Table of Contents
Reporting Vs Analytics
The chessboard has 32 pieces and 64 boxes and given the scale of production of chess boards, it can be assumed that there are quite a few people in the world who would know how the pieces move and how chess is played. However, there are only a few people who can really play chess in a manner that is worth watching.
We are going to wildly extrapolate the same thing to working with data. Yes, there is an enormous amount of data and there is a sizable population of the workforce across different industries who have the word analyst in their designation. So the question is what really differentiates a report from analysis and what differentiates an analyst from a data cruncher/visualizer. This our experience tells us is a challenge that dogs the analytics industry. The ability to write a piece of SQL code makes you an analyst as much as having a telescope makes you an astronomer.
We profusely apologize for the overtly critical and pompous tone of this blog- but this needed to be said. But there are far too many people who are crunching data and not enough people learning from it. Don’t get us wrong but this article is deep-rooted in the experience of working with data & analytics. There is a great anecdote from world war 2 times that tells the story of how data interpretation can change with perspectives and intellect. We are not going to narrate the story here but very highly encourage you to have a read (Abraham Wald & Missing Bullet Holes).
We have seen an enormous amount of human intellect invested in how to visualize data and very little on how to turn that into an analysis. Speaking purely hypothetically let’s assume that we do get to actually analyze data. Even that process is prone to issues that have been classified so well in the book by Morgan Jones called The Thinker’s Toolkit.
Here is what Morgan has to say about some of the most common issues with how we approach an analysis:
- Beginning with a conclusion in mind- this is one of the biggest issues in any analytics project. We usually have a hunch and we fit data to validate that. There is seldom an open-minded exploration of data
- The second issue is also an extension of the first one- this makes us favor one analytical outcome over another based on a hunch and no data backing
- Then comes a cool term called satisficing (yes it’s actually a word)- this also has been classified as neologism- a term coined by Herbert Simon in 1955 that describes a phenomenon in which a manager will settle for a satisfactory solution rather than a solution thrown up by analytical model
This article on the subject from Twink.org is a great read on this subject. That is enough bashing and growling for now. We will meet you next time with probably a lighter subject in a more “satisficing mood”.