I recently had lunch with some new friends, one of whom has a 17-year-old son entering college. He’s really interested in math and statistics.

Given the insane cost of college education, why not graduate with a degree in a white-hot field? At the top of the list these days is data science. Demand far exceeds supply. If I were a sophomore at Carnegie Mellon with a $63,000 annual bill, you can bet that I’d strongly consider going down this road.

This begs the question: Does an organization really need a data scientist to “do” Big Data?

Culture and Data Science

The answer is an equivocal no. You can rent them on sites like Kaggle and an emerging group of competitors.

More generally, rare is the organization that fully taps the potential of its employees. No, you can’t send a bright analyst to a two-day training course and expect her to become Nate Silver. You can, however, encourage employees to experiment and reward data-oriented thinking and solutions. And you can hire for numeracy. Why should any position not require basic fluency with statistics?

Above all, you can reject the notion of a center of excellence, defined as:

a team, a shared facility or an entity that provides leadership, evangelization, best practices, research, support and/or training for a focus area. The focus area in this case might be a technology (e.g. Java), a business concept (e.g. BPM), a skill (e.g. negotiation) or a broad area of study (e.g. women’s health). A center of excellence may also be aimed at revitalizing stalled initiatives.

Why centralize knowledge in one pocket of the organization? Why not diffuse it as much as possible? On HBR, Michael Schrage argues against black boxes and centers of analytic excellence.

Simon Says

I think back to the time that I spoke at Netflix headquarters. With the exception of the receptionist at the front desk, just about everyone I met there struck me as analytical, even the company’s head of PR. Is it any wonder that it is so successful?

Reject the notion of a center of excellence.

Data-wise, your organization may not be able to catch up to Netflix next week, next month, or next decade. Be that as it may, relying upon only a small cadre of “experts” to help understand, interpret, and act upon data is rife with peril. A single defection may hamstring a key department, product launch, or even the entire organization. Do what you can to diffuse expertise.

IBM sponsored this post. The opinions in it are mine alone.