The analytic insights function is valuable in providing data-driven recommendations and information. On many occasions in my current role, we have used data analysis to identify trends, issues, root causes, or to dispel a ‘gut feeling’. Unfortunately, in large fortune 500 organizations with complex data infrastructures, many teams use many sources to come up with hypothesis. Our goal in these situations is to avoid re analyzing or disproving a prior analysis. However, this still happens.

What business and other teams often don’t realize are the tens + decisions that go into a thorough, thoughtful analysis. I personally performed these types of analysis in the pharmacy division of my current employer, so I have first-hand experience into these decisions. Our insights teams asks probing, detailed questions while defining and shaping the analysis. To me, this is where we truly differentiate ourselves from casual analysis. We also have a strong foundation in statistical concepts. What might some of these tens of decisions include? You want a report or analysis by provider specialty, but we know the data quality of that data is less reliable. We might recommend another way at getting at the same information. Have you thought about which provider identifier to use? Not all pieces of information are equal. These are just a few simple examples that can drastically change the value and accuracy of an analysis.

Although there may be questions from time to time as to whether or not an analytics team ‘should do analysis’, having subject matter experts who are extremely knowledgeable of the data, the data sources, and the right probing questions to ask makes all the difference between an analysis and a deep, thoughtful, accurate analysis.

A book on a slightly different, but related topic is ‘How to Lie with Statistics’ and ‘Statistics Done Wrong.’ I know this post wasn’t primarily about statistics, but these readings show how intentional or unintentional decisions lead to erroneous results.