What SL Does 45-Second ASA Represent?

Answers

• Braheem Stafford Posted at 2:05PM on Jul 15, 2011

  Well said Brent... you actually answered my question inside your response...lol

• Rose Polchin Posted at 9:25AM on Jul 25, 2011

  I second Braheem, "Well said Brent!" The only other additional piece of information I want to offer is that even though both service level and ASA are calculated off the data set, they are two different calculations. Service level is "x% answered in Y seconds" and ASA is total delay of ALL calls (regardless of whether they were answered in zero seconds or waited) divided by calls answered. Often folks outside the contact center believe the "seconds" piece of service level should be what the ASA is...and Brent does an excellent job of explaining why a) that won't be the case and b) why there isn't a "pat" formula to calculate what ASA will be given a certain service level.

• Zeynep Biricik Posted at 11:22AM on Jan 27, 2012

  Hi all, I have the same problem in our callcenter. My manager wants me to calculate how many agents is needed to decrease our average speed of answer 1 seconds. I tried to make some simulations by using previous day's values but there are some many additional factors that I cannot find the same values as occured. And also I realised that erlang formula calculates occupancy rates above 100% and service levels below 0 which is impossible. Is there anyone who could realised this problem and knows any solution? Thanks Zeynep Biricik Özak

• Brent Haferkamp Posted at 12:00AM on Aug 11, 2008

  There really isn't a decoder ring to convert ASA [Average Speed of Answer] to SL [Service Level] or SL to ASA. Many factors determine this, including Calls Per Half Hour, AHT [Average Handle Time], the "normalness" of the ASA curve/call arrival pattern, etc. The example below shows the results of altering only Calls Per Half Hour and AHT: Calls AHT ASA Agents XX/10 XX/20 XX/30 290 120 45 21 46% 53% 59% 1080 120 45 74 37% 47% 55% 35 180 45 5 65% 68% 71% 190 240 45 28 55% 60% 64% 380 300 45 67 52% 57% 62% 930 360 45 191 46% 53% 59% A 45 second ASA in this example can range from 68/20 SL all the way down to a 47/20 SL. In other words, a 45 second ASA in an off-peak interval will more than likely result in a much different SL that the same ASA at a peak volume interval. You could try to narrow it down by saying that your volume and AHT normally fall in a specific range and calculate an average for that range, but it will still be a very rough estimate, as the pattern of arrival (thus pattern of individual call speed to answer) will also have an impact. For example, if all of your calls are answered in exactly 45 seconds, you have a 0/20 SL, but if 1/2 are answered in exactly 20 seconds and the other 1/2 are answered in exactly 70 seconds, you would have a 50/20 at the same 45 ASA. Both examples and the same volume, AHT and ASA, just different arrival pattern, creating different SL. These two examples are extreme and would never happen in real life, but they demonstrate the effect of arrival pattern on attempting to correlate ASA and SL. If you are interested in knowing what your SL is, you are better off, in the long run, to begin tracking actual service level and not attempting to correlate ASA to SL. -- Brent Haferkamp, Sr. Consultant ICMI