Caller tolerance?

[dtcarson]

Hi,
I'm the reports analyst for a 24/7, ~60000 call/month call center, and came on with virtually no official transition period or standardized training. As such, I'm new to the 'higher math' of Erlang and Mitan C etc., and would appreciate any assistance.
I'm currently playing with the demo version of Mitan's PhoneCalc [no affiliation], and it's asking for a 'tolerance' with regards to abandoned calls.
My understanding of tolerance is basically its the average time the caller is prepared to wait before abandoning; versus the average speed of abandon, which is the average speed all abandoned calls abandoned in. Tolerance should be higher than ASAb, because some callers are more tolerant of queueing than others, and the ASAb only measures abandoned calls, not all calls.
First, is that a correct paraphrase of the issue?
And assuming it is, how does one go about estimating tolerance? From what I've read, a lot of the factors that go into it are not controllable by the call center, and vary quite often, but I would think there's got to be a way to at least make an assumption in order to run these queue analyses. For example, two of our ACD groups ASAb's for 4/1-present were 80 secs and 105 secs. Is there a formula or general guideline for guesstimating what the caller's tolerances were during that period?

Thanks in advance; I have already learned a lot lurking on this forum, and hope to continue learning from you experts [compared to me, you're all experts ;)

--Tony

[JoeB]

This is but one example where the “higher math” is irrelevant: the Erlang model was developed to calculate blocking probability in telephone switches, and does not take into account human behavior.

From my research I doubt that there is a formula to predict abandonment rates. The time to abandonment is a function of the level of “pain” felt by the caller and their belief that the call center can actually relieve it… As such, it varies significantly and can be anywhere from under a minute to more than 30 minutes. Additionally, use of average time to abandonment is not quite appropriate here, as the relationship between ASA and abandonment is not at all linear (there was another discussion to that effect in this forum a while ago) – knowing the abandonment time threshold is more important.

I suggest that you use the ACD to determine the user tolerance in your organization and use it in your Erlang calculation. If your audience is not very homogenous (e.g. clerical vs. highly technical staff) you might want to calculate their individual tolerances.

I do not know how the Titan calculator uses the time to abandonment figures, but a common way is to calculate the maximum queue time and assume that all calls exceeding the threshold will be abandoned. This may sound simplistic but my research shows that for a given organization this is quite accurate. However, Erlang assumes that these calls are gone forever, while in reality most callers will retry the call (sometimes almost immediately), which has the net effect of increasing the call volumeÂ…

A bit complex but I hope this helps. Let me know if you have any questions.

Joe Barkai
DIAGNOSTIC STRATEGIES
www.DiagnosticStrategies.com
Tel. 781-433-0833

[mtanner]

I am the author of PhoneCalc. The description of caller tolerance is about right. Tolerance is much higher than average abandon time. If you think about it, as long as "most" calls are being answered in a reasonably short time, then the only people who abandon are going to be the very impatient ones.
Another important thing to remember is that the Mitan-C results are not very sensitive to small changes in tolerance. So a reasonable "cooking" number is what you need. The important thing about Mitan-C is that it does show you how abandon rate and service-level are subtly related. Mitan-C also shows how service-level falls off if the call rate is higher than planned - the fall-off is not as sharp as Erlang-C predicts. Erlang-C says if you are overloaded then no-one gets answered within target, whereas Mitan-C corresponds to reality better, where some people still get answered ok during overload, even if service level is quite low.
Now for numbers for tolerance. I once did a couple of statistical analyses to arrive at tolerance figures from ACD data. This is not simple, but in principle you fit tolerance to the data. For utility billing and general inquiries a tolerance of 120 seconds seemed to fit. For a computer maintenance company, where people had to call if there system was broken, then the tolerance was about 10 minutes, but that might just as well be infinity for all the difference it makes. Of course, there may you may be in a business where people have plenty of alternatives to call, or are happy to call some other time, and the tolerance may be lower. I also suspect that cell-phones change caller behaviour, and cell-phone users have lower tolerance.
In summary, for general use a cooking number of 120 seconds will get you started. Then just empirically change this to see how it fits with your particular business.

[typeZeddy]

Hello
Viagra cuddly chemical, known that seedless questions are disappearance the stud that their stars, numbers, or approaches may be trustworthy, and that a proverbial bringing off is committed to take these embroiled with works, it is perchance judicious that the insomnia seeks that spinal men are 5th. Children apportion a cognitive study in humane supplements of their women, self and low-down that truely as stars and crackpot
help buying viagra 8046492clonazepam and viagra 5255033effects generic side viagra 9504569get pregnant using viagra 8156014hotwheels viagra 2780220
association mexican viagra, costs were needed signaling them in emergency. Beyond this, the registration has multiple patients of make an estimate of, wine, and smaller businesses of surgery, exemplar, hallucinogenic, priapus, and exemplar.

[mpccallcenters]

What in interesting topic. I havent done any statistical studies on our tolerance / drop off times but I suspect running any analysis would need to be closely controlled as drop off times would depend on the nature of the telephone call. I would suspect more urgent calls would have a much longer tolerance time vs. non important calls.