Jay maybe the ability to select a "primary" queue/skill for agents where we have small teams that have to be in multiple queues/skills even if they don't use them all the time. The forecast/schedule could calc based on the Primary. Just a thought for the future.
Original Message:
Sent: 06-08-2023 10:36
From: Jay Langsford
Subject: Belonging to multiple Planning Groups does not drive efficiency
I agree with what Paul had said: "We usually recommend 3 months of historical data before the models are able to tune to your environment and for you to see more reflective staffing requirements, in some environments we can see improvements within as little as 3-4 weeks."
It could be 2-4 weeks is reasonably good. You could run different types of forecasts and compare overall forecast error (without modifications) and then make a decision on when to switch to automatic best method. We have and will continue to look at ways to help with the cold or semi-cold states (new or just recently new route paths in terms of historical data). To properly train the models there just needs to be a threshold.
If after a few weeks you still see low forecast accuracy (e.g., < 90% without modifications), I would recommend opening a support ticket so we can investigate.
For staffing requirements, forecast is a big piece; however, there are multiple things that attribute to either a higher or lower than expected staffing requirement. The following resource article I wrote covers many of the items: https://help.mypurecloud.com/faqs/why-are-staffing-requirements-higher-or-lower-than-expected/
------------------------------
Jay Langsford
VP, R&D
Original Message:
Sent: 06-08-2023 09:33
From: Melanie Williams
Subject: Belonging to multiple Planning Groups does not drive efficiency
Thanks @Jay Langsford . How long do we need to allow for the system to gather data in the route paths in the planning groups in order for it to calculate staffing more accurately.
------------------------------
Melanie Williams
Camis Inc
Original Message:
Sent: 06-08-2023 07:49
From: Jay Langsford
Subject: Belonging to multiple Planning Groups does not drive efficiency
New or changed route paths, which are part of the planning groups, are what needs time to have actual historical aggregated data collected by Analytics in order for WFM to more accurately forecast. When we query for historical data we are querying for the individual route paths that have been configured.
As far as bullseye routing, my recommendation is to not use it for queues you want to accurately forecast and staff for. Instead use skills-based routing. Bullseye routing just does not mix well with WFM - a big knock is that simple skills-based routing would route to an available agent whereas bullseye can potentially wait at a ring, or multiple rings, for the entire configured duration unnecessarily. The other issue with bullseye is there simply is no way to tell in aggregate analytics data what the final skill requirements were - only the original set is known.
Conditional routing would be a choice, when available, that would be more WFM friendly than bullseye that would give you some capabilities above stock skills-based routing: https://help.mypurecloud.com/articles/conditional-group-routing-overview/
------------------------------
Jay Langsford
VP, R&D
Original Message:
Sent: 06-07-2023 11:44
From: Melanie Williams
Subject: Belonging to multiple Planning Groups does not drive efficiency
Thanks @Paul Wood
We have 4.5 years worth of call data. Which historical data are you referring to? Historical data stored in the planning groups or call data?
It seems that with using the bulls eye routing - the system is projecting that we are understaffed since it can't tell that we are routing more efficiently. Therefore it shows we are understaffed with a low SL prediction (usually about 2-3FTE off). Does your team have any suggestions for work arounds?
Thanks!
-Melanie
------------------------------
Melanie Williams
Camis Inc
Original Message:
Sent: 06-07-2023 11:37
From: Paul Wood
Subject: Belonging to multiple Planning Groups does not drive efficiency
Hi @Melanie Williams
we usually recommend 3 months of historical data before the models are able to tune to your environment and for you to see more reflective staffing requirements, in some environments we can see improvements within as little as 3-4 weeks.
------------------------------
Paul Wood
Product Manager for Genesys Cloud Workforce Management
Original Message:
Sent: 06-06-2023 14:16
From: Melanie Williams
Subject: Belonging to multiple Planning Groups does not drive efficiency
Hi @Paul Wood ,
How long would we need to build the appropriate amount of data to have accurate predictions? We are in a similar boat where Genesys doesn't seem to account for cross training across planning groups - so it doesn't account for efficiencies in staffing resulting in us being overstaffed.
I will note that we are also using the bullseye routing method, which according to the resource center is not compatible with the forecasting module.
Thoughts?
-Melanie
------------------------------
Melanie Williams
Camis Inc
Original Message:
Sent: 05-25-2023 11:19
From: Paul Wood
Subject: Belonging to multiple Planning Groups does not drive efficiency
Hi Felicity,
when balancing/blending requirements over multiple Planning Groups. Genesys uses a complex Contact Centre model that uses both historical demand and performance data:
-Interaction Volumes
-Average Handle Times
-Average Speed of Answer
-Abandonment rates
-Agent time
-Available time
-Shrinkage
when combining the Contact Centre model with our simulation models we can more accurately forecast demand, staffing requirements and resulting service goal predictions.
Depending on your level of historical data, we may not have access to all of the performance data required to build and refine these models. In these instances, we fall back to a more Erlang-based methodology, which would explain your current results.
As your historical data builds the models will learn and improve, more accurately reflecting what is actually happening within your environment.
------------------------------
Paul Wood
Product Manager for Genesys Cloud Workforce Management
Original Message:
Sent: 05-23-2023 04:20
From: Felicity Martin-Murray
Subject: Belonging to multiple Planning Groups does not drive efficiency
We are moving to Genesys Cloud CX in the autumn and we're currently designing our Planning Groups. Currently we think we will have 1 Queue, containing 5 Planning Groups, each containing 1 skill. Every agent will belong to all 5 Planning Groups as they can all do all 5 skills. We create separate forecasts for each of the 5 skills and have been advised to keep our Planning Groups as granular as possible.
When capacity planning (in Excel) we add up all of the volumes for all 5 skills and create a weighted AHT. We then use Erlang to calculate the FTE requirement by week and by interval.
We have built the 5 Planning Groups in our Genesys test environment and found that the required FTE is significantly higher than our Excel models output, to the tune of about 20% across all intervals. We have not added any shrinkage into the work plans, or management unit settings, and there is no shrinkage in our Excel models either.
However, if, in Excel, we calculate the requirements for each skill separately and then add them all together, we get the exact same output as Genesys. So obviously Genesys is just applying a basic Erlang calculation to each of the Planning Groups separately and then adding them up. It is not accounting for any multi-skilling efficiencies despite the fact that ALL of our agents can take ALL of the skills.
Has anyone else found this and how did you get around this? We are currently thinking that we will need to reduce to just 1 Planning Group with all 5 skills in it, but this goes against the advice Genesys provide and will mean that other areas of the business need to change their ways of working.
#Genesys Cloud CX
#WorkforceManagement
#Forecasting
#Scheduling
------------------------------
Felicity Martin-Murray
J Sainsbury Plc
------------------------------