Workforce Engagement Management

Discussion Thread View

Possible reasons why predicted staffing requirement is higher than expected

  • 1.  Possible reasons why predicted staffing requirement is higher than expected

    GENESYS
    Posted 08-10-2020 08:14
    Edited by Jay Langsford 08-19-2020 13:53

    Higher predicted staffing requirement than expected can be due to one or more of many upstream causes. Here are the ones that come to mind:

    1. Higher forecast volume than actuals. E.g., you are forecasting more load than actual so therefore predicted staffing will be pessimistic
    2. Higher forecast AHT than actuals. E.g., you are forecasting more load than actual so therefore predicted staffing will be pessimistic
    3. Unknown/rogue, from the business unit's perspective, agents working the business unit's volume. E.g., you have agents not associated with the business unit that take interactions associated with planning groups of the business units
    4. Historically low adherence for the business unit's perspective; e.g., you've got an overly optimistic view on agent adherence, but when predicting staffing requirement we actually model past performance and account for bad adherence
    5. Lower actual service level than you plan for (i.e., actuals far less than what you set as service goals for the business unit). E.g., you have an overly optimistic view on what reasonable SLs can be achieved versus what actually occurs and in order to hit that overly optimistic goal it requires higher staffing requirements than you expect.
    6. Higher actual ASA than you plan for (i.e., actuals far greater than what you set as service goals for the business unit). E.g., you have an overly optimistic view on what reasonable ASAs can be achieved versus what actually occurs and in order to hit that overly optimistic goal it requires higher staffing requirements than you expect.
    7. Higher abandonment rate than you plan for (i.e., actuals far greater than what you set as service goals for the business unit) or maybe no abandonment rate goals set in your service goals for the business unit. E.g., you have an overly optimistic view on what reasonable abandonment rates can be achieved versus what actually occurs and in order to hit that overly optimistic goal it requires higher staffing requirements than you expect.
    8. Unrealistic service performance goals (i.e., service level, ASA, abandonment rate) compared to what you get historically. E.g., you set service level goal of 80% per interval, but historically you don't exceed 60%
    9. Horribly micro-fractured load across many planning groups and/or filtering on low volume/staffed planning groups. E.g., you have many planning groups with low single-digit total offered volume per hour or, worse, per day. Economies of scale and better forecasts and staffing requirement predictions occur with larger volumes and more homogeneous load/resource sets. The more fractured, smaller, and more heterogeneous you configure your system, the worse forecasts and staffing requirement prediction is going to be.


    Basic things to look at:

    • Forecast versus actuals in short-term forecast: look at your forecast accuracy. If forecast volume and/or AHT is higher than actuals by even a few percentage points, it will drive up staffing requirement predictions.
    • Intraday monitoring: look at scheduled versus actual for agents, abandoned, predicted versus actual service level, forecast versus actual offered and AHT.
    • Real-time and historical adherence: how good is your agents' adherence.


    In a recent case a customer had between 30% and 60% actual abandonment rate. That meant essentially they were telling WFM to staff for a completely unrealistic scenario, 10% abandonment rate in their case, when they actually abandoned 3-6 times that amount historically with all of their staff. I have also seen forecast accuracies sub 80% where they were way over forecasting volume and/or AHT. Many times, simply removing forecast modifications, gives them a 10 point or more boost in accuracy (e.g., just using a stock forecast).


    #WorkforceManagement

    ------------------------------
    Jay Langsford
    Senior Director, Workforce Optimization Engineering
    ------------------------------