Genesys Cloud (formerly PureCloud)

Discussion Thread View
Expand all | Collapse all

WFM Forecasted Staffing and Planning Groups

  • 1.  WFM Forecasted Staffing and Planning Groups

    Top 25 Contributor
    Posted 08-16-2020 17:04
      |   view attached

    I've found that the latest changes to Planning Groups have increased the stated forecast requirements significantly.
    I believe this is because each group does not share resources with other planning groups which would occur in multi-queue or multi-skill configurations.

    I ran two forecasts/schedules and compared the outputs using the Weighted Historical Import:

    Scenario one: Two planning groups, both with identical forecasts
    Scenario two: One planning group with double offered (i.e. the sum of Scenario One above) and the same AHT
    Both scenarios use the same service goals.

    The output was that Scenario one has a total required of between 11% - 36% higher than Scenario two.
    It's as if there is no multi-skill efficiency for agents across multiple planning groups.
    See the attached Excel sheet.

    I can't see any documentation online saying something to the effect of "Planning Groups should not share resources, otherwise forecasted staffing will be overstated".

    This is impossible to work in environments where agents are multi-skilled across queues.
    We need to have individual Planning Groups as our forecasts need modifying by queue (e.g. multi-skilled agents with Sales and Service), yet we can't have multiple Planning Groups as our staffing forecasts are wrong.

    Really keen for some feedback, I might be completely wrong

    Will Bellerby
    Pyrios NZ Ltd


    Forecast Test 22.xlsm   35 KB 1 version

  • 2.  RE: WFM Forecasted Staffing and Planning Groups

    Top 25 Contributor
    Posted 08-16-2020 20:41
    Will be interested in this outcome too.  We're about to build our first GC WFM deployment a part of a migration from Engage.  This customer has agents in multiple queues where there is a 1:1 skill:queue relationship but not every agent is in every queue/skill, plus there are varying skill levels.

    Vaun McCarthy
    NTT New Zealand Limited

  • 3.  RE: WFM Forecasted Staffing and Planning Groups

    Posted 08-17-2020 08:01
    Agents can belong to multiple planning groups. Shared resources/load is fine as long as it is within a business unit; shared resources/load that span business units should be avoided.

    I would suggest reading the following recent post:

    In regards to your two scenarios, of course creating a scenario with a more homogenous set of load and resources will produce a lower requirement than a more heterogenous mix. Multi- queue, media type, and/or skill capabilities and efficiencies are taken into account, but if you have isolated 'sets' of load and or resources, especially many/small ones, it will drive up staffing requirement.

    Jay Langsford
    Senior Director, Workforce Optimization Engineering

  • 4.  RE: WFM Forecasted Staffing and Planning Groups

    Top 25 Contributor
    Posted 08-17-2020 16:43
    Thanks Jay,

    How do you suggest we make forecast modifications to a single queue, inside of a planning group which has multiple queues?


    Will Bellerby
    Pyrios NZ Ltd

  • 5.  RE: WFM Forecasted Staffing and Planning Groups

    Posted 08-17-2020 17:05

    Modifications are done at a planning group level. If a planning group has multiple queues, then you should consider that one thing not multiple individual things (e.g., if you want to increase volume by 10% for one queue in a planning group with multiple queues, you'd have to realize that it would need to be discounted). If it is really necessary for a modification of just one queue, then your option is to have the one queue in its own planning group.

    In general I would say less is more with modifications. I rarely see real world examples where modifications made for a more accurate forecast when actuals come in. Weighted historical index method with equal weighting of just the past 4w tends to be generally good; automatic best method would be preferable after there has been a chance for patterns and such to be learned (so not right after a cold start case, but 2-3w after) - and both with either no or very few modifications.

    Jay Langsford
    Senior Director, Workforce Optimization Engineering

  • 6.  RE: WFM Forecasted Staffing and Planning Groups

    Posted 12-01-2020 11:17

    Hello Jay,

    Please allow me to jump in this discussion.

    You mention that shared resources/load that span business units should be avoided.

    Also, in the Resource Center under Business unit key principles, it says Do not configure agents who handle the same queue into different business units.

    Can you please elaborate more why? And what would happened if this recommendation is not followed?

    I am currently preparing the WFM configuration for a client who currently has 3 business units but some resources have the skills to answer calls across the 3 business units.

    The customer's desire is to keep these 3 units separate because their forecasts and schedules are generated by 3 different teams. What do you recommend In order to resolve the situation/remedy the customer's need?

    Thank you.

    caroline mbanga
    i3Vision Technologies Inc.

