I am teaching service operations this quarter so I was more than a little intrigued when last week’s McKinsey Operations Extranet featured an article titled “Relationship service cells” (it’s free but you have to register). Here is the idea in graphic form:
The idea is much like any cell implementation: replace a functional layout in which jobs jump from function to function with a process oriented layout in which co-located resources smoothly move work through. The focus here is on financial services. Cells have been tried before in financial services (see, for example, The Lean Service Machine, HBR, Oct 2003) but those applications have generally been in high volume back office operations like issuing life insurance policies. McKinsey is pitching a more forward facing setting in which clients call into an appropriate cell that can handle virtually any of their transactions:
[R]elationship service cell (RSC): a cross-functional, collocated team, or cell, that handles most of the major value streams that affect the client. The RSC integrates back-, middle-, and front-office processing into a seamless unit: for example, at any moment an asset manager’s RSC might be opening a new account, overseeing asset distributions, and resolving problems over the phone, all for one registered investment adviser.
Part of the benefit is that the service level delivered by each cell is tailored to the needs (and profitability) of the clients. High value clients get their calls answered right away while lower value ones may have to wait for a call back.
So to my mind, what differentiates this set up from your standard manufacturing cell is that there are many processes happening at once within the same cell. Opening an account or completing a trade are two different activities and may touch on different parts of the firm. How can one easily balance the load across workers when there is little control over the flow of jobs coming in? Cross-training is obviously one answer, but there has got to be some limits (whether regulatory or human) if you are going to handle virtually all transactions for customers. That is, I am suspicious that it will be easy to “match capacity across steps” as the figure above claims. Let me also add that I wonder how easy it will be to generate continuous improvement in this environment. In manufacturing cells, every worker touches every item (OK, that is an exaggeration) and has a view of the entire process. Here there are distinct processes and not every worker is necessarily involved. How can you improve one process while being sure you do not adversely impact another?