Forecasting What-If scenario

[WindsAloft]

I currently use eWFM (Aspect) and the cc-Modeler which has been suggested by this forum on multiple occassions, but I am having trouble using these tools to answer a couple of problems of mine. Anyone have any suggestions?

The Scenario: Our contact centers are currently struggling with low headcount and high shrinkage and unproductivity... not to mention increased call volume due to a marketing effort. The question is not regarding these problems, as we already have action plans to correct this in the long term. My question is regarding the short term... in other words, right now.

Facts: We have x training employees. They can either sign in and take customer calls, or they can walk the call center floors, assisting where possible (to hopefully reduce AHT).

Question: How much of a reduction in AHT would the team have to drive, in order to outweigh the benefit of them answering calls? (benefit measured in reduction of ASA)(assuming their performance is equal to an average agent)

I am aware that the answer might be different depending on call loads for that part of the day, so I was hoping to solve this problem for 2 different parts of the day.


Actual Data for Period 1:

Period = 60 minutes
Calls Offered = 1745
Calls Handled = 1541
Avg Handle Time = 467 seconds
Headcount = 240
ASA = 116 seconds

Actual Data for Period 2:

Period = 60 minutes
Calls Offered = 2344
Calls Handled = 1691
Avg Handle Time = 536 seconds
Headcount = 281
ASA = 327 seconds



I know I can somehow breakdown Erlang to manageable algebra, but I'm not sure I trust myself. Any experts out there willing to assist?

[TheTerminator]

Basically, during period ! you are understaffed by 10% or 24 FTEs. If you wanted to decrease you AHT you would need to reduce it by almost a minute to reach a goal of effectively answering calls in an 80/20 mode.

Period 2 is obviously more understaffed than period 1. Because of this, period 2s under staffing estimates are much more exaggerated. Your AHT, ACW, Abandonment rate increase more exponentially and not linear due to increases in being understaffed.

To answer your question: If your trainers are not really needed to answer questions by agents, put them on the phone digging for calls. If your agents need more mentoring have them work with the agents.

I hope this helps.

[WhatWouldReaganDo]

You could do some erlang c algebra, but I think a much simpler approach is to realize that if you are taking away an hour of resource (i.e. the x-training agent spends the hour assisting other agents instead of taking calls), you'd better be reducing the amount of work by an hour or more to compensate. However, each second of handle time actually creates more than a second of work (it will create 1/occupancy second of work). Based on the numbers you've given though, I'll bet you're occupancy is well over 90%. Let's say it's 95%. Then each hour an agent is crosstraining, he or she needs to reduce the workload by an hour or more and the total handle time of the other agents by 95% of an hour, or 57 minutes. 57 minutes is 3,420 seconds. So divide that by your calls handled (either 1541 or 1691) to get your required impact. It will be 2.0 to 2.2 seconds. Two cross training agents would need to impact AHT by 4.0 to 4.4 seconds, and so on.

Hope that helps. If you need clarification on any of the assumptions let me know.

Regards