I think a lot of customers use callback mainly as a virtual hold, because that is the given solution for that, and since its such a common use case for callbacks we also expect that use case to work seamlessly with the rest of the platform, specifically WFM, which it doesn't, since offered is double counted, but if you don't include it you don't get the AHT time.
For those who would also like callbacks to be treated as a virtual hold as far as forecasting is concerned, that is 1 offered, 1 wait, 1 AHT from the time the original call entered the queue, please vote on this idea:
As for position in queue, you can certainly add priority to the callback if required to make it seem like it has waited the full length of time.
Original Message:
Sent: 08-01-2023 18:07
From: Gina Palmer
Subject: Callback Planning Groups Forecasting Offered Count
I believe there is a difference between Queue Callbacks and Virtual Hold. I don't think Genesys offers a true virtual hold, does it? Doesn't the waiting clock start over when you opt for a callback, so your place in line is really not the same? I believe that was the case in a previous version of Genesys I worked with but I'm not sure in this one.
------------------------------
Gina Palmer
Papa, Inc.
Original Message:
Sent: 07-26-2023 11:39
From: Brent Price
Subject: Callback Planning Groups Forecasting Offered Count
Isn't a callback just a virtual placeholder in line? (Not talking about ones where you are scheduling backoffice callbacks).
------------------------------
Brent Price
Applied Systems
Original Message:
Sent: 07-25-2023 11:08
From: Jay Langsford
Subject: Callback Planning Groups Forecasting Offered Count
"do you recommend that EVERY call that holds should be counted twice" - no and I do not see them as being the same thing. That scenario does not have a portion awaiting an agent for the initial call and then again the callback awaiting an agent some time later (potentially significantly later).
If you don't want callbacks to be forecast, then do not include them in route paths configured for your planning groups. That is essentially you saying you do not care about knowing the demand, staffing requirement, or staffing for callbacks. It is certainly an interesting stance.
Other than hiding the callback load, an alternative would be to apply a decrease by percentage modification to offered count by the percent of interactions that end in a callback.
In any event you have the power to ignore callbacks or apply gross modifications to forecasts. Other than that it would be an Idea entered on Ideas Lab and see if it garners widespread (from different customer accounts) demand and ultimately approval by Product Management to tackle.
Take care.
------------------------------
Jay Langsford
VP, R&D
Original Message:
Sent: 07-25-2023 10:27
From: Brent Price
Subject: Callback Planning Groups Forecasting Offered Count
True demand the demand as it would have naturally arrived. The call and AHT should be (for future forecasting reasons) be considered at the time it came in.
Best example:
Assume a model environment where everything comes in the exact same time every day (simplifying the model to teach the principle):
There is 1 agent staffed. AHT is 15 minutes per call. 4 calls arrive every day at 10am.
- Model 1: No callbacks/no abandons
- Reporting: NCO 4. Answered 10, 10:15: 10:30, 10:45
- True demand/staffing need: I need 4 agents staffed at 10am (plus shrink).
- Model 2: Callbacks/no abandons (the 3 other calls took callbacks, but were answered every 15 minutes)
- Reporting NCO 7 with a 15 minute AHT for 4 calls and 3 flow outs. (We were only offered 4 calls...this is just wrong)
- True demand/staffing need: I need 4 agents staffed at 10am (plus shrink).
- Model 2 would show I need 4 agents staffed at 10am and 1 extra each other interval...(or maybe it shows just 1 agent required at each 15 minute interval). Either way, the model is wrong.
Now, extrapolate this into the 1,000s and it gets way off and hiring/staffing decisions get skewed, reinforcing the bad loop of not answering the calls when they actually arrive in the first place. Now, for some, this is a strategy they are good with, for sure, but when scheduling, I would want to have meetings/lunches, etc. placed where we don't expect calls to arrive versus "knowing" that if I schedule those activities there, callers will get callbacks later (which is a far worse experience then a 30 second hold time).
Being "blind" to the callback is no different than being "blind" to the ASA of the held call. They are literally the same, just one is a virtual queue, or do you recommend that EVERY call that holds should be counted twice (which is what this argument is saying)?
------------------------------
Brent Price
Applied Systems
Original Message:
Sent: 07-25-2023 08:30
From: Jay Langsford
Subject: Callback Planning Groups Forecasting Offered Count
I am not sure if Ideas Lab has anything related.
Scenario:
- initial interaction comes in at 9:08am
- caller requests a callback at 9:11am
- queued callback offered to agent at 9:18am
Which offered (1 or 3) are you wanting to be blind to? Notice also that the two offered items happen in different 15m intervals (the granularity we forecast volume and AHT and staff for). Ignoring 1 or 2, means that interval will not be representing actual demand and staffing requirement.
What if 3's time was 9:38am? 10:03am? 2pm? The further apart they are the stronger the argument is that they should both be considered for forecasting and staffing.
"feature that requires a business to have the right staffing levels certainly sounds more like bug than a feature" - not following... The need is forecast and we show you staffing requirement predictions for meeting service performance goals. We attempt to utilize available resources within their work plan constraints. Nothing 'requires' right-staffing.
Interesting you noted a thread where a seasoned industry person agrees with current approach.
If you find a relevant idea in Ideas Lab, I would recommend voting on it. If none can be found, I would recommend creating one.
"Forecasting should be the true demand" - completely agree. The disagreement here is in what true demand means.
