How to handle call volume fluctuations?

FOLLOWING TOPIC PASTED FROM YAHOO GROUPS

ORIGINATOR: Nirmalya Kumar Chaudhuri

DATE: Thu Jul 24, 2008

Team,

Pls find attach 3months data.

I am totally new to Service Industry / Call Center. I was with mfg industry for last 14yrs.

Our problem is , we are providing daywise forecast for call and we have given target for service level; abandant% & AHT to our call center.

Now since the call volume fluctuates , it is affecting SL ; abandant% ;AHT etc. Everytime Call Center is telling us that this is due to fluctuations in our forecast [ although they are maintaining extra manpower to handle some fluctuation ].

How to handle this scenario ? How can we validate the same ?

Can we do some regression with PI to fix the acceptable target ?

What should be the confidence level ?

I want some statistical method to explain my call center mgmt about the Acceptance Level [ with due consideration of fluctuations in forcecasted call ]. Service Level target is now 75%.

Regards,

Nirmalya

_________________

FOLLOWING RESPONSES WERE PASTED TO YAHOO GROUPS

Please find some preliminary analysis of your data attached below. As discussed over the phone, I have only done the analysis for SL. Similar analysis can be performed for AHT and Abandonment %.

From the attached analysis, I concluded the following:

a) The correlation between forecasted volume and actual volume is very low. The forecasted volume does not have much predictive value.

b.) There seems to be a decent correlation between call volume received and service level. As call volume increases, the service level decreases. Due to the amount of variation in the data, the prediction limits are pretty wide. For example, for a call volume of 40,000, the predicted service level was around 60% but for new data this could wary anywhere from 30% to 90%.

c) If you work with means, then the confidence intervals will be much narrower.

d) Looks like if you want to assure a service level of at least 75%, the current setup is not able to handle it, if the call volume is greater than 38,000 (approx).

Please review the above and we can discuss more later. If anyone else has a different take on this data, please post your analysis to the group.

Best Regards,

SJ.

_______________

Hi All

Just wanted to added , basis the forecast is the manpower planned . Then this is the problem as the manpower is fixed on the basis of Forecasted . Volumes > the forecast and the manpower is not able to handle the Volume . 1) need to built better forecasting model ( use statistics method like regression / time series analysis / use of minitab ) 2) Variance of 10 to 15 % is acceptable in the forecasting 3) Manpower needs to planned accordingly 4) Need to check is there any unplaned event which is trigerring call volumes 5) Use the Erlang calculator to calculate the manpower for a realistic picture .

Regards

A S

_________________