Analytics that is not Prescriptive, is a Hobby

There, I have said it. We all know it instinctively, but sometimes it just needs to be said. If data analytics is not able to guide your organization in making decisions and taking actions, then it is not delivering value. Yes, I know that there have been attempts to lay out Analytics along the continuum of Descriptive, Predictive, Prescriptive with additional layers thrown in for good measure - and these are still valuable from a practitioner/learning perspective - but when it comes to delivering value to your organization - all Analytics must be Prescriptive. 

And yes, human intelligence has a role to play and machine learning has a role to play. Indeed, every form of analytics must use every type of available resource until the output is prescriptive - guiding decisions and actions, hopefully in a fruitful direction.

Let us examine this in some more detail:

• Business Intelligence & Prescriptive Guidance - the original and still foundational Analytics capability, now pushing 25+ years. It tells you what is happening in your business right now. It also tells you what has been happening in the last year, or last 5 years. The delivery of Business Intelligence usually requires:

      a) Reporting - Useful information, automatically updated, easily accessible

       b) Analysis - a human Analyst that develops prescriptive steps to drive decisions

      Over time, BI tools include some of the more common forms of analysis that Analysts undertake, allowing them to reach their prescriptive decisions faster. 

• Predictive Analytics & Prescriptive Guidance - includes forecasts, likelihood probabilities and more - and these add the “looking ahead” flavor of Analytics. Now, we all predict the future all the time (“it will take me 25 minutes to drive home from work this evening”, “our web traffic will drop in December owing to seasonality”, “H&R Block Tax preparation’s business in the US will peak in the first 4 months of the year”). And as human beings, we are expert pattern recognizers and can extrapolate these patterns reasonably well. However, the real power of predictive analytics emerges when these patterns are not clear, and the factors influencing these outcomes is not clear either - and predictive analytics and related modeling can help us understand these. But, even with predictive analytics, we need additional human intelligence to look at the forecasts and the factors involved in making the final set of prescriptive steps to guide decisions. 
• Prescriptive Analytics & Prescriptive Guidance - frequently the domain of optimization and traditional Operations Research - this form of analysis yields very specific recommendations under well-defined constraints. Coming from the world of manufacturing, this form of analytics is very useful and very precise under those exacting conditions. In other spaces, where is there far greater flexibility, the recommendations from this analysis must be seen under the light of the organization’s ability to make significant changes to its operating model. Thus a measure of human intelligence must be brought to bear in these cases as well before a final set of prescriptive decisions can be made.

As you may infer, what does happen as we increase an Analytics organization’s capability and maturity is that the underlying Analytics can provide increasingly growing support in making those final, prescriptive, decisions. However, irrespective of level of maturity, every form of Analytics must be prescriptive before it is usable - otherwise, it remains an interesting hobby!