The APQC Blog

Business Case for Building a Data-Driven Culture

I recently interviewed Eric Siegel, founder of Predictive Analytics World and author of Predictive Analytics: The Power to Predict Who Will Click, Buy, Lie, or Die, about how to build a data-driven organization.

Eric will be a keynote speaker at Predictive Analytics World for Business in Chicago, June 20-23, 2016. The conference is the leading cross-vendor event for predictive analytics professionals, managers and commercial practitioners, focused on delivering on the promise of data science. PAW Business covers a wide range of business applications across industry sectors, including marketing, credit scoring, insurance, fraud detection, optimization, and more. Join Eric Siegel and APQC when you register today with 15% off code APQC15.

Q: What compels organizations to switch to predictive-analytics-based decision making?

 Siegel: There are millions of process decisions that all organizations make every day. The way to use data to drive and optimize those decisions is through predictive analytics. This technology informs decision makers of the probability of specific outcomes. It guides decisions makers as to whom to target for marketing campaigns, what content to send, who to approve for applications for credit, when to conduct audits, and so on.

At a fundamental level, predictive analytics ascribes a probability to an outcome. Decision makers have to decide what their risk threshold is (usually quantified as a specific probability value) when using predictive analytics. For example, is a five percent likelihood that sending a customer a brochure will result in a sale high enough, vis-à-vis the marketing cost of doing so?

Q. Do you have any advice on points someone should include if they are making a business case to move to using predictive analytics?

Siegel: A strong business case for implementing predictive analytics should include the following three elements:

  1. Value—an estimation of the impact on the bottom line or on other key performance indicators that will be achieved.

  2. Cost—all anticipated resource requirements.  This empowers decision makers to balance the resource investment against the potential value.

  3. Risk—for example, the risk that the predicted model will not perform as well as expected. Include information on how you can mitigate the identified risks, such as deploying the predictive models in a regimented way (e.g., only using it for 10 percent of decisions at first).  

Q. What are the common pitfalls you see most organizations face when they move to a data-driven decision making culture and how can organizations overcome these pitfalls?

Siegel: The most common pitfall is not effectively defining the deployment objective in the first place. You can spend a great deal of effort conducting robust analysis, but it will all be wasted if you don’t have an agreed upon plan for how the output will be used. This requires a commitment across the organization to change the way decisions are made so that predictive scores are made use of. Predictive analytics projects are usually more likely to fail due to an organizational failure to do this than because of analytical or technological shortcomings.

Check out the full interview with Siegel and learn more about analytics with the following APQC resources.