Workforce Engagement Management

Hello Linsey, 

It can at first sight looks inflated, but it's actually consistent with how queueing models behave under low volume.

Staffing is calculated per interval, not averaged.

Genesys uses queueing theory to meet your service goal in every interval.

With:

~0.32 calls per 30 minutes
A 60-second ASA target

The system isn't thinking "7 calls per day." It's thinking:

"What staffing ensures I can answer any call that might arrive right now within 60 seconds?"

At such low volume:

Many intervals = 0 calls
Some = 1 call
Occasionally = 2 calls close together

Even though the average is tiny, the variance is huge relative to the mean.

To protect your ASA target, the model prepares for:

"What if a second call arrives while the first is still in progress?"

Your key difference:

Queue A - ~7 min AHT
Queue B - ~9 min AHT

That ~2-minute increase significantly raises the chance that:

Call #2 arrives while Call #1 is still active

So the model reacts by increasing staffing toward having:

"someone always available just in case"

That's why Queue B jumps toward ~0.7 FTE, even though workload is still tiny.

A 60s ASA at this volume is very strict.

At low volume, hitting that target often requires:

An agent being idle much of the time because even one delayed call can break service level.

So staffing shifts from:

"Cover the workload" to "Guarantee immediate availability"

It inflates your total FTE because many small, separate queues each require their own "just-in-case" capacity.

So instead of:

1 agent handling all low-volume work you get 0.5–0.8 FTE per queue - which adds up artificially.

So the reason Queue B is so much higher than Queue A, even with similar total volume is:

Higher AHT
Same strict ASA
Same low arrival rate

This leads to a much higher probability of overlap, which forces staffing up disproportionately.

I hope this helps you.

Hi Camila, that helps a lot. Thank you for the explanation. I'm thinking of increase the service goal to something like 240 seconds, but I suspect that won't make a big difference on a low-volume queue. What do you recommend in this situation? We have the separate planning groups because it requires a separate agent training.

Hello Linsey, 

It can at first sight looks inflated, but it's actually consistent with how queueing models behave under low volume.

Staffing is calculated per interval, not averaged.

Genesys uses queueing theory to meet your service goal in every interval.

With:

~0.32 calls per 30 minutes
A 60-second ASA target

The system isn't thinking "7 calls per day." It's thinking:

"What staffing ensures I can answer any call that might arrive right now within 60 seconds?"

At such low volume:

Many intervals = 0 calls
Some = 1 call
Occasionally = 2 calls close together

Even though the average is tiny, the variance is huge relative to the mean.

To protect your ASA target, the model prepares for:

"What if a second call arrives while the first is still in progress?"

Your key difference:

Queue A - ~7 min AHT
Queue B - ~9 min AHT

That ~2-minute increase significantly raises the chance that:

Call #2 arrives while Call #1 is still active

So the model reacts by increasing staffing toward having:

"someone always available just in case"

That's why Queue B jumps toward ~0.7 FTE, even though workload is still tiny.

A 60s ASA at this volume is very strict.

At low volume, hitting that target often requires:

An agent being idle much of the time because even one delayed call can break service level.

So staffing shifts from:

"Cover the workload" to "Guarantee immediate availability"

It inflates your total FTE because many small, separate queues each require their own "just-in-case" capacity.

So instead of:

1 agent handling all low-volume work you get 0.5–0.8 FTE per queue - which adds up artificially.

So the reason Queue B is so much higher than Queue A, even with similar total volume is:

Higher AHT
Same strict ASA
Same low arrival rate

This leads to a much higher probability of overlap, which forces staffing up disproportionately.

I hope this helps you.

Hi Linsey,

What sometimes helps in our case is to reforecast until Genesys returns a closer to reality Staffing requirement. We have raised several cased to Genesys support and the only "solution" we received so far is indeed to reforecast and change the Service goal templates.

Regards,

Daniela

Hi Camila, that helps a lot. Thank you for the explanation. I'm thinking of increase the service goal to something like 240 seconds, but I suspect that won't make a big difference on a low-volume queue. What do you recommend in this situation? We have the separate planning groups because it requires a separate agent training.

