IT pros know how to install and maintain analytical business intelligence systems. Business pros know (or think they know) what they need to do to succeed. A frequent question asked by BI practitioners is, how do you really know if analytics are helping to gauge business success?

The answer lies in the metrics. The problem with designing metrics is that, more often than not, it requires a core expertise that neither IT nor business staff possess. Metrics are about neither technology nor business strategy, per se. The questions that come up around metrics design are almost philosophical: How do you define success? How do you apply quantifiable measures to business processes -- especially qualitative ones like customer service? What kind of data best reflects progress, or the lack of it?

As BPM Partners CEO Craig Schiff says in a recent article we ran on Business Intelligence Pipeline, the analytic dashboards that display metrics are nothing more than tools. Each is "a highly visual interface that includes graphs, color-coded gauges, stoplights, and other cues to highlight variance from performance targets." But more relevant than the tool itself is the information it displays. "What's most important," Shiff says, "is the content: that is, what's being measured and displayed on the dashboard."

He goes on to tell how to design the metrics that will fill those dashboards. The overall business strategy breaks down into objectives, which in turn divide into key business drivers, or the steps that must be executed to meet those objectives. KPIs, or key performance indicators, are the measures that are tied to those drivers. Metrics, finally, are the detailed measures that feed those KPIs. Schiff outlines this top-down approach in his Intelligent Enterprise story, Taking It From The Top. While you're at it, also check out consultant Gary Smith's performance metrics how-to article.

Performance metrics fall into a nebulous area of business intelligence that is neither technology- nor business-centered, but requires input from both IT and business people to come off successfully. (Metrics design is, in this way, similar to tackling data quality problems.) These types of BI challenges are the hardest to overcome, if only because nobody feels qualified to stand up and take the lead on them. But a thoughtful examination of the task at hand, including a componentization of its elements, is at least a good place to start.