Minerva raised her voice aloud, and made every one pause. "Men of Ithaca," she cried, "cease this dreadful war, and settle the matter at once without further bloodshed." - The Odyssey by Homer
I still remember it like it was yesterday; the first time I won an argument - not with logic or reason - but with data.
Our weekly Call Center Operations meeting had deteriorated into an argument about the quality of service our customers were receiving.
The argument began when the COO demanded that we explain why two of our Key Performance Indicators, Abandon Rate and Service Level were so poor [the metrics are explained at the bottom of this article].
I suggested that we had problems with our scheduling, but our Call Center manager insisted that because our Average Speed of Answer and Average Handle Time Key Performance Metrics were really good it didn’t matter if those other metrics were performing badly.
After a few minutes of arguing and a bit of name calling I asked them to give me a moment. I retreated to the far end of the conference table, opened up MS Access, pulled in some call data from my personal stash, and spun up a pivot chart.
Ben the Analyst: “…OK, take a look at this chart I just put together:
"Let’s look at yesterday, it’s a pretty good example. In the morning we’re overstaffed so our call capacity exceeds our call volume. After lunch we bump up our staffing, but not enough to cover the early afternoon peak. During the late afternoon we’re overstaffed again, then understaffed at dinner time.
On average we may have good handle times, but our typical caller experience after lunch and around dinner is terrible, and customers are hanging up like crazy! Even worse, in the morning and night time we’re overstaffed, so we’re paying for employees to just sit around.”
COO: OK, so how do we fix our scheduling?
***
Now, why the term Discussion-Time Analytics? This episode is an example of an important discussion that was destined to take place with or without analytics. In this case I had two good things going for me:
1. I had the data readily at hand, and
2. I was fluent with the tools for visualizing and analyzing the data
These assets allowed us to quickly perform some simple analysis right there within the conversation. The blocking question “Is there a problem with our Scheduling” was answered, and we immediately began seeking a solution.
In summary, we use Discussion-Time Analytics to fast-forward conversations by removing barriers of ignorance, and reducing the actions required to identify problems, make decisions, and build consensus.
How do you make Discussion-Time Analytics a reality? What’s the difference between Discussion-Time Analytics and Decision Support or other Business Intelligence applications? Good questions. Stay tuned.
***
A quick Call Center KPI primer
Abandon Rate is the percent of inbound calls that hang up before they are answered by a Call Center representative before the caller hangs up or “abandons”. If 10 out of 100 calls hang up before they are answered your abandon rate is:
10 abandoned/100 attempted = 10% abandon rate (this would be pretty high)
Average Speed of Answer (ASA) is the Number of Inbound Calls divided by their cumulative time to answer.
Service Level is the principal KPI for Call Centers, and is defined as a Percentage of Calls answered within a target Number of Seconds.
So, an 80/30 Service Level means 80% of incoming calls were answered within 30 seconds. Service Level is superior to Average Speed of Answer, because Abandoned calls immediately count as a call that was not answered within the target time.
This is important when you have periods of very high wait times (like in the example chart above) which drive abandoned calls way up for a short block of time. ASA masks this crummy performance several ways, but most importantly by dinging the Call Center for missing a potential customer contact.
Some public agencies (for example those whose initials include the letters I, R, and S) use long wait times and high abandoned calls to weed out all but the desperate and determined.