Hello Linsey, 

It can at first sight looks inflated, but it's actually consistent with how queueing models behave under low volume.

Staffing is calculated per interval, not averaged.

Genesys uses queueing theory to meet your service goal in every interval.

With:

~0.32 calls per 30 minutes
A 60-second ASA target

The system isn't thinking "7 calls per day." It's thinking:

"What staffing ensures I can answer any call that might arrive right now within 60 seconds?"

At such low volume:

Many intervals = 0 calls
Some = 1 call
Occasionally = 2 calls close together

Even though the average is tiny, the variance is huge relative to the mean.

To protect your ASA target, the model prepares for:

"What if a second call arrives while the first is still in progress?"

Your key difference:

Queue A - ~7 min AHT
Queue B - ~9 min AHT

That ~2-minute increase significantly raises the chance that:

Call #2 arrives while Call #1 is still active

So the model reacts by increasing staffing toward having:

"someone always available just in case"

That's why Queue B jumps toward ~0.7 FTE, even though workload is still tiny.

A 60s ASA at this volume is very strict.

At low volume, hitting that target often requires:

An agent being idle much of the time because even one delayed call can break service level.

So staffing shifts from:

"Cover the workload" to "Guarantee immediate availability"

It inflates your total FTE because many small, separate queues each require their own "just-in-case" capacity.

So instead of:

1 agent handling all low-volume work you get 0.5–0.8 FTE per queue - which adds up artificially.

So the reason Queue B is so much higher than Queue A, even with similar total volume is:

Higher AHT
Same strict ASA
Same low arrival rate

This leads to a much higher probability of overlap, which forces staffing up disproportionately.

I hope this helps you.

No forecasting methodology is going to do well if a time series has 7 data points in a 24h period. The "small numbers" issue in forecasting is a real thing. Fractured routing config (many queues, many skills) ends up negating economies of scale type benefits. The more homogenous your load and resources the better, the more heterogenous those are the more perceived inflation of staffing requirements.

10 interactions per 30m for a planning group may still be low, but a reasonable floor. 7 interactions per day...why even forecast or staff as it is in the noise and likely handled by whatever base staff (minimum paid / on queue hours) scheduled? Very unlikely there is any predictability with volumes that low. Is 1 FTE enough? What if the 7 interactions occur over 15h? What if they all happen to land in the same 15m window? What if that 15m window is in the morning one week, but the next it drifts to the evening?

Also, a correction to something said in a prior reply: we only use queueing theory when deferred work (e.g., emails) is in the mix.

Hi Linsey,

What sometimes helps in our case is to reforecast until Genesys returns a closer to reality Staffing requirement. We have raised several cased to Genesys support and the only "solution" we received so far is indeed to reforecast and change the Service goal templates.

Regards,

Daniela

Hi Camila, that helps a lot. Thank you for the explanation. I'm thinking of increase the service goal to something like 240 seconds, but I suspect that won't make a big difference on a low-volume queue. What do you recommend in this situation? We have the separate planning groups because it requires a separate agent training.

Hello Linsey, 

It can at first sight looks inflated, but it's actually consistent with how queueing models behave under low volume.

Staffing is calculated per interval, not averaged.

Genesys uses queueing theory to meet your service goal in every interval.

With:

~0.32 calls per 30 minutes
A 60-second ASA target

The system isn't thinking "7 calls per day." It's thinking:

"What staffing ensures I can answer any call that might arrive right now within 60 seconds?"

At such low volume:

Many intervals = 0 calls
Some = 1 call
Occasionally = 2 calls close together

Even though the average is tiny, the variance is huge relative to the mean.

To protect your ASA target, the model prepares for:

"What if a second call arrives while the first is still in progress?"

Your key difference:

Queue A - ~7 min AHT
Queue B - ~9 min AHT

That ~2-minute increase significantly raises the chance that:

Call #2 arrives while Call #1 is still active

So the model reacts by increasing staffing toward having:

"someone always available just in case"

That's why Queue B jumps toward ~0.7 FTE, even though workload is still tiny.

