Some Erlang C tools will give you service level, queue lengths, occupancy and other figures for agent levels around the most optimal level. For example, if Erlang C tells you that you need 35 agents, these tools will tell you what would happen if you had used 29, 30, 31, 32, 33, 34, 36, 37, 38, 39 or 40 agents in a tabular or graphic format. When looking at such a display the dynamics of queuing theory really jump out at you. You see that the addition or subtraction of a single agent can make a big difference on average speed to answer and queue levels, but hardly any difference on agent occupancy. This is why managing a call center by "peeking over the cubicles" doesn't work. How busy your agents are is not directly related to what callers are experiencing. We make two Erlang C and ACD simulation tools that ought to convince your doubter. -- Stu Harris, Portage Communications

Dear Harried & Hopeless: I understand you are harried, but I do belive that there is hope for you after all. The very "evidence" - of adding one more agent and getting an "absurd" difference - is in fact what the Erlang formula is all about. If you use an Erlang calculator - like ICMI's QueueView Calculator software, and plug in the input data (your AHT & half-hour call volume), the software output will be a table which will clearly show that for each new headcount added, the impact on Service Level is dramatic initially (at low Service Levels), and as the number of people you throw in the equation increases, there is a point of "diminishing returns" - which means that the Service Level will not improve as much even if you add an "absurdly large" number of people (say 10 or 20). So the Erlang calculator does move from an "absurd" Service Level figure to an "absurd" agent figure!!! This is by design - not accident - because Erlang is taking into consideration the 'random' nature of call arrivals. Hence the task of the contact center manager is precisely that of figuring out what is the minimum number of "bodies in chair" required during the half-hour to "just meet" the target Service Level. After all, your employees cannot have everything in their control, the least of which is the pattern in which callers arrive in your call center. A good idea would be to ask your boss the reason for his belief. For instance, does he see that the agents are 'less occupied' and that there is room for improvement? Erlang, (specifically QueueView from ICMI) also clearly shows the relationship between occupancy levels and Service Level at given number of agents. The principle is simple, if Service Level target is high, then agents need to be 'more available' to handle the probability of the next random call. If Service Level is set lower, then agents can be 'less available', or in other words more occupied. So I think that your objective of 'winning him over' needs to take all the above into consideration, and more. If you can be more specific with: what is your service objective, how far are you from it, whether other things like quality have been taken in consideration to establish that agents indeed perform to their potential, etc. etc., I will be happy to delve deeper into the issue. Thanks. -- Rajas Daithankar, Avaya GlobalConnect Limited