James Taylor (@jamet123), CEO of Decision Management Solutions, told an audience in Boston, “There are really only four things businesses can use analytics to predict: risk, opportunity, fraud, and demand.” [“The Four Traps of Predictive Analytics,” by Michael Fitzgerald, MIT Sloan Management Review, 25 August 2014] I would add two items to that list: personal preferences and diagnoses. Although some pundits may argue that preferences and diagnoses are sub-categories of opportunity, I believe they are different enough to warrant categories of their own. All six of those categories should be of interest to any business executive who wants a peek into the future. I base some of my observations on work that my company, Enterra Solutions^®, LLC, has done in partnership with Conair, the multi-billion dollar consumer packaged goods (CPG) company, and McCormick & Company, the global leader in packaged herbs and spices. Let me provide a few example of how predictive analytics could change the business landscape forever in each of the six categories mentioned above — beginning with risk.

Risk

Companies face all kinds of risks from supplier failures, natural disasters, shipping delays, climate change, and so forth. It doesn’t take a genius to recognize that if an agricultural area is suffering from drought that the availability and price of food commodities is going to be affected down the road. You might recall back in October 2013 that massive flooding in Thailand disrupted supply chains supporting the electronics, automobile, and other commercial manufacturing sectors. Many companies were caught off guard. That may soon change. Rachelle Flick reports, “Scientists have found a way to anticipate the saturation of water basins by analyzing data taken by NASA’s Grace satellites.” [“Satellite data can predict floods on Earth, researchers say,” The Space Reporter, 7 July 2014] Flick asserts, “Major floods can be accurately predicted 5 months in advance, now that researchers have found a way to measure the amount of water built up in river basins.” What risk manager wouldn’t love to have a five-month warning of a potential supply chain disruption? In our work with Conair, we have been able to help the company predict when orders have potential order fulfillment risks. Predictions are based on a number of factors including inventory on hand, shipping schedules, and production rates.

Opportunity

There has been a lot written about the opportunities that can be created by analyzing big data and making predictions based on that analysis. Perhaps the most written about opportunity involves targeted marketing. Companies are getting so good at predicting what online users are going to do that they are conducting real-time bidding for ad placements in the fraction of a second before a web surfer makes his or her next click. Opportunities don’t have to involve real-time predictive analysis; but, the clock speed of business is certainly accelerating. Another opportunity involved with predictive analytics is the ability to adapt ad content based on past behavior. The likelihood of providing the right offer to the right person at the right time is greatly increased if ads can be adopted based on factors like time and location as well as past history.

Fraud

Each of us benefits daily from predictive analytics in the area of fraud prevention. Financial services companies are always looking for suspicious activity that could indicate fraud is being perpetrated. Spencer E. Ante reports that IBM is working on a new approach that involves “spotting threats before the crown jewels are stolen.” [“IBM’s New Cybersecurity Plan: Find Bad Guys Before They Steal,” Spencer E. Ante, The Wall Street Journal, 5 May 2014] It doesn’t take much imagination to appreciate how much money could be saved if fraud could be predicted before a loss actually occurs. The prospect is staggering.

Demand

Demand has always been a difficult thing for supply chain executives to predict. In fact, demand has been so difficult that supply chain professionals have a name for the turmoil caused when demand forecasts are wrong: the bullwhip effect. As an article in the Financial Times explained, “This frustrating phenomenon occurs when falling customer demand prompts retailers to under-order so as to reduce their inventories. In turn, wholesalers under-order even further to reduce theirs and the effect amplifies up the supply chain until suppliers experience stock-outs – and then over-order in response. The effect can ripple up and down the supply chain many times.” [“Inventories: the bullwhip effect,” Lex, Financial Times, 31 July 2011 (registration required)] One of the great promises of predictive analytics is that it will eliminate or significantly mitigate the bullwhip effect. The goal, of course, is to reduce lost sales and increase profits. Because so many variables affect demand, it will take advanced cognitive computing systems to really master this area.

Personal Preferences

I mentioned targeted marketing in the opportunity discussion; but, Enterra’s work with McCormick and Company opened another category of interactive customer experience that is worth noting. McCormick, of course, wants to increase sales of its products; but, it also wants to make interactions with its customers more enjoyable. To accomplish this, McCormick worked with Enterra to create a personalized way to use big data to increase consumers’ eating enjoyment. The program is called FlavorPrint™. Based on answers to a few questions about flavor presences, participants in the program can get a personalized FlavorPrint that helps McCormick recommend recipes that have a high degree of likelihood the user will enjoy. The FlavorPrint is generated using McCormick’s deep knowledge base of flavors, user responses, and artificial intelligence embedded in the Enterra® Cognitive Computing Platform™. Reactions to the award winning program have been universally positive. FlavorPrint is just one way that companies can use predictive analytics to make interactions with customers more effective. The goal is to enhance interactions not to make customers feel uneasy about them.

Diagnoses

As noted above, diagnoses deserve a predicative analytic category of their own because they don’t neatly fit under the opportunity category. It is well known that IBM’s Watson system is now being used to help doctor’s diagnose cancers and suggest treatment. The system complements rather than supplants a doctor’s medical expertise. Diagnoses, however, don’t have to be limited to the medical field. For example, predictive analytics can be used to diagnose problems in agriculture to help experts recommend corrective courses of action such as which crops to plant and how fertilizer to use. Any problem that involves a large number of variables can benefit from predictive analytics that utilize a cognitive computing system.

Conclusion

If predictive analytics become as effective as I believe they will, the word “forecast” may soon be an anachronism of past business practices. Predictive analytics are not magical solutions that create insights ex nihilo. Taylor insists that companies often make mistakes when trying to implement predicative analytic programs. He suggests “four steps companies need to build the predictive enterprise.” They are:

• Focus on the decisions you want to improve.
• Make analytics as broadly usable as you can.
• Start with models you can scale.
• Land and expand — Taylor’s slang for continuous improvement in the models: use them repeatedly, make them better, and expand models into new areas of the business.

His first step may be the one that is most profound. If you don’t ask the right questions before you conduct analytics, you’ll never get the predictions you desire. Good answers always begin with good questions. Laying the groundwork is important, and, done right, predictive analytics can help solve problems involving everything from food to fraud.