Analyticsintermediate

Scorecards & KPIs

Explain how OBIEE Scorecard functionality models KPIs, targets, strategy trees, and initiatives for performance-management reporting.

Scorecards are OBIEE's performance management side: tracking whether the business is meeting defined goals, rather than exploring data.

A scorecard is like a car's dashboard warning-light system: the engine light (a KPI) turns red based on a defined threshold, and a mechanic's diagnostic tree (the strategy tree) lets you trace that single red light back to a specific underlying sensor reading, rather than guessing at what's wrong with the whole car.

Key Concepts

1
A KPI is a defined object, not a report.
2
    actual      the measure
    target      a fixed number, another calculation, or a prior period
    thresholds  green within 5% of target, amber 5-15% below,
                red beyond
    trend       is it improving
    dimensions  the KPI can be evaluated by region, by product
3
So "Revenue" is a measure, and "Revenue against Target" with thresholds and an owner is a KPI.
4
A strategy tree arranges them. Objectives with KPIs beneath them, rolling up:
5
    Grow Profitably
      Increase Revenue          <- KPI: Revenue vs Target   amber
      Improve Margin            <- KPI: Gross Margin %      green
      Retain Customers          <- KPI: Churn Rate          red
6
The colour at the top is derived from the ones below it, with weightings. That is the difference from a dashboard of charts: the roll-up is defined, so "are we on track" has an answer rather than an interpretation.
7
Initiatives are the actions intended to move a KPI, tracked alongside it. The scorecard then links a goal, its measurement, and what is being done about it.
8
Who uses this. Finance, strategy and PMO teams, rather than general analysts. It is a governance tool, and that is what an interviewer is checking exposure to.
9
The honest caveat. Scorecards need agreement before they need configuration.
10
Defining a KPI means agreeing the measure, the target, who owns it, and what amber means. That is a business conversation, and it is where these projects stall — the tooling is the easy part.
11
A scorecard built without that agreement becomes a wall of red that everybody disputes, and it is abandoned.
12
The technical dependency. KPIs sit on analyses, so they inherit the semantic model's definitions and its security. A KPI is only as trustworthy as the subject area beneath it, which is the argument for building them on the governed model rather than on ad hoc data.
13
What the interviewer is probing.1. "What distinguishes a KPI from a measure?" Probing: the concept. Stalls: "Nothing." Moves up: a KPI has a target, thresholds and a trend direction, so it carries a judgement about good and bad, not just a number.
14
2. "What does a scorecard add?" Probing: the structure. Stalls: "A dashboard." Moves up: strategy structure — objectives, perspectives and initiatives with KPIs attached, so performance maps to goals.
15
3. "What is the usual reason a scorecard project fails?" Probing: honesty. Stalls: "The tool." Moves up: no agreed targets or owners — the technology works, but a KPI nobody owns changes nothing.
16
4. "How many KPIs should a scorecard carry?" Probing: judgement. Stalls: "As many as possible." Moves up: few enough that each is acted on; a hundred-KPI scorecard is a data dump with a dashboard on top.