Think back to the last time you went to the doctor. If something was wrong you needed medicine. Once the doctor figured out what was wrong with you, they wrote a prescription! This prescription is made up of two things – a prediction and a description. The description describes what has been happening – why have you been coughing? The prediction describes what will happen – your virus causing your cough could spread to affect other parts of your body. With these two things in hand the doctor can write his prescription. The prescription tells you what you should do – take this medicine twice daily.
In much the same way prescriptive analytics work for businesses.
If something is or was going wrong for an organization, it’s time to call the doctor! In this case, the doctor would be a data analyst. A data analyst will, in much the same way as our doctor, look to the past via data. Typically, businesses will begin looking for help with recurring issues when they really begin to feel the problem it is causing.
Perhaps inventory levels have been inconsistent, too high one month then too low the next. One month, there’s nowhere to put the stuff. It’s racking up warehousing costs, or perhaps just sitting idly in a corner somewhere. If it is perishable the goods are approaching the point of being unsellable. Regardless, it is a pain for your business.
Even worse is when there is insufficient inventory. Customers are complaining that they cannot make their purchases. Perhaps you have to issue rain checks. The brand is tarnished and starts to be seen as unreliable. Rather than being a one-time problem, this damages the company indefinitely.
Making the Prognosis
The analyst will seek to describe what is happening. This is more than just looking and saying “Yep, there are inventory issues!” The analyst seeks to understand what is happening and more importantly WHY. This can be done in a number of ways, but most likely would begin with observational study using past data. Seeing the problem from different angles within the organization, the analyst can begin the work of descriptive analytics.
The inventory issues are the resulting pains of some other proximate factors the analyst will seek to understand. Maybe inventory shipments are not scheduled when they should be. Perhaps the shipping company is unreliable, having too many accidents or issues of their own. Regardless, the observational study allows the proximate causes to be found and traced back as far as possible, ideally to the ultimate cause. This is the description, what is actually happening.
The other half of the equation the analyst seeks is prediction. What will happen next? Typically we would assume the problem would just continue on indefinitely. But once the ultimate cause is known, we can make better estimates. If the ultimate cause of your inventory issues is that a supplier is delivering inconsistently to your constraint operation. Despite meeting the demand volume, the timing is causing line stoppages that could not be recovered!
Additional Testing Required
This is of course a simple illustration meant to be accessible. When actual data analysts perform their work, they are using a litany of technical mathematical, statistical, and data tools such as regression analysis, ANOVA, and Design of Experiments to determine the cause and effect relationships. Much like a trained physician, the analyst is trained to know what to look for and to determine the appropriate tests and analyses to make a sound prognosis and corrective prescription.
With the descriptive and the predictive side of the equation in place, an analyst can now begin to prescribe corrective actions for the company. This first involves understanding the causation of events, something you wouldn’t be able to do with predictive or descriptive analytics alone. Having united both, we can provide reasons why some future event might happen, good or bad.
Knowing the why of course provides us the best course of action for our desired results. It is important to remember that analytics of any kind – especially prescriptive – cannot provide one hundred percent certainty. They simply give us the best chances of success, and taken in aggregate over many different decisions ought to lead the company where it needs to go.
With this we’ve touched on each of the four aspects of data analytics. It should be apparent that utilizing data is no longer optional, it is required. Companies that don’t embrace this are depriving themselves of necessary insights that both they and their customers will benefit from. Ultimately, a company without data analytics will not be a company for long. Sometimes you have to take the medicine!