The Human Side of Analytics at #AMDays

April 17th, 2013 by

“Lies, damned lies and statistics” Mark Twain

This session does not focus on technology but rather how the human works.  Instead it focuses on how to be a great analyst versus how to use a new tool. Right away Jim had my attention with his magnetic personality and excitement on the subject.  I knew this was not going to be a boring statistical presentation.

How to be a great analyst you need to master these skills

  • Understand the raw material: The worst thing you can have is really smart people making smart decisions on bad data
  • Understand the tools: You have to understand what all the tools do for you and how they will solve your problems
  • Understand the problem to be solved.
    • We have to have a specific problem that everyone agrees upon and has a specific timeline
    • Everyone has to be onboard from upper management to the implementer.
    • There are only 3 business goals: raise revenue, save time and  customer satisfaction
    • Understand the art of analysis (the wonderful world of puzzles)
      • Look for anomalies (errors and omissions, complaints, spikes & troughs, oddities.  Things that makes you think ”hum that is funny”)
      • Segmentation is key.  If you only take an average of anything is always wrong.
      • Work backwards.  Look at that people who are completing the call to action and see how they got there.  From there you can find out how to find more of them.
      • Microconversions are essential.  It is about getting the users to do one more click.
      • Avoid the traps: if you torture statistics you can get it to say whatever you want it to say like .  Correlation is not causation.
      • At the end you need to have intelligence and intuition.  For the intuition, you need to tone it down..day dream, relax
      • Create a representation of reality through a model. All models are wrong, some models are useful – George Box.  Whatever you figure out will change.  They have a limited lifespan.
      • Keep the model as simple as possible.
  • Understand the art of communication
    • Have an opinion, data to back it and a way to test it. Your opinion is your value!
    • Do not deliver reports, tell stories
    • Help individuals achieve goals.  What’s in it for me.
    • Don’t describe the data and be confident.  It’s about probability and trends.

I have always loved analytis and recomendations.  This session reaffirmed my passion.  Well and there is one more thing I learned.  According to Jim, if I review my data with a glass of wine, I will have better results.  Now that is a new tactic I plan to implement immediately.

Leave a Reply