  • 7.  RE: WFM Forecasted Staffing and Planning Groups

    Posted 12-01-2020 11:35
    Edited by Jay Langsford 12-01-2020 11:57
    Because staffing (scheduling, predicted requirement, predicted performance (service level, ASA, abandonment), etc.) will be completely blind to what is essentially a rogue set of resources handling interactions specifically forecast for the BU. E.g., we think you have 100 agents, but in actuality you have 150 or something

    This will result in what will be perceived as:
    • inflated staffing requirement predictions
    • higher actual service level than predicted
    • lower actual ASA than predicted
    • lower actual abandonment than predicted

    One of our fine folks on the consulting side can jump in (tagging @Chip Funk​) with suggestions but I will note mine...

    Customer may wish BU2 to handle some BU1 queues if BU1 resources are too busy. I would recommend against BU1 resources and BU2 resources having common queues to implement this behavior. Instead I would recommend the following:

    • BU2 would have a BU1OverflowToBU2 queue
    • BU1 would overflow/transfer interactions to the BU1OverflowToBU2 queue so that BU2 resources could handle

    By implementing the overflow/transfer model each BU would have independent forecasts and staffing. E.g.,
    • BU1 has the following queues: BU1Q1, BU1Q2, ...
    • BU2 has the following queues: BU2Q1, BU2Q2, ..., BU1OverflowToBU2

    Jay Langsford
    Senior Director, Workforce Optimization Engineering

  • 8.  RE: WFM Forecasted Staffing and Planning Groups

    Posted 01-04-2021 12:06
    Edited by Jacques Tiberghien 01-04-2021 12:52
    Happy new year Jay,
    This topic is really interesting when considering worldwide or multi country deployment. I am currently working on a big prospect that has CCs in more than 10 countries and the scheduling feature is a must for them. but I don't see how to be accurate in their scenario.
    Let me oversimplify the scenario. Let s consider 1 queue "main" and 2 countries (which we set as skills) "country1" and "country2" and those countries are also separated business units. But keep in mind that there is more countries. As of now this is how we configured the routing :
    * all channels are placed on the "main" queue
    * depending on the DID or destination@, we set the appropriate (country) skill
    * each country provide 24h operations but with reduced agents at night
    * ​Within this queue, we use the bullseye routing to first target the "local" agents, then after a certain amount of time, we remove the country skill.

    The problem with the solution you gave is that splitting the resources to multiple queues and transfering the overflow to an "BU1OverflowToBU2" would prevent agents from BU1 to answer the interaction at that point. Also this can be apply if we have just 2 BUs, but in my customer case, there would be more than 10 that need to answer the same overflow.

    Do you see any solution to this scenario?
    Also how work the planning groups/Wieghted historical forecast with the bullseye routing? I assume that with the bullseye, there won't be any route that contains only the overflow (queue without the country skill) as bulsseye is just a routing method?
    Thank a lot

    Jacques Tiberghien
    Paxyl Solutions

  • 9.  RE: WFM Forecasted Staffing and Planning Groups

    Posted 01-04-2021 13:24
    Edited by Jay Langsford 01-04-2021 13:24

    I would suggest not to have one 'main' queue and instead drive with DID/country specific queues (rather than use country skills). It's likely you'll want to see analytics views at the country level and many of our views are queue specific. I'm not sure you'd find that if you went the skill level at all or as easily.

    I would also encourage you to engage @Chip Funk and his team on consultation services.

    Genesys Cloud WFM does not support bullseye routing (BER) forecasting/scheduling, so I would recommend avoiding use of it for queues you want to forecast/staff (at least accurately so).

    As skills are stripped in BER, historical data from analytics _only_ shows the original requirements. This means that if a BER queue is included in WFM, the forecasting and staffing will only see the original requirements (e.g., all the skills) and forecast and staff accordingly.

    BER, in general, isn't a great practice. It can do unnecessary waits when able agents are available immediately (e.g., you are waiting until the time you strip a skill). BER is easy to describe/explain and that simplicity is a trap folks fall into not realizing the deficiencies. Skills-based routing evaluates all of the agents and determines the best agent without artificial time delays that BER has.
    > Currently, workforce management forecasting and scheduling only supports inbound voice, chat, callbacks, email, and message media types in queues configured for standard ACD routing. Forecasting and scheduling does not support queues configured for bullseye routing.

    Jay Langsford
    Senior Director, Workforce Optimization Engineering

  • 10.  RE: WFM Forecasted Staffing and Planning Groups

    GCAP Member
    Posted 01-05-2021 09:29
    We decided against bulls-eye routing and are using skill proficiencies in combination with Best Available Skills routing. Overflow agents have a lower proficiency for the skill associated to the call. Like Jay states above, this avoids the wait for a skill-requirement to be dropped, and also allows to have multiple layers of overflow if so desired. We don't use WFM yet, so I cannot comment on the effects this may have on workforce planning.

    Sven Schiller