Chris Crosby

I’ve been pondering lately some the practices and methodologies in the Contact Center industry that are out dated and in need of a new approach. One thing that comes to mind is how people measure and manage “Agent Occupancy”. My first lesson in Occupancy came in the summer of 1995 when I transitioned from the domain of outbound telemarketing into the then foreign world of Inbound Call Centers. My manager sat me down and tried to explain to me the basic concepts of running an inbound call center. Somewhere between describing Service Level and Calls Per Hour, the concept of Agent Occupancy came up. There are many variants to the notion of Occupancy, but for simplicity and purposes of this discussion let’s define it as the percentage of time that an agent is occupied, or working, during any given period of time. For example, if an agent spent 45 minutes of the last hour talking to customers then their occupancy rate was 75%. Occupancy is traditionally, and almost always, measured at the Skill Group Level and not by individual agents. Urban mythology says that the higher your Occupancy the more efficient your staffing plan is, and thus your profitability. However, my manager cautioned me that, you “never want to run your occupancy rate at higher than 80% because you will burn your agents out, and your attrition will be higher.”

Now, as I mentioned, since “agent occupancy” is really measured by “skill group” and not by agent, let’s call it what it is Skill Group Occupancy. Coming back to 1995 and my manager’s (mostly correct) notion that the higher the rate of occupancy the more likely you are to burn out your agents (I also believe the inverse of that to be true); certainly a general rule of 80% doesn’t pass the acid test and there are other variables that impact an agent’s decision to quit; for example, the type of calls an agent handles (billing inquiries typically have a higher rate of agent churn than, say, general technical support). Additionally you will find that Skill Group Occupancy, as we’ve defined it, does not mean the work load is evenly distributed amongst the agents within that Skill Group; and you inevitably find that some agents are busier than others (the problem is compounded if they are multi-skilled).

So here’s the deep thought of the day: Could one build a model that can predict an agent’s likelihood to churn based on their individual “occupancy” data points? My sense is that we can. By taking Skill Group Occupancy down to the Agent level, we should be able to build an algorithm that evaluates factors like the amount of time an agent spends by each Call Type (not Skill Group) and overlay it to when similar agents churn. We could also ascertain their comfort level, and thus “stress level”,  with specific Call Types by looking at the amount/percentage of time the agent puts customers on hold, conference, has longer than normal talk time, or transfers calls to their supervisor or escalation queue as compared to other agents. If this notion is correct, then one could build staffing plans that not only optimize call handling but also prevent attrition.

Any thoughts, feelings or opinions?