About the forecasting process

To create a forecast, you must complete each of the following tasks.

Generate a distribution

Generate a distribution (see Create a distribution) or use an existing distribution for the service queue and reference period you select. Choose six to eight recent weeks for the reference period, taking care not to use a week with unusual volume, such as a holiday week.

Distributions are responsible for determining the expected percentage of the daily contact volume that occurs in each interval of the day. Select a reference period that best represents the expected contact distribution over the course of a day.

Edit the distribution

If needed, edit the distribution (see Edit distributions and forecasts). For example, if you expect the contact volume to be lower on a certain day of the week in your forecast period, you can decrease the contact volume on that day in the distribution.

NOTE   If a special event is assigned to that day (see Account for days with abnormal contact volumes), then WFM normalizes the contact volume so that distributions are more accurate. In that case, there is no need for you to manually adjust the distribution.

Select a forecasting method

There are two forecasting methods, Standard and Advanced. The Standard method is the legacy method used by WFM. This method treats all weeks the same: it is not possible to represent differences in contact volumes or handle times on a week-by-week basis inside a single month. Therefore it does not reflect seasonal variations in contact volume at different times within the week and month. This method is recommended if you have one year or less of historical data to use as reference data.

The Advanced forecasting method is able to more accurately reflect seasonal variations in contact volume and handle time. This method is recommended for contact centers that do experience seasonal fluctuations in contact volume, and that have more than one year of historical data available. The longer the reference period, the more accurately this forecasting method is able to reflect weekly, monthly, and yearly variations.

Select a reference reriod

WFM uses historical data to project future requirements. The reference period you choose depends on which forecasting method you choose.

• Standard Forecasting Method. Identify the reference period with the historical data that most closely resembles the period for which you want to generate a forecast. The reference period you choose should reflect any weekly patterns that are likely to occur during the forecast period. Choosing a reference period from a year prior to the forecast period often provides the best reflection of weekly patterns.
• Advanced Forecasting Method. This forecast method performs better with a longer reference period. Identify the reference period of a year or longer with the historical data that most closely reflects the weekly, monthly, and yearly seasonal patterns that are likely to occur during the forecast period.

To view the historical data available to you for the service queue and the reference period you are considering, use the View and Edit Historical Data page (see View and edit historical data).

Generate the forecast

When you launch the forecast request, WFM performs these steps:

1. Applies any special event adjustments you assigned for the service queue in the historical reference period.

2. Generates the average contact volume for each day of the week using data from the reference period (Standard forecasting method) or generates the average daily contact volume using data from the reference period (Advanced forecasting method).

   NOTE   When calculating contact volumes, WFM rounds near (up or down to the nearest whole number) except when the value would be rounded to zero. In that case, the value is rounded up to 1.

3. If you opted to use a historical trend, WFM determines the annual growth rate based on the two selected reference periods. The annual growth rate comes from the change in contact volume over the change in time between the two reference periods. The change in time is measured by the distance between the mid-date of each reference period. If you opted to use an expected trend, WFM adjusts the volume projections according to the percent annual growth rate you provide. Historical Trend and Expected Trend are applied to the forecase period differently. See Use trends in forecasting for more information.
4. Adjusts the volume projection for each day by the contact adjustment factor.
5. Applies the contact ratios from the distribution for each interval in the day. Contacts are redistributed for non-interactive service queues that use even or proportional redistribution during closed hours. Otherwise, the number of contacts is taken from the contacts arriving during service queue open hours. If the service queue is closed during an interval, the number of contacts is zero.
6. For non-interactive service queues, multiplies the projected contacts for each interval by the average handling time in order to estimate the amount of agent handling time required.

7. Uses an algorithm to determine the number of agents required per interval to meet the service level goals.

   NOTE   If a service queue is closed on a day of the week, you must edit the service queue on the Service Queues page to clear the Active check box for that day of the week. If this is not done, the forecast request will fail.

   NOTE   If there is a date with a firm date association within the forecast period, WFM uses the volume from the firm date association reference date, but the distribution of that contact volume by interval for the day comes from the distribution.

Review the forecast

If you do not think the forecast values are on target, you can edit the forecast (see Edit distributions and forecasts).