational act. "More recently, DQ is beginning to be thought of as an enterprise issue requiring a formal strategy," confirms Vermuri. "I have a CEO who includes a data quality report along with his other items to the board," seconds Katie Fabiszak, vice-president of marketing with DataFlux, an SAS company based in Cary, N.C. "The bank also has a data monitoring system and formal method for record clean-up. You just didn't see that as recently as three years ago." Six Sigma and clean data Don Carlson, senior vice-president, compliance project manager, Bank of America, was hired three years ago to take what had been a less formal approach to cleaning up data to higher--and more systematic--ground. BofA now uses Six Sigma as a backbone for a rigorous program with the following edicts: focus and finish; link and leverage (line of business responsibility with that of a centralized group of data stewards); rigorous, statistical measurement on error and improvement; and ROI measurement of project value. …