A 60s ASA at this volume is very strict.

At low volume, hitting that target often requires:

An agent being idle much of the time because even one delayed call can break service level.

So staffing shifts from:

"Cover the workload" to "Guarantee immediate availability"

It inflates your total FTE because many small, separate queues each require their own "just-in-case" capacity.

So instead of:

1 agent handling all low-volume work you get 0.5–0.8 FTE per queue - which adds up artificially.

So the reason Queue B is so much higher than Queue A, even with similar total volume is:

Higher AHT
Same strict ASA
Same low arrival rate

This leads to a much higher probability of overlap, which forces staffing up disproportionately.

I hope this helps you.

Actually, the forecasting/staffing requirement in Genesys Engage is very reliable. It will return is 0.5 FTEs if that is the case for very low volumes.

So, there are other forecasting methodologies doing very well on this part. 

No forecasting methodology is going to do well if a time series has 7 data points in a 24h period. The "small numbers" issue in forecasting is a real thing. Fractured routing config (many queues, many skills) ends up negating economies of scale type benefits. The more homogenous your load and resources the better, the more heterogenous those are the more perceived inflation of staffing requirements.

10 interactions per 30m for a planning group may still be low, but a reasonable floor. 7 interactions per day...why even forecast or staff as it is in the noise and likely handled by whatever base staff (minimum paid / on queue hours) scheduled? Very unlikely there is any predictability with volumes that low. Is 1 FTE enough? What if the 7 interactions occur over 15h? What if they all happen to land in the same 15m window? What if that 15m window is in the morning one week, but the next it drifts to the evening?

Also, a correction to something said in a prior reply: we only use queueing theory when deferred work (e.g., emails) is in the mix.

Hi Linsey,

What sometimes helps in our case is to reforecast until Genesys returns a closer to reality Staffing requirement. We have raised several cased to Genesys support and the only "solution" we received so far is indeed to reforecast and change the Service goal templates.

Regards,

Daniela

Hi Camila, that helps a lot. Thank you for the explanation. I'm thinking of increase the service goal to something like 240 seconds, but I suspect that won't make a big difference on a low-volume queue. What do you recommend in this situation? We have the separate planning groups because it requires a separate agent training.

Hello Linsey, 

It can at first sight looks inflated, but it's actually consistent with how queueing models behave under low volume.

Staffing is calculated per interval, not averaged.

Genesys uses queueing theory to meet your service goal in every interval.

With:

~0.32 calls per 30 minutes
A 60-second ASA target

The system isn't thinking "7 calls per day." It's thinking:

"What staffing ensures I can answer any call that might arrive right now within 60 seconds?"

At such low volume:

Many intervals = 0 calls
Some = 1 call
Occasionally = 2 calls close together

Even though the average is tiny, the variance is huge relative to the mean.

To protect your ASA target, the model prepares for:

"What if a second call arrives while the first is still in progress?"

Your key difference:

Queue A - ~7 min AHT
Queue B - ~9 min AHT

That ~2-minute increase significantly raises the chance that:

Call #2 arrives while Call #1 is still active

So the model reacts by increasing staffing toward having:

"someone always available just in case"

That's why Queue B jumps toward ~0.7 FTE, even though workload is still tiny.

A 60s ASA at this volume is very strict.

At low volume, hitting that target often requires:

An agent being idle much of the time because even one delayed call can break service level.

So staffing shifts from:

"Cover the workload" to "Guarantee immediate availability"

It inflates your total FTE because many small, separate queues each require their own "just-in-case" capacity.

So instead of:

1 agent handling all low-volume work you get 0.5–0.8 FTE per queue - which adds up artificially.

So the reason Queue B is so much higher than Queue A, even with similar total volume is:

Higher AHT
Same strict ASA
Same low arrival rate

This leads to a much higher probability of overlap, which forces staffing up disproportionately.

I hope this helps you.