The CIO’s role in IT project portfolio governance is well established. The 2004 book IT Governance, by Peter Weill and Jeanne Ross, captured the benefits of effective IT governance: higher return on assets, investments tied to strategic priorities, greater agility and reduced duplication. The principles of IT portfolio governance served as a road map for CIOs in taming what was often the Wild West of IT spending and decision-making in organizations where investments were not necessarily aligned with business strategies and often led to inflated expenses.
We now face a new organizational Wild West, brought on by the promise of big data and increasingly sophisticated predictive analytics. Data projects are sprouting up throughout the organization. Multiple business units in every level of the company are developing a growing appetite for using analytics to answer such questions as: What happened? Why did it happen? What will happen? How can we make it happen?
Companies like eBay have predictive analytics experiments running all the time, with every customer involved in at least one. American Express mines structured and unstructured behavioral data to predict responsiveness to a set of offers that are quickly executed.
Multiple uncoordinated experiments and campaigns can yield impressive learning, agility and profit. But they can also lead to unintended consequences. Knight Trading lost $440 million by launching the wrong version of its software when testing algorithms. Target identified a teenager’s pregnancy through data analytics and sent her coupons for baby supplies–before she had told her parents, prompting an irate visit from the girl’s father.
According to researcher Michael Goul of Arizona State University, CIOs should play a key role in aligning business and IT when deploying predictive analytics campaigns–taming this new Wild West. In a recent presentation to the Society for Information Management’s Advanced Practices Council, Goul discussed the various roles that the IT group can handle, such as establishing security practices, assessing risk, designing data-management policies, encouraging collaboration between data scientists and business leaders, and looking for opportunities to turn predictive analytics into new businesses or improve existing ones.
Goul also described a six-step process for deploying predictive analytics campaigns that are coordinated, tracked, measured and quickly turned into action:
Design: Determine the duration and pace of the campaign, and the strategy for performance measurement.
Embed: Develop and test user interfaces and ensure that results appear at the right time and for the right people.
Empower: Train employees to interpret and use the model results as intended. If the content is highly dynamic, ensure that model results refresh fast enough to keep up.
Measure performance: Make results available to campaign managers in real time. Establish dashboards for monitoring progress. Send managers alerts if campaigns stray from intended paths.
Evaluate: Study the effectiveness or progress of a campaign, addressing such questions as: Are error levels acceptable? Were campaign results worth the investment in the predictive analytics system?
Re-target: Reflect on the results and decide what changes are needed for future modeling efforts.
CIOs who take an early leadership role in analytics governance will further demonstrate their own value and help invent their organizations’ future.
Madeline Weiss is director of the Society for Information Management’s Advanced Practices Council (APC). June Drewry is a former CIO of Chubb and an adviser to the APC.