It’s hard to open up the Wall Street Journal these days without running across an article on artificial intelligence, big data or predictive analytics. All of these terms are interrelated, and part of what could generally be called data science today. Here is the reality though – despite all the hype, data science is still in its infancy and offers unique opportunities for companies that want to capitalize on it. So far, most firms have not figured out the right way to use data with their business yet.
Tools like predictive analytics – what I call Moneyball for Business – have a lot to offer the business world: price optimization, better decision making around investments, improved AP/AR functions, more effective marketing, and more. Yet most firms aren’t ready to put predictive analytics into practice, and predictive analytics is really just the most basic entree into data science – big data and AI come later.
So why can’t business get predictive analytics right? Two reasons stand out.
First, it’s easy to talk about big picture ways to use predictive analytics in theory, but actually getting data together into a tangible, usable format is hard. Even in the investments world, where data is much more available than the corporate world, data collection is still hard. For instance, on a recent project in litigation finance, I was asked to use predictive analytics to forecast duration of individual claims in a large portfolio of consumer litigation claims. Doing that effectively required putting together a lot of data that the firm did not have – not to mention dealing with the fact that more than 90% of cases are settled prior to trial.
If that sounds time-consuming, it’s because it is. In other words, predictive analytics is hard because putting together good data is hard, but the rewards for doing it right are substantial.
The second major hurdle firms face in predictive analytics is not being able to separate the noise from the signal. In a recent project I did for a major consumer products company, the goal was to use data to determine which groups of consumers the company should send coupons to. The key was to send coupons to those customers who are most likely to use them, but who also would not have bought from the consumer products company without the coupon. If you send a coupon to a customer who would have bought the product anyway, it needlessly lowers the average selling price (ASP) and wastes marketing dollars.
To efficiently solve that problem for the consumer goods company, just having good data is not enough. I needed to forecast who would buy and who would not, based on the customer characteristics. Now truly doing that prediction with 100% accuracy is impossible. Instead, we have to think in terms of probabilities – who is likely to buy or unlikely to buy. Once you start thinking in those terms, you can start looking for ways to separate the noise from the signal and build an accurate model of purchase propensity.
The key here is to not just dive right into crunching a number as soon as you have a basic dataset. Instead, really consider how to structure that data in a meaningful way, and what you are trying to learn from the data. This point was elegantly made in a recent Harvard Business Review article by Eric Siegel.
Predictive analytics is really just the start of the data science path; ultimately, whether you’re looking at giants like Amazon and Cisco, mid-sized firms like software provider Factset, or tiny firms like litigation finance provider Town Center Partners, the leaders in data science in their fields have a clear advantage. They can offer better pricing to clients because they have a better grasp of the risks, costs and opportunities in business.
Putting data analytics into practice is not easy, but our TRI Corp. team can help get your professionals up to speed by providing customized executive education to your company’s needs and integrating TRI’s custom business simulations with data analytics tools such as Tableau, Alteryx, Crystal Ball, and others. The best way to get your team up to speed is practicing in a safe environment where they can learn through experience.
