Key Performance Indicators for Information Quality
Part 2 — How to Measure
Daragh O Brien
In the July 2009 issue of the IAIDQ newsletter, we showed that information quality metrics must be aligned with key strategic goals, otherwise we are in danger of measuring the wrong things and creating incentives for the wrong behavior.
Assuming that we have determined what should be measured as an information quality KPI and we understand why we are going to measure it, the next challenge is to figure out how to measure it. In this context, I am not talking about the detailed technical issues involved in profiling datasets either using hand-coded SQL checks or one of the increasingly excellent data profiling tools that are available. Rather I will be discussing some of the practicalities involved in determining how to turn raw metrics data into KPI information.
There are a few factors that you need to have in mind when considering the How of Information Quality KPIs:
- Not all the information required for your checking will or can live in a database. Some of it may only exist in “hard copy” paper forms or in the form of real people or things
- The purpose of a KPI is to be an indicator of how things are going, not a forensically exhaustive insight of why they got that way. However, the process of collating those indicators can (and should) provide base data for 'drill down' insights on the 'why' of the 'how'
- On their own, KPIs are not predictive of future performance, rather they are evidence of performance in some prior period. However, statistical techniques can use KPI trends to help suggest possible future problems.