Analytics success starts with a center of excellence

Data-driven organizations accelerate data transformation with an analytics COE.

idc center of excellence analytics numbers data
Getty Images

Artificial intelligence and data science offer great promise for delivering business value across the enterprise. IT organizations and lines of business units alike are rushing to make good on this potential, but AI and data science initiatives can’t be rushed. Yes, there are instances in which an individual with some expertise in data science grabs available data and surprises business leaders with answers to questions they hadn’t thought to ask. But more often than not, this approach leads to misleading recommendations and wasted time.

According to a recent IDC survey of CIOs and senior IT executives, 93 percent of IT executives in the U.S. say their enterprise is leveraging some form of center of excellence (COE) to drive AI and data science initiatives. In many, if not most, cases, these COEs are small, nascent and experimental. In every case, they play an essential role.

How have companies created successful analytics COEs that propel measurable business outcomes?

Building a successful analytics COE

Successful COEs focus on collaborating with the business. Philip Jenkins, director of Verizon’s Analytics Center of Excellence, says Verizon’s COE was created because the IT department and business departments like marketing, finance and operations were all doing data work in an uncoordinated fashion. The center now acts as the hub in a hub-and-spoke model, with the spokes being data consumers in the business units. Jenkins says that, from the outset, the goal of the center has been “to make our data more powerful so that we can have better outcomes — what we call ‘simple, smart and connected experiences’ — with our customers, so that we don’t waste their time, we make more personalized offers, and whatever actions we take are more relevant to what’s important to them.”

It’s also critical for the COE to focus on strategic business priorities. Margery Connor, founding manager of Chevron’s Modeling and Analytics Center of Excellence, recommends setting up a prioritization system based on business value. Otherwise, “you get bombarded with all these ideas and some of them are higher value, some of them are lower value,” she says. In fact, Chevron’s center is guided by an “enterprise data science steering committee,” with representatives from procurement, finance, business units, and the CIO and chief technology officer.

As a result, “we get to show them what we are working on, and they identify areas where we could be more opportunistic. To work on a project, we need a well-defined business problem, a reasonable data set and a [business] champion,” Connor says. “If you don’t have a vocal champion in the business, then, more than likely, even if you solve the problem, it won’t get implemented.” 

Best-in-class COEs

Analytics COEs support both business and IT. They enable a transformation from department to enterprise scope and from tactical to strategic. At the same time, they create a pathway for an operational transformation for enterprise-wide data quality, strong decision-making, optimized value and business alignment. At their best, analytics COEs focus on enterprise-wide coordination; they thrive with both central and distributed resources. Because analytics expertise is scarce, the COE owns or creates a network of experts who can help anybody in the enterprise. The COE acts as an influencer node on a network of distributed experts or competency centers, bringing resources and connecting networks of people engaged in AI projects.

The COE is particularly valuable when the enterprise already has some isolated learnings and successes and wants to reach a more global scope. This COE will align, advise, support, communicate, educate, govern, architect and standardize all aspects of enterprise data transformation.

Despite, or perhaps because of, the excitement about these technologies, AI and data science face an uphill battle. Data experts look like auditors, and sooner or later, a machine learning algorithm hits a land mine and gets somebody fired. However, a data-driven culture is not a “nice to have”; it’s essential for business success. For most enterprises, the analytics COE offers the best hope of achieving the promises of data science. 

Serge Findling is VP of research with IDC’s IT Executive Programs. Martha Rounds is research director for IDC’s IT Executive Programs. IDC’s Mitch Betts also contributed to this article.

This article originally appeared in the Spring 2019 digital issue of CIO magazine.

Copyright © 2019 IDG Communications, Inc.

Download CIO's Roadmap Report: 5G in the Enterprise