Characteristics of the business unit dimension


We introduced the dimensional structure of MI in this blog post earlier in the week.  It turns out that the following dimensions are common to virtually all organisations, irrespective of their size or sector.

Unit: Responsibility centre, such as subsidiary or product group

ItemPerformance indicator, such as sales volume, profit or headcount

ModeDistinction such as actual, forecast or budget

Base: Measurement yardstick, such as local or parent currency

TimeReporting period, such as day, week, month or year

Transformation: Aggregate or average of reporting periods, such as year to date or moving annual total

Understanding the dimensional structure of our data helps us to choose the most appropriate way to present it in graphic or tabular form. A series of blog posts will look at each of these dimensions in turn, and we will start by examining the characteristics of the unit dimension.

A Unit is a responsibility centre, such as a subsidiary or product group. It is a distinct enterprise or part of an enterprise (not necessarily one’s own), characterised by its own set of performance indicators and often with its own manager(s) taking responsibility for its performance.

There are many different types of unit, such as:

Functions: Sales, Production, Research, Finance…

Territories: Chicago, North-East, Europe…

Products: Detergents, Pharmaceuticals, Electronics, Savings…

Brands: Dove, SEAT, Pom-Bears…

Customers: Ford, Tesco, Willis, Mosaic Group A01…

Programmes: High Speed 2, Project Verde…

Competitors: Mazda, Mitsubishi, Unilever…

Economies: EU, US, UK, Japan, ASEAN…

Units are often linked by one or more hierarchies. Corporate structures are typically hierarchies of functions, territories and product groups, with both the elements of the hierarchy and their interconnections changing over time. The ability to be able to recognise and management multiple simultaneous hierarchies of units is often very important, as in the example below pertaining to a supermarkets group.

multiple hierarchy

Each hierarchy provides a particular drill-down path through the data for that particular unit dimension.

Matrix-based organisations require performance to be measured along multiple unit dimensions simultaneously (for example territory by product), as is the case in this heatmap:



In general we can treat units adequately by using generic dimensions in our underling data store that do not  need any particular business knowledge built in to them. However, there are some exceptions:

  • In the case of subsidiary companies, acquisition and disposal dates may be important for analysis.
  • The date of the financial year-end may differ between units (eg. for competitor comparison) so appropriate adjustments to the data would need to be made
  • For projects or programmes treatments of cumulative costs against budget are as relevant on the basis of a full project life-cycle, rather than purely for an accounting year.

Leave a Reply