Data Quality Dashboard

Business problem
Every process fails sometimes, whether from human error or process flaws. We needed a reliable, consistent way to spot problems with critical data about people and accounts. Unlike people, computers never forget to run a report or send an e-mail, and they always do things the same way, so automated audit reports were the best solution.

What gets audited?
Initially, we tried to identify all the places where important processes might fail and then report any failures. Of course, it seems like there's always one more way that any process can fail, and the list of audit reports continues to grow.

Audit reports are e-mailed to the people who are directly responsible for the process and for correcting any defects. Many of the reports are also displayed on a IT data quality dashboard, along with selected statistics. The dashboard currently contains the following measures:

In addition to the measures on the dashboard, there are many other automated audit reports that are e-mailed but do not lend themselves to being displayed on the data quality dashboard. For example, a report such as the "Missing ITID in Meditech provider dictionary" mentioned in Integrating Provider Directories is not listed on the dashboard because it is a list of possible issues that are not necessarily defects — a useful audit tool but not a defect report that needs to be managed to zero defects.

Lessons learned

• Simply e-mailing exception reports does not always achieve the desired results; other priorities may cause the reports to be ignored.
• A dashboard, visible to a wider audience, can help ensure attention to important measures.
• Nothing — not even a dashboard — guarantees that defects will be addressed without consistent management attention.

Posted 23 March 2008