For companies to benefit from mining to find the business insight gems in big data, they certainly need data scientists and analysts, but it is also vital for managers to be well-versed in predictive analytics to bolster the bottom line.
Many managers might be reluctant to delve into the world of predictive analytics because of the perceived “quantitative wizardry” it takes to cull through vast amounts of data to predict customer behavior, market shifts, or other factors to gain a competitive advantage.
But, many managers already are immersed in forms of predictive analytics without knowing it, noted Tom Davenport, in a recent Harvard Business Review blog post. Davenport is a professor at Babson College and a research fellow at the MIT Center for Digital Business.
Companies that have developed a customer lifetime value measure, for example, are using predictive analytics to determine how much a customer may buy in the future, he notes.
Furthermore, making a “next best offer” to a customer or using digital models to determine where to place advertising are all forms of predictive analytics.
So, what do managers really need to leverage predictive analytics to make better decisions? According to Davenport, managers need to understand the basics of three areas: data, statistics, and assumptions.
A lack of good data is the most common challenge for organizations seeking to leverage predictive analytics, because firms need data on what customers are buying, what they have bought, details of those products, and demographic attributes of customers.
“If you have multiple channels or customer touch points, you need to make sure that they capture data on customer purchases in the same way your previous channels did,” according to Davenport.
Next, it is important for managers to understand the statistics used for predictive analytics.
Regressions analytics – the most common tool used for predictive analytics – works when an analyst takes a set of variables like gender, income, visits to a web site, etc. and analyzes how correlated each variable is.
When an analyst succeeds in finding variables that explain a product purchase, this information can be used to create a score predicting the likelihood of a purchase.
“You have created a predictive model for other customers who weren’t in the sample,” Davenport notes. “All you have to do is compute their score, and offer the product to them if their score exceeds a certain level. It’s quite likely that the high scoring customers will want to buy the product – assuming the analyst did the statistical work well and that the data were of good quality.”
Finally, managers need to understand the role of assumptions in predictive analytics, Davenport notes. The biggest assumption in predictive analytics is that the future will continue to be like the past. However, this assumption can be invalid.
“If your model was created several years ago, it may no longer accurately predict current behavior. The greater the elapsed time, the more likely customer behavior has changed.”
After managers grasp the basics of predictive analytics, there are a series of questions to ask analysts to ensure that predictive analytics efforts are successful:
- What is the data source for the analytics?
- Is the sample data representative of the population?
- Are there outliers in data distribution? How will these affect the results?
- What are the assumptions behind the analysis and are there conditions that might make the assumptions invalid?