------------------------------
Jay Langsford
VP, R&D
Original Message:
Sent: 07-24-2023 21:22
From: Anton Vroon
Subject: Callback Planning Groups Forecasting Offered Count
Thanks for the responses @Jay Langsford and @Brent Price
Clearly the callback offered being double counted by WFM is an issue for many other community members, another recent post on the same subject:
https://community.genesys.com/digestviewer29/viewthread?GroupId=133&MessageKey=25c283f7-312e-469c-ad3a-5da22b288d72&CommunityKey=f6e20f73-2c94-4b44-be95-118216aafb4f&ReturnUrl=%2fcommunities%2fcommunity-home1%2fdigestviewer%3fCommunityKey%3df6e20f73-2c94-4b44-be95-118216aafb4f
To be fair, in the current staffing economy, not a lot of contact centers are at levels they want to be, and having a feature that requires a business to have the right staffing levels certainly sounds more like bug than a feature.
Callbacks being offered in Queue, is the standard, the solution of offer callback before hand is a bit odd in that regard, you want the opportunity for the call to be answered first before offering callback. Its not really a solution but a workaround for feature that doesn't work with all the other features across the platform.
Does anyone know if there is an existing idea open to address this?
------------------------------
Anton Vroon
Original Message:
Sent: 07-24-2023 10:39
From: Jay Langsford
Subject: Callback Planning Groups Forecasting Offered Count
If 'for a bit' is a short duration and reasonably uncommon, then naturally those periods will be heavily muted in the forecast due to them being outliers.
"since staffing for the original call removes the 2nd call", but during the original it was not staffed correctly thus the callback request and we have to account for staffing the queued callback. If the original call portion is sufficiently staffed you would avoid callbacks and if you avoided callbacks, then callbacks would have less offered count in the forecast.
The other thing to remember is the time of the initial call and the eventual queued callback could be minutes to many hours separated. You must forecast need and staff for both because there is load and service performance goals for both periods.
The more right-staffed you are the lower the callback requests will be. Again you could offer a callback option in the IVR, say after playing estimated wait time or estimated position in queue to the caller, and avoid double counting.
Also, if a small percentage of total interactions end up requesting a callback, then the impact will be in the noise (likely less than the forecast error outside of these scenarios).
------------------------------
Jay Langsford
VP, R&D
Original Message:
Sent: 07-24-2023 10:22
From: Brent Price
Subject: Callback Planning Groups Forecasting Offered Count
100% you don't want to double count those, ideally, since staffing for the original call removes the 2nd call. The fact they are counted twice in Genesys is a flaw, not a feature in my opinion. For example, if I had a queue where every 15 minutes 1 agent could handle 1 call and we get 1 call per interval, I would want to staff 1 agent per every 15 minutes (yeah, ignore shrinkage for this). Now, if each of the calls hit call-back because he was really long on his first call that day, then my normal forecast of 32 calls (the actual demand) would show as 64 calls (demand + callback) and I would think I need twice as many agents.
If you are staffed incorrectly for a bit and get a lot of callbacks, that spike in callbacks will falsely inflate not only the callbacks, but make it look like you need more staff later in the day (callback time) versus when you really need them (original arrival time).
------------------------------
Brent Price
Applied Systems
Original Message:
Sent: 07-24-2023 07:45
From: Jay Langsford
Subject: Callback Planning Groups Forecasting Offered Count
We have to forecast and attempt to staff for the initial call. If there could be enough scheduled staff, then the caller would most likely not have opted for a callback. Not doing so would be wrong.
We have to forecast and attempt to staff for the queued callback. Not doing so would be wrong.
It comes down to what percentage of total interactions eventually request a callback. The higher it is, then it points to an issue in staffing for the initial call in period. If it is chronically high, then it points to chronically low staffing compared to need.
To completely mitigate then the caller should be presented a callback option prior to enqueuing the call to avoid the initial offered count.
------------------------------
Jay Langsford
VP, R&D
Original Message:
Sent: 07-23-2023 19:14
From: Anton Vroon
Subject: Callback Planning Groups Forecasting Offered Count
Thanks @Jay Langsford
So we have voice offer count, and callback offer count, so 1 interaction but 2 offer counts. Does this not put the forecast out since it is using offered and therfore will be expecting more interactions to staff than there will actually be?
------------------------------
Anton Vroon
Original Message:
Sent: 07-21-2023 16:08
From: Jay Langsford
Subject: Callback Planning Groups Forecasting Offered Count
The initial interaction would have an offered count - e.g., I called in and prior to requesting a callback. The queued callback would have an offered count. WFM does not include the flow out as part of offered count and thus would not be forecast.
Example scenario with metrics:
- call waits in queue = voice offer count
- callback created after request = voice flow out count
... sometime later
- callback waits in queue = callback offer count
- callback answered = callback answer count
------------------------------
Jay Langsford
VP, R&D
Original Message:
Sent: 07-20-2023 19:05
From: Anton Vroon
Subject: Callback Planning Groups Forecasting Offered Count
Hi team
A couple of question for our WFM experts.
We currently have callbacks and the associated queue/skills configured within planning groups and use this in our forecast along with the voice queues/skills planning groups.
Since callbacks start as voice and get create while in Queue and then flow out, is forecast taking that into account and not double counting the interactions for the offered amount?
What is the best practice for taking in to account the callback portion in forecasting to ensure offered, aht, wait times etc are all taken in to account accurately?
Anton
#Genesys Cloud CX
#Workforce Management
#Forecasting
------------------------------
Anton Vroon
------------------------------