I get many questions on this subject and it often turns into almost a religious debate. Let's throw some structure into it. Here's a decision-to-raw-data stack.
- Objectives (e.g., clear understanding of what is driving revenue performance)
- Goals (e.g., achieve x% income growth)
- Calculated metrics (any combination, variation of the standard metrics or KPIs)
- KPIs (e.g., profitability, liquidity, shareholders value)
- KPMs (e.g., enterprise value, trailing/forward price/earnings)
- Metrics (e.g., fee income growth %, non fee income growth %)
- Dimensions (e.g., customers, customer segments, products, time, region)
- Pre-calculated attributes (standard, cross enterprise metrics, KPIs and KPMs)
- Pre-built aggregates (used to speed up reports and queries)
- Analytical data (DW, DM)
- Operational data (ERP, CRM, financials, HR)
Obviously, it's never a clear cut, binary decision, but in my humble opinion
- 1-6 should emphasize business ownership
- 10-14 should emphasize IT ownership
- 7-9 is where it gets murky, and ownership depends on whether metric/KPI/KPM is 1) standard and fixed, 2) fluid and changes frequently, 3) different by product, line of business, region
What did I miss? Thoughts?
Boris also blogs at http://blogs.forrester.com/boris_evelson/.