The possibilities for predictive analytics are tantalizing. The path for getting there, however, is far from clear. Just getting started can seem paralyzing. But it starts with asking the right business questions.
“What could you do if you could apply analytics across your entire data sphere with no constraints? What would you learn by connecting marketing response data, customer feedback, transactional data, predictive analysis, social sentiment, and product support records? How could you improve inventory management by looking across retail sales, marketing analytics, web analytics, and supply chain data?” asks Hannah Smalltree, Director of Product Marketing, ParAccel in a piece for The Data Warehousing Institute(TDWI) early this year.
The possibilities, for sure, are tantalizing. The path for getting there, however, is far from clear. Just getting started can seem paralyzing. “There is a lot of confusion around analytics, especially predictive analytics,” observes Eugene Roytburg, managing partner at 4i, Inc., a growth and foresight analytics firm best known for industry leading predictive analytics.
Predictive analytics, Roytburg explains, begins with asking the right business questions. For example, say you want to grow 5% next year. Your questions then might be, where will that growth most likely come from and how can you sustain that growth. Most managers try to answer those questions qualitatively, rather than using predictive analytics.
Other questions include:
W hat should be the level of investment to support future growth?
Where will the next innovation come from?
How can I maximize innovation goals and ROI?
How should we maximize innovation success?
What should be optimum strategy to maximize customer and client goals?
Which categories to find more customers and how should we grow them?
Who is our most profitable long term consumer and how we should target them?
How with consumer needs and behaviors change in future and how we should adjust our s trategy to take those into account ?
Predictive analytics as Roytburg describes it also can focus on various geographies, especially in search of growth. Maybe a manager needs to decide which particular emerging countries offer the best target for their particular mix of products. Will Brazil be a better bet than China or Russia?
For CFOs, Roytburg can boil it down even more: “The CFO needs to ask how we should grow and where?” Typically, the CFO goes to Wall St. and promises growth, sometimes even adding a specific numeric goal. Now they need to respond when financial analysts or investors push back. At that point they need to come up with specific numbers and details. How much? Which markets? What products?
The challenge is finding the optimum way to achieve growth. It needs to generate the growth you want in the timeframe you are looking at and be cost efficient to produce the best ROI. For example, you could try to generate more money from existing customers or invest in completely new markets. Each has advantages and disadvantages.
That’s where predictive analytics the way 4i does it comes in. It will enable you to determine where the growth should best come from. Once you determine the where and how for optimal growth and are ready to commit the resources then growth simply becomes a case of how well you execute your strategic plan. Ok, maybe that’s not so simple.
The predictive analytics part, however, is simple; at least once you asked the right questions at the outset. The rest comes down to collecting the necessary data, applying the right analytical models, and correctly interpreting the results. There is no magic here, no crystal ball the way 4i does it. Rather, it’s a question of good data combined with solid mathematical models.
Well, maybe there is a little magic after all. “You start with the understanding of the business problems and drivers, that’s 70% of the magic right there,” says Roytburg. Then you need to understand and identify the 5-6 key performance drivers. From there, the mathematicians model both the individual and joint drivers.
This is not something most organizations would do themselves. There are many organizations like 4i that say they can do this for you, ranging from giant consulting firms to small, mathematical predictive analytics boutiques. The latest trend, however, is for a firm like 4i to go the extra mile and package proven analytics models as a service you can deploy on your own. Are you ready for your own crystal ball?