Expert Exchange: Creating Community with Embedded Analytics Teams

Client Inquiry:

I'm leading the effort to embed analytics teams into other business functions, but I want their allegiance to be to analytics rather than the function they support. How do you create enough autonomy to ensure integration with the business units, while at the same time ensuring the teams don't go native?

Expert Takeaways:

Continuum of Operating Models

The challenge with embedded analytics teams is that you effectively have a collection of autonomous states that don't adhere to a central hub. On the continuum of possible operating models, which would suit your situation?

• Fully embedded and self-sustaining analytics team spread over the organization: anarchy. There's no control and no economies of scale.
• Hub and spoke model that’s led by business units: analysts align to enterprise-wide analytic guidelines, but they set their own analytic agenda and own the execution and adoption.
• Utopian "ideal" of perfect balance between business units and centralized analytics: Center of Excellence essentially acts as a consultant. The central deployment of resources becomes a managerial and bureaucratic nightmare.
• Hub and spoke model that’s led by a central analytics team: hub sets analytics strategy across the business units and deploys capabilities to come up with solutions.
• A fully centralized analytics group: ineffective because analysts have no feeling of connection or responsibility to the business units and because you can't centrally plan everything.

Hub and Spoke Model

For your organization, I recommend the hub and spoke model that's led by a central analytics team.

• Think of the hub as a special forces team whose members can be dropped into the business units to help out the embedded analysts with specific tasks.
• The spoke teams having the foundational ability to support their business units with the hub teams augmenting them with specialized or more complete data science capabilities, such as machine learning.
• The hub analysts are trained in the same way and have the same allegiance.
• The ability to move the hub analysts between teams allows you to modularize.

One of the challenges of this hub-and-spoke model is how you determine who in the hub team is supporting what and where their time is spent. You can use an Agile project management methodology to:

• Assign responsibility for task completion and move hub analysts between spoke teams.
• Organize cascaded standup meetings starting at the beginning of each day: ground-level analysts meet for 15-minute standups with their direct supervisors, then these supervisors meet with their supervisors, and so on up the chain.
• All new requests coming from the business units cascade up to central command in the hub, who can triage the tasks.

The Human Social Dynamic: Common Identity

To maintain loyalty from the spoke analysts, hyper-secure their identities as members of the hub organization. Empirically, people leave because they don’t get recognition and a sense of fulfillment from their supervisors.

• As a consequence of the cascading daily standup, I make it a habit to send notes to individual contributors thanking them for their contributions. You can create boilerplate messages for efficiency.
• Create mechanisms to reinforce their sense of attachment to the collective. This can be anything from totems they keep at their desks, updated business cards, and silly awards, to gift cards and quarterly retreats.
• Not everyone wants to be the center of attention, but no one wants to be ignored either. Most people need to hear how they've done well.