Quality Matters: Seven TipsQuality Matters: Seven Tips

Learn about your third-party or affiliated data sources. At OneSource Information Services Inc., which works with more than 25 information suppliers, the data stewards learn where suppliers regularly make mistakes or lack information, and they ask for changes. "Some of them are good about improving, some we just drop," says Bill Schumacher, senior VP of content.

Clean your databases before starting a data warehouse, data mart, or other database-integration project. Before running multiple passes of a data-cleansing tool, try to understand what assumptions were made when old databases were built.

Structure databases across a company so that fields for customers, vendors, etc., are consistent and reference the same table of choices or use the same data ranges.

Assign a person in each business area to maintain data quality. This data steward might not own or enter the data but should be familiar enough with the data and the business the data supports to understand how clean things need to be kept.

Do your maintenance. "A manager might spend three-quarters of a million dollars to get the company's data cleaned up, then two years later say, 'Why isn't it fixed?'" says Steve Brown, executive director of product management at Ascential Software Corp. "This isn't [a set-it-and-forget-it] problem."

Human intervention is almost always necessary to assure near-flawless data. At FleetBoston Financial, when database updates are done, changes are reviewed by account executives. "If we want our 2,000 agents to use our applications, they have to be flexible and easy to use," says Jim Eardley, managing director of business development and strategy. "But we also need to proactively alert the users to potential duplicates and control the data they enter."