By Eric Siegel, Ph.D.
Conference Chair, Predictive Analytics World
This article summarizes the wide range of business applications of predictive analytics, each of which predicts a different type of customer behavior in order to automate operational decisions.
A named case study is linked for each of eight pervasive commercial applications of predictive analytics.
Predictive analytics – the concept is dazzling yet daunting. To foresee what each customer will do next may sound like an overwhelmingly complex feat of math and statistics. Is such a thing even possible? And, if so, tell me specifically how it helps my business!
Here’s the bottom line: the predictions don’t actually have to be all that accurate to deliver great value to your business. For example, if predictive analytics identifies a customer segment 3 times more likely to defect than average, you can target a retention offer such as a discount accordingly, and avoid incurring the revenue loss that results from providing the discount to those customers not destined to defect.
This “numbers game” translates into a high return, and doesn’t depend on highly accurate predictions. If your overall defection rate is 5%, the segment we’ve discovered has a higher defection rate of 15%, so, in this case, we’re not actually confident in predicting whether any one individual customer will defect. Rather, the value comes from identifying this kind of customer segment which is predicted in aggregate to behave much differently than your overall customer base.
How It Works
The “magic” of predictive analytics is in how these valuable segments are discovered. The optimal segment is often arcane and involved, the kind of esoteric combination of features you’d expect only a computer to uncover. Core predictive modeling methods do this very well by crunching historical data, thus learning from the collective experience of your organization. To get a concise overview of how this actually works in simple terms, see my DMReview article, Predictive Analytics with Data Mining: How It Works.
Customer Predictions Drive Operational Decisions
Predictive scores are the golden eggs produced by predictive analytics – one predictive score per customer or prospect. Each customer’s score, in turn, informs what action to take with that customer. Business intelligence just doesn’t get more actionable than this kind of decision automation.
Predictive analytics is applied in many ways to help businesses overcome a plethora of challenges. The core difference in one mode of application to another is in what’s being predicted. Predicting customer response, click, or defection are each very different things, and deliver business value in different ways.
Business Applications of Predictive Analytics
Organizations of all sizes apply predictive analytics to automate operational decisions, both online and off-, across marketing, sales and beyond. Which business application of predictive analytics is best for you is a strategic question, and depends on which type of decision you choose to to automate – that is, how predictive scores will best serve to drive decisions within your organization.
The table below lists predictive analytics business applications. The first column names the flavor of business benefit, and the second column tells you the type of customer prediction required – that it, which behavior or action must be predicted to undertake each business application. There are many such applications – this list includes only the most pervasive in commercial deployment to date.
| Business application: | What is predicted: | Read more about it: | Company case study at PAW-09: |
| Customer retention | customer defection/churn/attrition | Published article | Reed Elsevier, and Telenor |
| Direct marketing | customer response | DMReview article, podcast | Charles Schwab |
| Product recommendations | what each customer wants/likes | Netflix Prize leader, HSBC and Amazon.com | |
| Behavior-based advertising | which ad customer will click on | “$1 million” case study and Podcast | Google, Yahoo! and Click Forensics |
| Email targeting | which message customer will respond to | Arthur Hughes | |
| Credit scoring | debtor risk | Wells Fargo | |
| Fundraising for nonprofits | donation amount | NRA | |
| Insurance pricing and selection | applicant response, insured risk | Pinnacol Assurance |
And there are many more applications of predictive analytics, including collections, supply chain optimization, human resource decision support for recruitment and human capital retention,and market research survey analysis.
As you can see, the way predictive scores help your business depends on the customer behavior predicted – just aim predictive analytics towards the right customer prediction goal and fire away. This is why we like to say, if you predict it, you own it!
Learn More: Conference and Resources
Want to learn more? Since there’s no better way to learn about predictive analytics than from concrete case studies, look to Predictive Analytics World, February 18-19, 2009 in San Francisco, the conference packed with case studies from Fortune 500 analytics competitors and other top practitioners. A dozen examples are listed in the rightmost column of the table above, with links to more detail about these sessions at the conference.
For further reading, training options and other resources, see the Predictive Analytics Guide.
About the author: Eric Siegel, Ph.D.
The president of Prediction Impact, Inc., and the program chair for Predictive Analytics World, Eric Siegel is an expert in predictive analytics and data mining and a former computer science professorat Columbia University, where he won the engineering school’s award for teaching, including graduate-level courses in machine learning and intelligent systems – the academic terms for predictive analytics. After Columbia, Dr. Siegel co-founded two software companies for customer profiling and data mining, and then started Prediction Impact in 2003, providing predictive analytics services and training to mid-tier through Fortune 100 companies.
Dr. Siegel is the instructor of the acclaimed training program, Predictive Analytics for Business, Marketing and Web, and the online version, Predictive Analytics Applied. He has published over 20 papers and articles in data mining research and computer science education, has served on 10 conference program committees, and has cochaired an Association for the Advancement of Artificial Intelligence Symposium held at MIT.