Are the results of predictive analytics really that surprising?

Change

It’s common to hear claims about how businesses will be transformed through the use of predictive analytics techniques with surprising and shock results that they never would have imagined to be the case. It’s particularly common for product or service vendors to make these kinds of claims when in pursuit of a possible sale. After all, who could resist the prospect of a magic analytics bullet that promises to completely transform their organization?

I completely agree that the business benefits of analytics can be significant, but I think it’s a myth that predictive analytics will always produce surprising results that utterly revolutionize your business. More often the results of predictive analytics yield steady improvement to an already successful organization. Predictive analytics contributes important incremental changes rather than revolutionary back to basics ones. 

At Smart Vision we work with organisations at every stage of their journey with analytics. The process always starts with a deep look at the requirements from a business standpoint. Using the CRISP-DM methodology is a great help in these conversations and discovery phases. Knowing what the business problems are and helping organisations uncover specific challenges on which to focus is really the key to starting a successful project. It’s important to spend this time really getting to know the business at the start of the process, and this discovery phase can be valuable for the client as well as for us. If you spend some time really thinking about the problems that your business faces and about the data you’d need in order to start solving them, then the ultimate results of the analytics phase of the project really shouldn’t come as a complete surprise.

The point is this: if you’re prepared to look closely at your business’s challenges and take an open-minded view of what it does today, how you currently measure business performance and how well you’re achieving what you want to achieve, then you should get a pretty clear view of where analytics will make a difference.  In a well-orchestrated project the client ought to know ahead of time what kind of a difference predictive analytics will make before the modeling work even begins.

It’s generally not the case that predictive analytics produces some revelation that completely revolutionizes the business from the ground up. Instead, predictive analytics easily gives the talented business manager a small advantage with each customer, on each project, each year. Over time these small advantages turn into a substantial competitive edge. For example, a catalogue or online retailer that can better target its mailing list can increase profits by reducing the volume of expensive mailings while increasing the number of orders. Over time, this can result in a substantially more profitable business.

The associations found in your data may be a surprise but the impact of using these carefully to increase profits should not be.