Predictive Analytics has become the poster child of BigData. Today, there is hardly an industry or vertical that hasn’t seen some amount of innovation infused by predictive analytics. Among verticals, Marketing for example uses predictive analytics for learning customer preferences and propensities, Finance uses it for forecasting and optimizing cost and spend, Sales leverages predictions in demand and revenue predictions.
Most product offerings today aim to add a dash of smartness by making them predictive. However, users of applications have grown smarter being pampered by intelligent interfaces already, that being predictive has become the norm and is no longer the differentiator.
Data Scientists need to start planning for the next iteration of their data products to delight your already intelligent customers. In the next few years, we believe that predictive models need to also be combined with the following features –
- Contextual: Predictive apps will need to understand the context of the user and his overall journey with the data product. (For example, suggesting a FAQ page, when the user hits a roadblock with a feature)
- Cognizant: Such apps need to be cognizant of the user’s resources and constraints. E.g. playing a long podcast in the morning right before a meeting vs. in the evening during a jog
- Anticipatory: We are already seeing the rise of anticipatory predictions, be it the GoogleNow alerts, or Amazon’s brave foray into anticipatory shipping.
- Augmentative: Finally, my favorite is when predictive and smart apps can build themselves into the natural workflow of humans and augment their capabilities. E.g The predictive type-ahead feature on Google, is a simple yet powerful predictive model that augments a user’s capability
When predictive modeling is done right, you get happy customers and smug data scientists !