Help! I Don’t Think Dr. Erlang Loves Me Anymore!
1 May, 2008
By: Ric Kosiba Ph.D.Question:
Congratulations to me! I was just given responsibility for workforce planning of both of our call centers and our back office functions. The boss just asked me to do a what-if analysis on merging some of the different functions in several of our sites. My Erlang spreadsheet is going to blow up! What can I do? It looks like it is me, Erlang, and another boring late night at the office.
Dating Doctor Erlang
Answer:
Dear Dating:
Spending late nights in the office is almost written into our job descriptions! Look on the bright side, Dr. Erlang is a very cheap date. He’ll never turn you down and he’s willing to stay out all night. On the other hand, dating Erlang can be pretty unsatisfying.
Maybe it is time that you found a different beau. Dr. Erlang was fun when your life was much less complicated, when your work was easier, your networks simpler, and, to be frankly honest, when you had fewer other options. But let’s face it --- you do have other great options. It is time to drop Dr. Erlang for a real analysis engine. You need to find someone who is serious.
All kidding aside, the Erlang equation does not work when analyzing back-office functions. It was designed to solve a very simple problem: “How many switchboard operators do I need to handle all calls in the next small block of time?” It was designed for a simpler time and we all know there are serious problems with the Erlang equation when used for analyzing a modern contact center:
• It assumes that there are no abandoned calls.
• It assumes every contact center is exactly the same.
• It assumes that you have a single center and super simple routing rules.
• It begins to lose accuracy as the time period increases.
This last point is critical for longer-range planning. Over the next 15 minute interval, Erlang is a fairly good approximation, but over the next week, or month or year, Erlang falls apart. This is an important fact that our industry already knows: Erlang doesn’t even work well for long-term planning of simple contact centers.
And your problem is much more complex than that.
You have to plan for both traditional call centers and back-office/production functions. Erlang does not have a queuing equation for this type of setup. Trying to make Dr. Erlang into something he isn’t will …well… not work for either of you. You cannot change him.
So whom else can you bring to the dance? There are a number of technologies that have traditionally been used to model work processing centers --- regression modeling, queuing equations, workload estimates, etc.--- but they all crumble when it comes to complex processes. All of these modeling methods are problematic for answering what-ifs, especially if the operation is a bit complex.
This leaves us with only one other real option --- but don’t worry, it is a handsome option. That is, the shining knight of center analytics is discrete-event simulation. Discrete-event simulation is a technology that was designed to fulfill your needs. It was designed to perform what-ifs of almost any type, be it what-if analysis of call centers, e-mail functions, chat centers or production centers (like your back office). Further, with discrete-event simulation, you can evaluate what-ifs on virtually any combination of the different types of contact processing or queuing setup.
It is not to say that Mr. Simulation is easy; to some he might seem a bit deep and complicated. Building simulation models means you might have to mimic the complexity of your network into your simulation. But building good relationships takes time and is worth the effort.
Simulation allows you to:
• Model complex networks of any combination of contact types
• Define any performance metrics that are important to your organization
• Route contacts across multiple processing models (e.g. a call generating a back-office process)
• Test increased volumes or changes to handle times
• Test sharing of resources
• … and much more
I know it is hard. You’ve invested time in your relationship with Dr. Erlang. You’ve had a long history together. But it is finally time to ditch him and move onto someone who’s more dependable, serious and will return your (emotional) investment. Just let him down easy.
