Genesys Cloud - Main

Wondering if anyone else faces this challenge, for context, we are running a 12hr weekday operation for which our call service level is 80/60.

I am finding that even with 0% shrinkage applied, when generating a schedule based off a short term forecast for which we are using the Weighted Historical Index method and modifying to suit our needs, it is always scheduling a lot more staff than we really need.

It has got me thinking that Genesys Cloud is making sure that we are meeting our SLA per 15 minute interval as opposed to daily or weekly. Have I perhaps missed something in this process or is this as designed?
#Reporting/Analytics

------------------------------
Dylan Le Gros
Team Leader, Contact Centre Services | Aimia
------------------------------

We absolutely do attempt to staff correctly to hit service goals each 15m interval. Consistency of service performance is important. You don't want your customers to have radically different experience depending on if they call in the morning, mid-morning, afternoon, evening, Tue, Thu, etc.

There could be multiple factors in play here.

1. Forecast inaccuracy - specifically higher forecast volume and/or AHT than actual. Over-forecasting volume and/or AHT will raise staffing requirement and scheduled resources. I have seen many cases that simply disabling modifications results in much higher forecast accuracy - essentially a 'stock' forecast is better than humans making assumptions. Utilize the forecast viewer's/editor's built in forecast vs. actual information (hover over total offered and AHT actual stats in upper right on past forecasts; have show actuals enabled on display options).
2. Historically service performance is less than what is set as the goal. E.g., the service goal group is configured at 80%/20s, but actual service levels achieved are far less than 80% at the interval level up. I have seen many cases of folks stating a goal, but then admitting that they never hit that...but we staff for attempting to hit it. Utilize the intraday monitoring page and view past days.
3. Historically there is high abandonment rate with no corollary performance goal. E.g., 8, 9, 10% or more. Utilize the intraday monitoring page and view past days.
4. Historically there are staff not associated with the management unit that service some of the management unit's volume. In other words there are rogue agents unknown to the scheduling engine handling volume that is not accounted for fully (we attempt to model in operational efficiencies both good and bad) when determining staffing requirement and scheduling.

Also, consider using 'Automatic Best Method Selection' instead of 'Weighted Historical Index' method for forecasting.

------------------------------
Jay Langsford
Senior Director, Workforce Optimization Engineering
------------------------------

All valid points. In working in other WFM systems as well as Genesys I've always had to "adjust" the forecast (WFM) SLA to match reality. You see that when the system says you are understaffed by 5-7 agents but consistently hitting SLA. Or in other cases overstaffed by 5-7 agents and missing SLA. 

You have to look at this over a few week period and keep all other changes out of the process. Then I adjust my WFM SLA up or down to make sure when we are staffed at +/- 0 agents we hit our real SLA. For example real SLA goal is 80/120 but to get the correct staffing I set the WFM SLA to 80/160. It gives me better staffing results. Which is what I really want.

I have yet to work with any system that is 100% accurate out of the box with no tweaking. Which is fine. The system works. 


------------------------------
Greg Barrett
Outdoor Network, LLC
------------------------------

Original Message:
Sent: 03-31-2020 07:39
From: Jay Langsford
Subject: Scheduling Generation Accuracy

We absolutely do attempt to staff correctly to hit service goals each 15m interval. Consistency of service performance is important. You don't want your customers to have radically different experience depending on if they call in the morning, mid-morning, afternoon, evening, Tue, Thu, etc.

There could be multiple factors in play here.

1. Forecast inaccuracy - specifically higher forecast volume and/or AHT than actual. Over-forecasting volume and/or AHT will raise staffing requirement and scheduled resources. I have seen many cases that simply disabling modifications results in much higher forecast accuracy - essentially a 'stock' forecast is better than humans making assumptions. Utilize the forecast viewer's/editor's built in forecast vs. actual information (hover over total offered and AHT actual stats in upper right on past forecasts; have show actuals enabled on display options).
2. Historically service performance is less than what is set as the goal. E.g., the service goal group is configured at 80%/20s, but actual service levels achieved are far less than 80% at the interval level up. I have seen many cases of folks stating a goal, but then admitting that they never hit that...but we staff for attempting to hit it. Utilize the intraday monitoring page and view past days.
3. Historically there is high abandonment rate with no corollary performance goal. E.g., 8, 9, 10% or more. Utilize the intraday monitoring page and view past days.
4. Historically there are staff not associated with the management unit that service some of the management unit's volume. In other words there are rogue agents unknown to the scheduling engine handling volume that is not accounted for fully (we attempt to model in operational efficiencies both good and bad) when determining staffing requirement and scheduling.

Also, consider using 'Automatic Best Method Selection' instead of 'Weighted Historical Index' method for forecasting.

------------------------------
Jay Langsford
Senior Director, Workforce Optimization Engineering
------------------------------