Overdial rate during lunch time (predictive mode)

[8isim]

Hello! I observe very high overdial rate during lunch time because dialer doesn't expect that half of agents leave.
What are possible solutions to decrease the overdial rate?
I don't want to use a regular limit for overdial rate because overdial rate is calculated since the beginning of campaign and in my situation I will have to set quite low limit which will speed down all campaign during all time.
My ideas which I didn't find how to implement it:
1. Limit of overdial rate calculated for last 20-30 minutes: since there are many abondened calls only during lunch, I will can recognize it by abondan rate during last 20 minutes.
2. Dialer understands from Work Force Manager that half of agents have a lunch in 5 minutes and automatically decease the speed of dialing.

Any ideas are welcome.
Thank you.

[Tambo]

one solution is to stop dialling completely and get all agents to take lunch at same time. A few companies i know do this as they have a short dialling day (less than 10 hrs)

[cavagnaro]

Or by schedules at that time change to Progressive
Then, switch again to Predictive after 30 min or so

Or some super do a manual adjustment, like a Stop or slow down on campaign settings so those agents can go away.
Other is to request a OCS Logout and go to lunch only when the OK was received from OCS.

[Fra]

Does it happen also if you use the time_optimized_odr algorithm?
I also support Cav's idea: OCS negotiated logout protocol is designed to cater for similar scenarios.

[8isim]

Thank you for your replies.
We will try to run a pilot with a switch from Predictive to Progressive mode when the number of agents decreases.

[cavagnaro]

Yup, that will help a lot

Enviado de meu E6633 usando Tapatalk

[8isim]

If you have any other suggestions on how to decrease abandoned rate and keep or decrease the average waiting time, feel free to share
Maybe some parameters in Genesys to optimize. Or any other advise.
FYI: we have a call center with ~300 agents working with one call list per day.

[cavagnaro]

All depends in your filters and dialing methods settings. What does abandon means? You are dialing too much, which means you expect this as your main goal is to keep agent occupancy and not customer being reached necessarily.
So it is not a setting, is a management and adjusts during the campaign itself.
Good campaigns change dialing filters constantly based on customer phone numbers type, control agent behavior, set rules for not ready, manage dialing methods and adjust the level of each predictive, one 100% assured way to increase abandon is for example to put the agent busy factor dialing method to 100% just because in a few minutes there were no records for some agents. What does it means? Means you are dialing to the wrong places, home at 2pm? Rather than the wrong dialing speed.

Enviado de meu E6633 usando Tapatalk

[8isim]

Guys, thank you very much, especially cavagnaro.

Predictive more really helped a lot when number of agents sharply decreased.
For sharp increase of agents we already tried to restart the dialer and to do nothing but in both options it takes quite a lot of time until utilization of agents reaches normal level.
We are going to try to switch to Progressive during increase of agents as well. What do you think? Will it help? Any other ideas?

[Dionysis]

Here are a few ways to generally improve outbound.  Your specific situation will be different, but these apply everywhere.


1 - If you dial one list each day, are there repeats the following day?  If so, I'd bet it's in the same order each day, which means you're redialling the same number at approximately the same time, every day.  This causes very low list completion rates, randomise the record_id field each day when you load the data.

2 - Stop using Overdial Rate.  Agent Busy factor is far superior, I'd recommend setting it to ~80%.  Any higher and you will abandon lots of calls, and burn out your agents.

3 - Tune your buffer settings.  Each list has configuration for minimum and optimal buffer sizes, these need to be higher than the defaults.  I can't tell you what you should set them to, but higher than the default is almost always better.

4 - Make sure your record type priorities are correct.  Campaign Callback must be highest priority, followed by Campaign Rescheduled, then Fresh records.  New data should never be first, I can't count the number of times I've found call centres doing this and they have great hit rates for a while, then have nothing but old data left to dial and end up with huge wait times.

5 - Use negotiated logout.  Nothing improves abandonment rate more, and make sure your agents do what they're supposed to do and keep taking calls until the logout has been processed.  I've seen agents hit logout, and walk away, then you end up with potential sales speaking to dead air.

6 - Most important, tune your call progress detection settings regularly.  Depending on what you use for call progress analysis, you can get huge benefits from running some tuning a couple of times per year.  Defaults in this area are terrible, especially if you're using Dialogic or Genesys media server.  By default you wont get better than 80% accuracy on home phones (the most accurate), the environment I manage currently got 92% on mobile phones last time we tuned it (Mobile is generally the least accurate). 

[8isim]

[quote author=Dionysis link=topic=9974.msg45355#msg45355 date=1480715491]
Here are a few ways to generally improve outbound.  Your specific situation will be different, but these apply everywhere.


1 - If you dial one list each day, are there repeats the following day?  If so, I'd bet it's in the same order each day, which means you're redialling the same number at approximately the same time, every day.  This causes very low list completion rates, randomise the record_id field each day when you load the data.

2 - Stop using Overdial Rate.  Agent Busy factor is far superior, I'd recommend setting it to ~80%.  Any higher and you will abandon lots of calls, and burn out your agents.

3 - Tune your buffer settings.  Each list has configuration for minimum and optimal buffer sizes, these need to be higher than the defaults.  I can't tell you what you should set them to, but higher than the default is almost always better.

4 - Make sure your record type priorities are correct.  Campaign Callback must be highest priority, followed by Campaign Rescheduled, then Fresh records.  New data should never be first, I can't count the number of times I've found call centres doing this and they have great hit rates for a while, then have nothing but old data left to dial and end up with huge wait times.

5 - Use negotiated logout.  Nothing improves abandonment rate more, and make sure your agents do what they're supposed to do and keep taking calls until the logout has been processed.  I've seen agents hit logout, and walk away, then you end up with potential sales speaking to dead air.

6 - Most important, tune your call progress detection settings regularly.  Depending on what you use for call progress analysis, you can get huge benefits from running some tuning a couple of times per year.  Defaults in this area are terrible, especially if you're using Dialogic or Genesys media server.  By default you wont get better than 80% accuracy on home phones (the most accurate), the environment I manage currently got 92% on mobile phones last time we tuned it (Mobile is generally the least accurate).
[/quote]

Thank you very much for valuable advises. We have thought about some of them but didn't expect that they are so important.
I'd like to ask few question about your points:
point 2: why is it so? why Overdial rate is worse than ABF? We are currently using ABF and thinking to switch to Overdial rate.
point 5: is it Smart logout in Deployment guide? We are currently pushing agents to use "Now Ready" statuses during lunch and breaks and not to logout. So Smart logout will be applicable only for the time when agent is going home. Should we ask agents to logout during lunch to use Smart logout? Which option (Not Ready status or logout) better predicts/ behaves at the time when agents are coming back from lunch? Or in both cases it is better to switch to Progresive and performance will be the same?
point 6: I will definetely investigate it. So I'll have questions later
Thank you!