Providing the right semantics along with fact data is one critical User Experience factor and built into the layout design. Fact data i.e. Measures/Metrics are numeric and can describe anything unless the right semantic description is associated with it. A “KPI” is essentially a measure linked with right semantics and context, which also depends on underlying fact types for representation and behavior. KPI’s are typically represented as either a whole number or a ratio and this post delves into finer aspects of the latter.
A KPI that has to be represented as a ratio is typically done through percentages. Apart from ratios, there are several scenarios where a percentage based notation is apt. Any reporting related project will encounter sizable share of KPI’s in percentage format. Some examples of such KPI’s are
- Market share – How is my product (e.g. 32inch LED TV) is performing against all competitors for large screen TV’s segment (e.g. 28/32/36 inch LED & HDTV)
- Sales Mix – How is my product (e.g. Color TV) performing against other products (e.g. Home Entertainment, Gaming)
- Profit Margin – What is the profit on a product i.e. difference of cost and sales over sales.
- Growth – How is my performance compared to a different time period i.e. difference across time periods over older time period.
- Variance – What is the difference of the current value against a benchmark.
One important aspect of these type of KPI’s to keep in mind is that they are normalized calculations against a denominator or one hundred. Percentages are the starting point and other types of KPI’s are dependent on the base percentage metric.
- Percentage: This is the most simplest and easiest to calculate. The resultant value will fall anywhere between zero and hundred and are usually represented as two digit numbers along with ‘%’ symbol. (e.g. 34.53%). Values below and above one hundred are possible in many scenarios and it is up to the KPI definition to set upper and lower boundary conditions.
- Percentile: This is a derived measure that is comes into existence after the underlying data is normalized according to bell curve. (Note: Link to previous post on N-Tiles). These KPI’s can be represented both as a number and percentage based on user requirement. Key item to remember is that, when a percentile is represented as a number, it does not equate to the corresponding percentage. For example, 50th percentile is equal to 50%, whereas 80th percentile could be 85% based on underlying data.
- Percentage Points: A special type of KPI, which is heavily used in economics and finance and used to denote difference between two percentage values. Calculating a percentage of percentages is logically incorrect and hence percentage points shows the difference between two percentages in absolute value. If the baseline is 100 i.e. 4 is equivalent to 4% basis point uses a baseline of 1000 where 100 is equivalent to 1%. These are typically denoted with a suffix as deemed necessary e.g. 4p or 4pp or 4bp etc.
Knowing when to use which notation for a KPI will greatly enhance usability factor of a report or dashboard and drive successful BI adoption in an organization.