When a KPI values do not reflect expectations, you shall determine what went wrong. There are two ways to do that: cause-effect and drill down.

Cause-effect cascading is performed by associating KPIs in a cause-effect chain (or network). This way whenever a KPI presents a bad value we can check related KPIs to identify what went wrong. If margin is low, for example, this can be due to excessive competition that we can measure with another measure like price / average competitors price.

Cause-effect cascading can employ KPI formula as a starting point or consider constraining factors, like previous examples but always involves different measures. The other, non-exclusive alternative is hierarchical drill down.

Organizations are hierarchical in nature, or, more precisely multi-hierarchical, a retail chain, for example, is usually organized around two intersecting hierarchies: product and geographic location. Unless your organization is extremely small, you will usually require a hierarchy of same KPI measuring similar aspects but at different levels of different hierarchies.

Let's make an example with a classic KPI: net margin. Suppose that you realize that net margin is well under expectations (say 12% instead of planned 22%). Your next analytic step would probably consider the same KPI (net margin) for different regions of your retail chain, and, once identified the offending regions you will drill down to cities inside those regions and then to stores inside cities.

After exploring the geographic hierarchy you may realize that there is no better or worse store, so you start analyzing the product hierarchy: product categories and products under those categories or different suppliers.

Hierarchies can also been combined and you can possibly identify that certain products are poorly performing only in certain stores.

There are certain advantages in using a hierarchical approach when compared to cause-effect (in practice both approaches shall be used):

1. Responsibility reflects your organization hierarchy, so information becomes more relevant and specific.

2. Low-level measures can be combined and rolled-up into higher level measures providing perfect traceability.

3. No extra KPIs are needed.

Which is the best tool to visually perform such analysis? In practice I found treemaps best visual alternative for this kind of analysis because they show in a single shot:

1. The hierarchy (or combination of hierarchies)

2. The importance or impact of each hierarchy unit. In our case rectangle sizes can be proportional to amount sold.

3. KPI of each unit represented by rectangle color.

To allow hierarchical drill down you only need plan data collection at required detail level. If you define your KPIs with KPIStudio (a freely available tool), you will identify those hierarchies in the classification (dimensions) column.