CFOs are keenly aware that while the popularity of the term “predictive analytics” is relatively new, the concept is really nothing revolutionary. Predictive analytics is often described as data mining or advanced analytics that, as a concept, has existed for a long time. The only difference now is the abundance of technologies, tools and applications in the marketplace to analyze the large amount of data available.

This proliferation has pushed some finance executives to immediately utilize the tools without first understanding their available data and devising a proper strategy that allows the organization to truly benefit from predictive analytics.

Capgemini recently spoke with a CFO who said it best: “I can’t predict my organization’s financial future without knowing and truly understanding the past. 

Predictive analytics, at its core, means understanding the key relationships between business KPIs and their predictive variables, given past events. The real benefit lies in effectively leveraging that information to predict future outcomes or behavior. It’s all about identifying patterns in the past data, melding them with current data points and then using both to predict customer behaviors, wants and needs that ultimately lead to improved performance in key business domains. It can be applied across many industries; however, several basics must be in place before you can effectively employ predictive analytics.   

Availability of relevant data: Simply having large volumes of data, without effectively capturing enough history, will not provide enough measurement points to detect a rate or change of rate. You must have a sound data management strategy in place and have sufficient data, over time, to establish data behavior patterns. With a large data set, information must captured efficiently and effectively and also be spread out over enough time to establish logical patterns. Technology is an important factor here.

One example is CRM solutions, which offer a strong commercial application of predictive analysis. Accurate data is an organization’s most valuable asset; without insight into an organization’s historical data and into customer experiences, you can’t offer actionable predictions for specific customer accounts.

Data quality: An assessment of current and historical data quality is also very important when determining the data’s potential to make predictions and support decision-making. While most predictive modelers will employ some data cleaning, poor data quality can lead to a substantial increase in effort. While all organizations recognize that poor data leads to poor BI, the effect of poor data quality gets magnified in predictions.

Defining predictive analytics goals: It is important to define predictive analytics goals in terms of something quantifiable and measurable. While “improving customer insights” is a laudable goal, the predictive analytics target should be something like “increase (by cross-selling, up-selling) the number of customers who own two or more of our products by X percent, thereby increasing profitability by Y percent.”

Measuring the ROI: It is critical to measure and baseline the current metrics and predict additional gains around the predictive analytics initiative in order to determine the ROI. For example, if a retailer wanted to carry out a predictive analytics initiative around increasing the forecasting accuracy of demand, a mechanism should be in place that accurately captures the stock outs so the predicted benefit gained serves as the basis for ROI prior to the initiative getting off the ground. After launching and deploying new forecast models, the effect on stock outs needs to be captured and documented to either justify or change the predictive analytics initiative.

As mentioned earlier, predictive analytics is not a new concept. Before a new initiative is launched, it’s prudent to survey the current environment and see if any previous work done by other departments within the organization can be leveraged. For example, marketing departments also compute customer exposure as part of customer lifetime value calculations and so do credit risk departments. While both departments may do it for different purposes, there are synergies within the process that could be leveraged for both initiatives.

CFOs can help their companies profit from predictive analytics, as long as they ensure that the analytics are based on enough historical data to determine meaningful patterns, and analysts properly model and analyze it for decision-making support. This approach helps turn poor business decisions, made with hazardous guesswork, into informed and successful business decisions that improve performance.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access