OVERLY HIGH STAFF LEVELS USING ERLANG C

Call Centre Helper Forum » Call Centre Management

(8 posts)

Tags: | Erlang Calculator | measurement |

Hi All,

I am in need of some help badly.

I am working in my first ever forecasting role for a call centre.

I have downloaded the spreadsheet from here and entered in my call data, but I am getting extremely high "Agents required".

If I work it on a per line bases (we have 9 incoming lines) it tells me that I need in one particular hour 164 agents.

However, if using the same data I then group it by our phone skills set, it tells me that I only need 31 agents in the same hour.

Help!!!!!! Can someone please explain why the high difference?

Thanks for your help in advance
Belinda

Posted 2 years ago

applebyd

OK....

First things first....

Have you checked the data you're using?

Now, without the actual numbers I can't be too specific, however, it
is a function of probability (and Erlang in particular) that individual
lines will always require more agents than grouped.

Using the Call Centre Helper calculator and arbitry numbers...
(Both top Calculators being used)

100 calls in a 30 min period with an average handle time (AHT)
of 120 seconds and a target answer time of 20 seconds gives a
requirement for 10 agents to hit an 80%+ service level.

Now if you have 5 independant lines this would then scale up
to 50 agents.

Now if we look at the skillset side and these lines combined / blended
for the same service levels and answer times we get a requirement for
38 agents to hit the same service standards. at this point an immediate
difference of 12 agents.

Whilst what you have appears to be extreme, one area to look at
may be the service levels you are trying to hit. Above 90% the
additional agent requirements for a 1% increase in service level
quickly become financially prohibitive.

Hope it helps.

Good luck, and, let us know how you get on.

Regards

DaveA

Posted 2 years ago

lilroos

Hi DaveA,

Thanks for your information.

Can I throw some figures at you for a little more help.

Just to make sure that I am doing it all correctly (this is really starting to make the brain work overtime).

Do you perhaps have a good reference guide that you can recommend that will help me understand all of this better? I have searched the net and come up with alot of information but nothing that gives it to me in simple terms.

Again thanks for your help.

Belinda

Posted 2 years ago

applebyd

OK,

Why not give us the numbers in the format

For a half hour
Line 1 = x calls.
Line 2 = x calls.
and so on....

Also your target answer times and Svc Level.

I'll have a quick look and see if there's
anything odd about it.

Regards

DaveA

Posted 2 years ago

lilroos

Thanks DaveA,

Target Answer Time = 30 secs
Svc Level = 80%
AHT in Seconds

Line 1 = 102 calls, 193.8 AHT
Line 2 = 34 calls , 78.6 AHT
Line 3 = 6 calls, 84 AHT
Line 4 = 13 calls, 132.6 AHT
Line 5 = 482 calls, 73.20 AHT
Line 6 = 739 calls, 181.2 AHT
Line 7 = 3 calls, 33.6 AHT
Line 8 = 91 calls, 242.4 AHT
Line 9 = 39 calls, 205.2 AHT

For Grouping:

Group 1 = Lines 1, 6, 7, 8 and 9
Group 2 = Lines 2, 3, 4, 5

Really appreciate you having a look at this for me.

Thanks

Belinda

Posted 2 years ago

applebyd

OK, This is going to take a while to look at.

One of the initial issues would be the varying AHT
so when you average the line data you are effectively using
the average of an average,statistically speaking a pretty big no no,
to calculate your FTE requirements, however, what I'll do
is run this weighting the data and see what I get.

Not going to be today though :-)

Check back early next week and I'll see what I can do.

Regards

DaveA

Posted 2 years ago

applebyd

OK....

I get, using the numbers you've given.

With the raw data:
Line 1 = 15FTE
Line 2 = 3FTE
Line 3 = 2FTE
Line 4 = 3FTE
Line 5 = 23FTE
Line 6 = 50FTE
Line 7 = 1FTE
Line 8 = 16FTE
Line 9 = 7FTE

Which gives a total of 120 FTE required
resourcing on an individual line basis.

Now if we group and weight the averages we get

Group 1: 674 Calls 192sec AHT = 78 FTE
Group 2; 535 Calls 75sec AHT = 26 FTE

For a total requirement of 104 FTE

An immediate drop of 16 FTE.

Although it must be remembered that the queues, like
dice and the lottery*, have no memory! This is one situation where
whilst you have good data, the Erlang process is probably not going
to be enough for you. Modeling it this way gives an ideal solution
that will only work in a specific case. I think you have an argument
for looking into a full forecasting system / WFM product.

Hope it helps,

Drop a line back if you need any more info.

Regards

DaveA

* Something for people who look at 'hot' and 'cold' numbers
would do well to remember!

Posted 2 years ago

lilroos

Hi DaveA,

Thank you so much for your help.....at least now I know that I am on the right track and am doing it the correct way.

We are looking at getting a WFM system, but the bosses want to make sure I have my head around doing it manually before we move to that.

Have you (or anyone) ever heard of a system called Openwave. This is what the bosses are looking at.