I wonder if anyone can help me out. I have a dilemma and need some advice. I have a staff base of about 25 advisors handling calls against a forecasted workload within a call centre environment. I\'ve noticed that calls are queuing and several calls have already been abandoned. What data/MIS should I investigate to understand the cause of the increased call volumes? What actions do I take to mitigate the situation?

Thanks

Hi Darren,
having worked on a similar problem, I came up with a very simple solution. Colect data through through your CMS as to what is the call flow like on an hourly basis. i.e. For Every hour of the shift you would need data like Calls Received, Speed Of Answer, Calls Queued, Calls Abandoned, Service Level, Average Talk Time Per Call, Idle Time. and see where you need more manpower. You can keep updating this over a period of time, it will almost always predict the accurate days or time of the day you would need mmost of your manpower.

Darren

Question, if you dont already have a form of MIS how do you know you have abandoned & queueing calls.

Asumming that you are completely unable to identify these calls but you just know its happening then the best solution short term would be a call logger.

This would give you historic information on peaks & troughs and also be able to identify the numbers calling you. You would then be able to ensure that your staffing levels are adequate to answer the calls.

This is only a short term cost effective solution as I am guessing that a call logger would probably highlight a few issues. If this was not the case I would have to recommend an independant review.

Jason

Have a read here.

Penguin-Programming DOwnload

This should let you map the main problems and apply (and I can't stress
this enough at this point), SIMPLE solutions

HTH

DaveA

Hi Dave,
Read your document but are you saying what I said was wrong or are you saying its just a simpler way of saying it??

Amol,

Average is a very misleading word.

What I was trying to point out is that on a simple level with
a lower call volume yes a simple (average handle time * calls)= Agents
can be applied however as we all know 'calls bunch'.

The doc I linked to trys to run through these points and
then allow someone to see where it breaks down.

As a rule of thumb:
i) With low volumes, averages work, Erlangs don't

ii) With high volumes, Erlangs work, Averages don't

From my experience this starts to hit when the call volume goes
above about 7 per 1/2 hr with an AHT of 5 min or more. These seem
to be more or less (inversely) proportional.

At that point it's over to Agner...

HTH

DaveA

I liked Dave's document, good common sense. But there are two more things to consider.

All calls can be divided into two types; value demand and failure demand. If the call volume is increasing then it might be because you are getting more value demand i.e. more people are calling to request your service. Or because your failure demand is increasing, i.e. the business process is failing somewhere and people are calling to chase, report errors or complain. Or it might be a mixture of both.

If it just an increase in value demand then good stuff, read Dave's document and look at scheduling and resourcing etc. But if there is an increase in failure demand you need to find what is causing it and remove the source of the failure. With the failure demand gone (or reduced) call volume with drop and the scheduling might not be so pressing.

Also, do you know the capability of the system and of the demand? How predictable is it? The tool to use is the Statistical Process Control chart (or more recently the Process Behaviour chart). This is basically a run chart with calculated upper and lower limits which will tell you if the process and demand is under control. Put "Statistical Process Control" into Google and you will get plenty of results. If your process is not in control then work to remove the special causes, if it is, work to reduce variation. Some of Dave's process comments will help here (especially the one about getting the staff to suggest improvements).

Sometimes other factors influence call spikes. For example, one of my clients is a camera manufacturer. We have determined that not only do we get spikes on holidays (when people get or use new cameras), but we also get spikes after serious weather, and when there is a good heavy rain in an area. So if there is a major frontal system crossing a heavily populated portion of the US, we are flooded with calls. (Apparently everyone is bored inside and decides to figure out how their camera works.)
So you can use the averages and Erlangs, but some call overflow may still occur occassionally. You will need to think about your customers and when they are most likely to need your help and factor that in to your equations.

Marianne