The Analytics Organization – Part 1

Having been a part of large organizations, I have seen first hand the effect of both good and bad structures in Information Technology and Analytics. In analytics organizations, there are many different skill sets and the number of people calling themselves data analysts and data scientists it becomes increasingly difficult to determine how the analytics organization should be constructed.

This article will focus on the organization’s need for centralized or decentralized analytics team. Back in the early 2000′, many CIO’s struggled with the role and position Enterprise Architecture Team. To this day, I still hear CIO’s arguing about whether individual units should have their own Architect, while others have created specialized architecture teams that roam between projects.

In the field of analytics, the debate is basically the same. However, architects required specialized skills for development of systems, analysts may have limited skills in certain areas, only to find themselves thrust into making analytical decisions with serious consequences without having the requisite knowledge, toolsets or skills to adequately make those decisions.

Having a strong central figure for all analysts is the key to successful analytics organization.

One of the optimal methods for organizational structure is to create a hybrid analytics organization with a centralized head of the organization, who is unbiased and astute in dealing with complex political dynamics that can arise with contentious results from analysts. Within the group should be a small group of data scientists who are grounded in statistical techniques and who keep up on new techniques and technologies. These individuals should also possess the ability to teach other analysts within the group; furthermore, these individuals should act as mentors reviewing work, assessing methods and recommending approaches to other analysts.

The other analysts in the group, may have a variety of skills, but should be have a set skill and very good domain knowledge for given areas such as finance, marketing, etc, and then be collocated within the business unit. In this way, there are dedicated cycles of time for the given domain, but knowledge can be shared across the organization through this hub and spoke group with constant checks and balances. In addition, the central hub, acts as a balancer to ensure that the results derived are unbiased and free from influence of business units.

This structure forms the basis for an unbiased group with a strong central figure that ensures the integrity of results, cooperation and communication between the units, and allows for the proper allocations of resources for training. With the support of key data scientists in the group , individual analysts can then have sounding board and support from strong central group with the skills and political savvy to navigate the difficult and turbulent waters of data analysis and organizational behavior simultaneously.