I N S I D E
Robert Dobson Named
2012 Workforce Management Professional of the Year
Measuring Productivity in the Inbound Contact Center
WFM Survey Results
WFM Winter Survey
Scheduling Swap Ideas from the 2012 SWPP Conference
Strategic Service Optimization
10th Annual SWPP Conference a Huge Success – Were You There?
Customer Satisfaction Through Performance Management at Sprint
Events Calendar
Ask the Workforce Wizard
Our Sponsors
Join SWPP

swpplogo

Visit our Website at
www.swpp.org

s
 
 

Question: I was wondering if there is any traffic calculation model for e-mail and calls – as our call center handles customer queries via both channels.

Answer: There is a model for calls and another model for e-mails. Calls are randomly-arriving work and subject to speed of answer goals typically measured in seconds or at the most minutes. There is a risk of abandonment if callers wait too long. The metric here is all about how long it takes to answer the call but does not include the time to do the work. Therefore, the model most commonly used in simple queuing environments is the Erlang C model. However, this model does not work well in environments utilizing skill-based routing where more sophisticated algorithms or simulations are typically applied.

E-mail work is not randomly-arriving in that it typically sits in an inbox waiting for a response and then is worked in sequence as time permits. There is typically a response time goal measured in hours or even days. There is no risk of abandonment, and in fact the customer who does not receive a timely reply may well send another e-mail or call which simply adds to the workload. In this case, the response time goal must include not only picking up the e-mail but also doing the work since the sending of the reply is the first the customer knows of the activity. Given the sequential nature of this work and the long reply time, the Erlang C model does not work here so there are different math models for this type of work which consider the workload, the shrinkage, but do not have to consider speed of answer but rather response time.

The real challenge comes about when you mix the two types of work for a single team of frontline staff. Now you have some randomly arriving work that must be answered in seconds queued into the same line with e-mails that may have hours or days to be done. But if you put off the e-mail too long, the response time goal will be missed as well and the work will multiply as mentioned above. Theoretically, the idea is to use e-mail to “fill in any spare time” an agent might have between calls. But if the agent is working an e-mail and will not answer a call until it is done, then the e-mail makes the agent unavailable to the phone queue. If the agent can drop the e-mail and answer the ringing call promptly, then maybe it can work but the risk is the overall time it will take to finally finish the e-mail work and the potential quality issues of the interrupted work. So the simple answer is there are different models for each and combining them takes the models into the realm of “higher math.”

Question: I am exploring the advantages of expanding the hours of intra-day management in our call center, but would like to know what measurement(s) will help me identify if there is any value once the pilot is completed. We are a 24/7 operation and currently have a dedicated resource for intra-day management from 7:30 a.m. to 6 p.m. Monday to Friday. We are considering piloting additional hours until 9 p.m. and on weekends.  What metrics should be taken into consideration pre- and post-pilot to determine if there is any value in additional resource time?  Thank you for your consideration of this scenario.

Answer: The justification for extended hours of real-time monitoring should consider the volume of work (risk of failure if not monitored), volatility of the workload and ability to move resources around to match, workload for the analyst, and budget (ensuring the most effective use of analysts).
The Operational Effectiveness measure would be a good way to measure the before and after of a pilot. This is the comparison of the actual bodies in chairs versus the new forecast requirements once the actual call volume, shrinkage, and AHT are known. If this analysis is done on a half-hourly basis, the variance during the hours when the real-time center are open versus when it is not would give a clue as to the value of longer hours of operation. Where the match of staff to workload is already solid, the added effort and expense may buy little. Where there is a genuine cooperation between the WFM team and the operations group to make real-time adjustments and they appear to be needed during those hours, this could have a solid ROI. 

Have a tough question?

Send it to wizard@swpp.org and we’ll try to find an answer!