Demystifying Analytics – Part 2 – getting started

Michelle Bidwell
August 22, 2012

Getting started with analytics can be overwhelming – the language off putting, the size of data involved immense, the complexity of the problem staggering. However getting started is simply a matter of following these steps:

  • Define the business purpose
  • Define success
  • Percolate
  • Get data
  • Percolate some more!

The very first step (and it’s often forgotten) is to clearly state what it is that you want to know or do. For example you may want a way to determine if a claim made by a customer is fraudulent or you may want know which customers best respond to an online marketing campaign or you may want to determine the optimal price for a product.
Once the purpose has been defined you then need to articulate what success will look like. For example the number of fraudulent claims will decrease, the response rate to the marketing campaign improves, margin improves & customers stay.
Once you have a clear idea – take some time to percolate. Analytics is often said to be a mixture of art, intuition & science. This is where the art & intuition bits come in. What you are aiming for here is a rough idea/feeling about the sorts of things that could be important, what behaviours would you expect, what are common scenarios and outcomes. Draw from your own experiences/knowledge, chat to others, research what others have done – what works/what doesn’t.
All this thinking leads you to understand what sort of data you will need. At this stage, more is better than less –if it proves not to be useful you can always drop it. What you want is data that is relevant to the outcome, with as much history and granularity as possible. If there is data missing that would have been useful – is there a proxy that could be used in its place?
Percolate some more. …

Copyright © 2019 OptimalBI LTD.