Possible mistakes on erlang calc

Call Centre Helper Forum » Call Centre Management

(3 posts)

Tags: | call centre | Erlang Calculator |

Dawn1983

Hi - I'm using the Erlang calculator spreadsheet to calculate how many agents I need logged in to cover calls. The targets are:

80% of calls answered witin 30 seconds
30 second wrap time
Avg call time (inc wrap time) 360seconds

I've worked out the average calls taken for half hour periods over 1 day. Late on in the day, the average is only 1 call per half hour. However the spredsheet is telling me that 1 agent, to take that 1 call, will only result in an 89% predicted service level. Surely that should be 100%?

Also, for a different half hour period, if we take 15 calls, and we have 2 agents, the predicted service level is 0% - shouldn't that be at about 25% at least?

Any help is much appreciated!!

Posted 5 months ago

applebyd

Hi Dawn,

I'll try and do this without turning into the Math geek!

Erlang is always 'interesting' around the very low numbers, remember
you're telling it that ON AVERAGE you get one call, not, that you are
guaranteed to get one call. Therefore it actually starts to 'hunt' around
the probabilities hence the 89% you're getting.

Interestingly everyone I know has tried to put one agent in at some point
to see the result!

With regards to your second point, yes it would be roughly 25% IF calls arrived at
predictable intervals, however, the algorithm presupposes random arrivals, therefore you cannot expect (within the math) that a call will be queuing within
it's service level as an agent becomes available.

Hope it helps (a bit)

Regards

DaveA

Posted 5 months ago

Dawn1983

Hi Dave,

Thanks so much, that's helpded a lot!!

Cheers,
Dawn

Posted 5 months ago

RSS feed for this topic