Workforce Engagement Management

Hi guys

In Genesys Cloud - how have you approached a forecast/schedule build when the hours for a contact centre have changed?  ie were originally 9am-9pm but are now only 9am-5pm.

Forecasts include historical calls that came in between 5pm and 9pm but those calls will no longer be relevant going forward as the contact centre is now closed during those times.

Doing a manual modification on each forecast seems a painful way of dealing with it, and I guess maybe another way is to just leave those forecast calls in the forecast but end up with them being non-staffed due to workplan etc constraints only scheduling to 5pm anyway?
#Workforce Engagement Management
#Workforce Management

------------------------------
Vaun McCarthy
------------------------------

Hi Vaun - This is not something I've done on this platform but is something I've had to do in past roles.  I think there is a dependency on understanding what type of caller you have. For example, do you have a sticky caller that has to contact you, or do they have options to call a competitor.

For a sticky caller, like a technical support interaction or billing, you might expect the overall volume of daily/weekly calls to stay the same, just truncated into fewer work intervals.  So you could just proportionally inflate the remaining logging periods volume forecast as to balance overall volume. Anyway you slice it, it seems like a manual modification on each forecast is the name of the game.

For a non-sticky caller, like sales interaction, you might just need to end up with the previous open hours being forecasted but non-staffed.

------------------------------
Raymond Hicks
Sutter Health
------------------------------

Thanks Raymond

------------------------------
Vaun McCarthy
------------------------------

Original Message:
Sent: 11-11-2021 11:46
From: Raymond Hicks
Subject: Forecast after hours have changed?

Hi Vaun - This is not something I've done on this platform but is something I've had to do in past roles.  I think there is a dependency on understanding what type of caller you have. For example, do you have a sticky caller that has to contact you, or do they have options to call a competitor.

For a sticky caller, like a technical support interaction or billing, you might expect the overall volume of daily/weekly calls to stay the same, just truncated into fewer work intervals.  So you could just proportionally inflate the remaining logging periods volume forecast as to balance overall volume. Anyway you slice it, it seems like a manual modification on each forecast is the name of the game.

For a non-sticky caller, like sales interaction, you might just need to end up with the previous open hours being forecasted but non-staffed.

------------------------------
Raymond Hicks
Sutter Health
------------------------------

Afternoon Vaun, 

If you are using Automatic Best Method it will take about 90 days for that existing distribution volume to be removed from the AI model that is designed for that individual Planning Group. Its 90 days because we are managing the calculation of distribution variation across a wider calendar period, any quicker and can see spikes form in the forecast when you have actual spikes. Data that's sparse (random 2am test call for example) is automatically cleaned up. This change in hours would trigger that historical data to also be considered sparse however over that 90 day time period and then be ignored. 

This can be tricky though as we don't want to be overly aggressive and learn from previous week (maybe you had a holiday, or special black friday hours or other event) so we have to have sufficient enough data to say, yes this is a permanent change. 

There are obviously a few items to consider:

1. Does our volume simply compress now we are open for shorter hours?
2. Does our overall distribution pattern change? 

In both cases the answer is most likely yes however with no history to go on the human 'art' does need to make its way back into the AI Algorithms - some decisions will need to be made. So there is a motion where you would still need to manually adjust in the short-term until that new pattern is formed. 

You have highlighted a use case that we have noted now in our backlog which is to give the forecast user a modification around operating hours. One thing to note is the reverse scenario is accommodated immediately (say you expanded hours) as the models pick up that new Analytics traffic instantly. 

------------------------------
Cameron Smith
VP, Product Management - Workforce Engagement Management

cameron.smith@genesys.com
------------------------------

Thanks Cameron, is there a resource centre article that covers what you've explain here around the 90 days pattern, and treating random low volume calls as sparse and ignored?

------------------------------
Vaun McCarthy
------------------------------

Original Message:
Sent: 11-11-2021 14:52
From: Cameron Smith
Subject: Forecast after hours have changed?

Afternoon Vaun,

If you are using Automatic Best Method it will take about 90 days for that existing distribution volume to be removed from the AI model that is designed for that individual Planning Group. Its 90 days because we are managing the calculation of distribution variation across a wider calendar period, any quicker and can see spikes form in the forecast when you have actual spikes. Data that's sparse (random 2am test call for example) is automatically cleaned up. This change in hours would trigger that historical data to also be considered sparse however over that 90 day time period and then be ignored.

This can be tricky though as we don't want to be overly aggressive and learn from previous week (maybe you had a holiday, or special black friday hours or other event) so we have to have sufficient enough data to say, yes this is a permanent change. 

There are obviously a few items to consider:

1. Does our volume simply compress now we are open for shorter hours?
2. Does our overall distribution pattern change? 

In both cases the answer is most likely yes however with no history to go on the human 'art' does need to make its way back into the AI Algorithms - some decisions will need to be made. So there is a motion where you would still need to manually adjust in the short-term until that new pattern is formed.

You have highlighted a use case that we have noted now in our backlog which is to give the forecast user a modification around operating hours. One thing to note is the reverse scenario is accommodated immediately (say you expanded hours) as the models pick up that new Analytics traffic instantly. 

------------------------------
Cameron Smith
VP, Product Management - Workforce Engagement Management

cameron.smith@genesys.com
------------------------------

Not to that level of granularity unfortunately Vaun as its more how the process/algorithms function which gets fairly technical pretty quickly.

Is this something of general interest? We are looking to improve our documentation in this area and will add this to that feedback.

------------------------------
Cameron Smith
VP, Product Management - Workforce Engagement Management

cameron.smith@genesys.com
------------------------------

Original Message:
Sent: 11-12-2021 15:32
From: Vaun McCarthy
Subject: Forecast after hours have changed?

Thanks Cameron, is there a resource centre article that covers what you've explain here around the 90 days pattern, and treating random low volume calls as sparse and ignored?

------------------------------
Vaun McCarthy
------------------------------


Original Message:
Sent: 11-11-2021 14:52
From: Cameron Smith
Subject: Forecast after hours have changed?

Afternoon Vaun,

If you are using Automatic Best Method it will take about 90 days for that existing distribution volume to be removed from the AI model that is designed for that individual Planning Group. Its 90 days because we are managing the calculation of distribution variation across a wider calendar period, any quicker and can see spikes form in the forecast when you have actual spikes. Data that's sparse (random 2am test call for example) is automatically cleaned up. This change in hours would trigger that historical data to also be considered sparse however over that 90 day time period and then be ignored.

This can be tricky though as we don't want to be overly aggressive and learn from previous week (maybe you had a holiday, or special black friday hours or other event) so we have to have sufficient enough data to say, yes this is a permanent change. 

There are obviously a few items to consider:

1. Does our volume simply compress now we are open for shorter hours?
2. Does our overall distribution pattern change? 

In both cases the answer is most likely yes however with no history to go on the human 'art' does need to make its way back into the AI Algorithms - some decisions will need to be made. So there is a motion where you would still need to manually adjust in the short-term until that new pattern is formed.

You have highlighted a use case that we have noted now in our backlog which is to give the forecast user a modification around operating hours. One thing to note is the reverse scenario is accommodated immediately (say you expanded hours) as the models pick up that new Analytics traffic instantly. 

------------------------------
Cameron Smith
VP, Product Management - Workforce Engagement Management

cameron.smith@genesys.com