Which KPIs can be created on a data source?

There are often very valuable data sources which are not exploited at their full potential because of a poor KPIs... or I would say lack of imagination when thinking about KPIs.

Over years, we created some kind of deterministic approach to define almost an exhaustive list of the KPIs which could be defined. Out of this list, it is then your own appreciation to keep only the meaningful ones :-)

This approach works on any data source, for any business.


Let's imagine such a simple data source on Sales data:


How many KPIs can we create out of it? 3? 6?... much more actually... let's start!


0. Focus on what really matters

First, start to analyze all your the attributes of you data source and figure out which information is meaningful or not (based on your business knowledge, the quality of data...). Hide all non-meaningful information.


1. Define 'Basic' KPIs

The idea here is to build KPIs based on the raw information.

Enrich your data source with raw information

First, see how you can enrich your data with additional information based on the available data. Generally speaking, this could be done by creating calculations on source attributes. In our example, we can create an attribute 'Amount' being defined as 'Unit Price' * 'Quantity'. With the same approach, we can define 'Margin'. 


Aggregate your attributes

Now, list all your attributes (inluding the news ones we've been adding) and aggregate them.

For instance, we can aggregate attribute 'Quantity' by applying a SUM all data lines which would create a KPI we could name "Total Sold Quantity".

But not only this, we can aggregate attribute 'Customer'. Really? Yes, counting the number of Customers across your data source could be a very valuable information to look at. This would create an additional KPI "Number of active Customers" which would be defined as Count Distinct Of ('Customer').

Applying this rule for all attributes, you can get a first set of KPIs.


and this is the way you define basic KPIs in LinPack :-)





2. Combine your KPIs!

Congrats, you have reached the first step :-)

Now, the second and last step of your journey is about combining all those KPIs together.

For instance, you can compute the "Margin %" as "Total Margin" Divided By "Turnover". With the same approach, "Average Selling Price" would be defined as "Turnover" / "Total Quantity".

This approach is recursive, meaning that you can of course create a combined KPI from two other combined KPIs.

By using this method, here is a list of combined KPIs you COULD create:


And this is the way you define combined KPis in LinPack :-)



3. Only keep meaningful KPIs!

Such an approach has to be considered to get inspired in your KPIs research. The whole list should not be considered and created as it is. So, sort them out and focus only on the top meaningful KPIS for your business :-)



And when you create a dashboard with LinPack, it is the way to define as many KPIs as you need !