Let's Talk About Data Science's Role
View Comments: Newest First | Oldest First | Threaded View
Re: Our approach
  • 7/26/2014 8:24:52 AM

I would think the two roles would depend greatly on the size of the organization. A large one would almost need the "isolation" instance where they are separate, while a smaller company can easily form the analysts inside a department like sales for instance. Probably, a somewhere inbetween would be ideal. Or form a separate group that mediates between the two extremes.

Re: Our approach
  • 7/25/2014 1:37:24 PM

@MattH, it sounds like you're taking the approach of a company with a more mature analytics program -- so kudos to you. I especially think your point about knowledge sharing and loss of knowledge is a really important one given the analytics talent demand of today. You don't want to face a setback anytime one of your data analysts decides to leave the company!

Just in case you haven't seen this previous content, I thought I'd share a couple of recent posts that you might get some take-aways from.

 Creating an Analytics Nerve Center: This is an audio interview with Patrick McIntyre, senior vice president of healthcare analytics at WellPoint

Analytics SLAs Prove Healthy for WellPoint: This is recap on some of what he discussed

Zurich North America: Hot Analytics in an Unexpected Place & 4 Ways to Pick Out a Great Analytics Hire: These are blogs from an interview with Peter Hahn, who's building a center of analytics excellence at Zurich North America


Re: Multi skilled
  • 7/25/2014 11:28:03 AM

@Jamescon I agree. It's more likely that this will help the company to promote analytics. It will be difficult to change long held beliefs and attitudes. Using their own team members as change catalysts would no doubt make it easier.

Re: Multi skilled
  • 7/25/2014 9:08:16 AM

@Phoenix. Maybe giving business unit leaders (or someone on their team) the ability to do their own queries would whet their appetites for more analytics based apps and provide the analytics team with advocates in the business units.

Multi skilled
  • 7/24/2014 11:57:33 PM

Both models have their pros and cons but it might be interesting to see how team members themselves handle their own analytics like you mentioned. Team bias issues will be there, but I feel doing your own work would mean a better grasp of exactly what is needed. There could be a analytics professional who could then go through everything and make the final analysis.

Re: Our approach
  • 7/24/2014 1:28:41 PM

@Matth. Thanks for sharing, Matt. Great point about the potential to lose out on skills if people leave under the federated model.

That liaison in each department should help keep good ties with the business units.

Our approach
  • 7/24/2014 1:24:22 PM

This is something that has been a major topic of discussion within my organization.  We have never had any data analytics, nor data analysts/scientists to speak of, so we essentially had tabula rasa for instilling this competency once I joined the company.

At first, we thought the federated model was most appropriate.  After all, health insurance is a vastly diverse industry with widely varied data needs.  However, as we began to weigh the pros and cons, we quickly realized the best approach for a company our size (<200 employees) was a central Analysis Group. 

Our business is incredibly siloed already; the last thing we wanted to do was reinforce these silos.  Aggregating our data and analytics would allow us the most complete picture, as well as allow our analysts to freely exchange ideas. 

For a company our size, if we had one or two data experts in one line of business, that opens us up to a significant chance of losing intellectual capital if they were to leave.  By pooling knowledge, talent, and practices, our analysts become stronger and reinforce one another.  Having diverse perspectives keeps bias at bay and generates fresh ideas.

We have a liasion in each department, essentially the most data savvy business user in the area, that serves as an interface and domain expert when needed.

We are placing a large emphasis on empowering business users to work with their own data, however, we experience a large degree of resistance from our "legacy" users that are used to reports being giftwrapped and hand-delivered. 

I'd be interested to hear other approaches as well.