So announce this when they first enter.... once and only once. That sounds like a reasonable idea.
bkw On Fri, 5 Sep 2003, McAughan, Matt wrote: > There is one thing you have to look out for. Wait time is affected only by > the number of calls in front of you, not total calls, the number of agents > answering, and the length of calls. > > I say this because if you are going to update the announcer x seconds, > depending on the calculation the caller may experience, "your wait time is 5 > minutes", "your wait time is 25 minutes", "your wait time is 7 minutes". > That makes me want to hang up as well. On our (non Asterisk) phone system we > avoid this by just announcing the average wait time once, when the caller > enters the queue. > > I think the proper calculation should be a running average of time to answer > over the last X period, with a factor taking in to account average agents > logged in over that period. Something like average wait time per agent per > period. Then factor that against your current queue position (calls > entering behind you have no affect on your wait time) and current number of > agents (more could log in to help out or drop off). What the proper period > factor is I don't know. 30 seconds, 1 minute, after each hang up? Too > frequently it will fluctuate wildly. Too infrequent it will show residual > affects. > > -----Original Message----- > From: Steven Critchfield [mailto:[EMAIL PROTECTED] > Sent: Friday, September 05, 2003 02:25 > To: [EMAIL PROTECTED] > Subject: Re: [Asterisk-Users] app_queue input needed... > > > On Fri, 2003-09-05 at 14:05, Brian West wrote: > > A friend and I have recently added the ability to announce the callers > > position in the call queue every x seconds.. or even just inject an > > anouncement every x seconds. All setup in queues.conf and can be setup > > per queue. > > > > My next project is to add the ability to announce the callers estimated > > wait time. I want some feedback to see whats the best method to calculate > > that? What do you want just minutes? or minutes and seconds? Or the > > option to use one or the other? > > > > I'm thinking (totaltime / totalcalls) - (now - qe->start) = current > > estimated wait time. Which would update after each call is hungup. > > > I do not use queues, so accept my comments as only an opinion of how I > would like to experience them if I where a person in a queue. > > Your wait time is not very accurate unless you have sufficiently large > enough pools of people to service them to offset little abnormalities. > So I would say it would be good to define an acceptable list of > announcements, then round up to the first available announcement and > play from there. Have to look up be something like... > under 3 minutes > under 5 minutes > under 8 minutes > under 10 minutes > under 15 minutes > under 20 minutes > If I ever heard a time over 20 minutes I'd hang up and call back later, > or stop doing business with the company. This limits down your number of > prompts and lowers the expectation of wait time accuracy. > > -- > Steven Critchfield <[EMAIL PROTECTED]> > > _______________________________________________ > Asterisk-Users mailing list > [EMAIL PROTECTED] > http://lists.digium.com/mailman/listinfo/asterisk-users > _______________________________________________ Asterisk-Users mailing list [EMAIL PROTECTED] http://lists.digium.com/mailman/listinfo/asterisk-users
