© 2019 SourceMedia. All rights reserved.

Poor use of reference data management puts information quality at risk

Effective management of reference data such as conversion rates, calendars, country codes and units of measurement is a key part of data management overall. If it’s not done well or not done at all, the result can be poor data quality.

So it makes sense that reference data management (RDM) be a valued component of a company’s approach to using its data resources.

“The biggest piece of advice I can offer is to approach the problem holistically, taking business processes, applications, data and technology into consideration,” said Stewart Bond, director of data integration software at research firm International Data Corp. (IDC).

“Technology can be a solution to RDM, but organizations first need to understand what reference data is, where it is in the organization, and how it is managed within individual applications,” Bond said. “As with any data mastering effort, everyone needs to use the same terminology and vocabulary when discussing reference data.”

Green light illuminates coaxial cables inside a communications room at an office in London, U.K., on Monday, May 15, 2017. Governments and companies around the world began to gain the upper hand against the first wave of an unrivaled global cyberattack, even as the assault was poised to continue claiming victims this week. Photographer: Chris Ratcliffe/Bloomberg

Understanding where the systems of record, systems of entry, and systems of reference are within the context of business processes is critical to understanding how the data flows through and is used in the organization, Bond said.

The key to effective RDM is to have consistency of data across all of the systems that leverage it, Bond said. Several years ago he was in a workshop in which it was discovered that country codes contained within a data set coming into an organization were transformed into long-form country names, only to have a downstream process change the data back to codes.

“The people performing the downstream process never knew that the data had been transformed upstream, and quite often had issues with the reverse transformation because the upstream process was manual and inconsistent,” Bond said.

For reprint and licensing requests for this article, click here.