Introduction post is available “BI Essentials: Benchmarks“.
Organizations have different approaches towards setting benchmarks. The process of finalizing a benchmark, in several cases is a common process, but organizations have their own unique approach based on corporate strategy, operating industry and past experience. Broadly the different types of benchmarks can be classified into three categories as follows
First type of benchmark falls under the category of “self comparison”. This type of benchmark concept is heavily used and can be seen in majority of BI implementations. Here, the measure under scrutiny is evaluated against itself with a common baseline. In BI lingo, this type evaluates the same measure/metric against one or many dimensions. Couple of examples to illustrate this concept are Sales for current year versus previous year or Quantity produced for product A versus product B.
Once the data points are represented in graphical form, the human brain automatically does a context switch. Data consumers (i.e. report users) will automatically start comparing two values and start interpreting one value against the other. Primary effect of benchmarks of any type has on a user is that of comparative analysis, where users are ranking two or more dimensional elements based on a single measure/metric.
If we are adding more dimensions for analysis, only one among all dimensions can have two more more instances e.g. Current Year Apple Sales for USA versus Current Year Apple Sales for Canada. Denominator should be the same and can contain any number of dimensions. Since users are interested in the performance and due to context switch, the same measure acts as benchmark as well as actual values.
NOTE: Self benchmark type is implicit in nature and comes into prominence during actual analysis and interpretation by users. Having cognizance of different benchmark types has more impact in reporting area and helps in communicating the right message to end consumer.