For some time now, predictive analytics has been hailed as the next big thing.
A quick Google search shows everyone from Forbes to The Wall Street Journal and beyond have written about how predictive analytics is going to “transform business” and “turn analytics on its head” and even “make BI obsolete”.
However, the analytics market is still dominated by visualization and business intelligence tools like Tableau, Qlik, and Birst.
So if predictive analytics is the next best thing, why isn’t everyone using it? And what’s the solution?
The Value of Prediction
Predictive analytics is useful because it tells you what will happen in the future.
If your business knows what’s going to happen in the future — a potential borrower will default on their loan, tickets for a flight will be sold out, a network will be attacked — you can make decisions that mitigate future risk (decline the risky loan) or leverage future opportunity (increase ticket prices for the in-demand flight).
And current predictive analytics technology is capable of making thousands, even millions, of such predictions each second, giving you the power to optimize nearly limitless business decisions.
It helps you optimize decisions given a predicted future. It helps you react to the future – a future which you’ve accepted.
But what if simply reacting to the future isn’t good enough. What if you need to change the future instead?
Where Prediction Fails
The dirty secret is that, for many business problems and outcomes, predictive analytics alone holds little value.
In many cases, it’s not enough to know what’s going to happen, you need to change what’s going to happen.
And in order to do that, you need more than a simple prediction. You need to know the reasons behind every prediction.
An Example: Customer Churn
Consider the simple case of customer churn.
With predictive analytics, you will know which customers are likely to take their business elsewhere. Knowing this, you could choose to invest fewer resources in those customers, and step up new customer acquisition efforts to offset your predicted losses.
You’ll be more successful and profitable than if you didn’t know who was going to churn. But new customer acquisition can be difficult and costly – much more costly than finding a way to keep those customers who are predicted to churn.
If you don’t know why they are leaving, you won’t know the right actions to take to ensure loyalty.
If, for instance, a customer is leaving because her product keeps failing — but you don’t know that or really anything about why she’s leaving — this story will play out in one of two common scenarios.
One scenario is you guess what might make her stay. Maybe you offer her a longer contract and a free add-on. And you will fail at keeping her because you haven’t addressed her pain point, and you will fail at keeping some substantial percentage of all the other customers who are leaving for greener pastures, and your investment in predictive analytics technology will not have resulted in an acceptable reduction in churn.
Or, you could do some work. You can pick up where your predictive analytics technology left off, digging through the data — manually, frequently via a business intelligence dashboard — to determine what churning customers have in common, why they’re leaving and therefore what interventions might be the most effective. This requires expertise and attention that most businesses can’t afford, and time that customers aren’t going to give you.
Predictions are clearly not enough on their own. So what’s the solution?
Taking Prediction a Step Further
Prescriptive analytics gives you the ‘why’ with every prediction, so you (and your systems) know what to do about each and every future threat or opportunity.
Going back to the churn example, if you knew the reason why your customer was leaving (faulty product) you could intervene in a meaningful way, offering her a replacement product, effectively keeping her as a customer, and successfully reducing churn across your customer base.
No more guessing. No more manual analysis work. And you will unlock the true potential of predictive analytics that everyone is talking about, but that few have actually harnessed.