by Jill Dyché

The new enterprise analytics model

Feb 11, 2016
AnalyticsBusiness IntelligenceData Mining

Analytics Centers of Excellence were all the rage back in the day. But executives are considering a new model.

I started designing and working in Centers of Excellence (COEs) over a decade ago. At that time they represented the evolution of the everyone-does-everything model of organizational design, aiming directly at duplicate work efforts, vague accountability and turf wars.

I was particularly focused on the Analytics Center of Excellence, an organizational construct larger companies were forming to centralize key reporting, predictive and data skills. It was a shared service model relying on a hub-and-spoke engagement structure, in which business people across departments sought out the COE’s experts when they needed help building new reports, accessing or integrating data, or training users.

COEs work best when management believes in the economies of scale introduced by centralized, cross-functional teams. With the emergence of shadow IT business people managing their own technology portfolios has become commonplace, and members of analytics COEs find themselves looking for ways to help increasingly self-sufficient stakeholders.

Many companies are now re-examining the value of their COEs, looking for lighter-weight approaches to enabling enterprise analytics while trimming bureaucratic engagement processes and reducing request backlogs. One alternative is what I call the Analytics Marketplace.

Like its COE predecessor, the Analytics Marketplace takes the form of a single team that acts as a service—often funded from steady-state budget—to the rest of the company. But unlike the COE the marketplace doesn’t develop analytics solutions. No protracted requirements gathering meetings, user interviews, or prototyping. No “approved vendor lists.” No scrums.

Instead, members of the marketplace help various business units procure the right analytics solutions in order to streamline their own development activities. This includes negotiating and renegotiating vendor contracts, cultivating consulting partnerships, identifying third-party data providers, and arbitrating service level agreements. They are as externally-facing as they are internally-focused, and essential to their value proposition is ongoing education and trend-spotting.

They also track existing analytics capabilities and monitor progress. One analytics marketplace team I’ve worked with maintains what it calls a “Common Report Library,” which it avails to the entire company via a publish-and-subscribe model. Say a sales executive wants to run a “Profitability by Territory” report. The Analytics Marketplace team knows that the finance department runs that same report every month. They add the report to a centralized “library” that, after a quick tutorial, is shared with the sales team.

No new tools need to be procured, no reports built from scratch. The Analytics Marketplace ensures that existing capabilities and skills are understood and quickly dispatched. Software tools are optimized around user communities according to how analytics and data are consumed–simple dashboards for some, complex modeling algorithms for others, down the line to knowledge discovery and machine learning. Members of the marketplace help provision these capabilities as needed, leaving their use to the business units—often staffed with their own data scientists or analytics experts—that made the original request. 

Does the Analytics Marketplace model imply that BI and analytics have become commoditized? That depends on whether new business initiatives embed analytics at the outset or whether analytics is treated as an afterthought. A marketplace model offers knowledge workers assistance where they need it, and freedom to build the best solution to solve a business problem. Think of it as the evolution of the COE.