Forecasting Practices – Initial steps
Developing good forecasting practices for your workforce needs starts with understanding the steps in the process and building your capability. Those experienced in forecasting can quickly create a forecast for future weeks and incorporate this work into their regular work activities, however these practices need to be mastered individually by learning each step. You’ll never optimise your staffing if you aren’t forecasting your workload and staff requirements, so here’s an outline to help you get started, broken down to a level that you shouldn’t find complex to undertake.
The building of your capability in each of the steps will lead to you creating good forecasting processes overall for your multiskilled, multi-channelled contact centre/service centre/helpdesk. Here is one step in the early stages of forecasting call volume for a future week, using Daily Distribution curves and comparisons at an interval, day, and week level.
Gather and Present the Data
When preparing a forecast, start by gathering a number of weeks of historical data (say 6-8) and then create future interval data and daily distribution curves, as shown in the two attached screen shots of forecasts over two sequential weeks in August:
(Click to enlarge each graphic)
These graphics have been created in the injixo cloud-based WFM service, and show the weekly forecasts for the weeks of 3 August and 10 August, with individual distribution graphs per day. Call volume for each interval and day is forecast, and totalled to give the full week’s forecast by interval.
The first practice in analysing a future week’s forecast is to compare the future forecast you have at the full week’s level, a daily level, and an interval level. In this case the analysis is being carried out on the week of 10 August. In the first instance, if the forecast for Monday 3 August was 300 calls, will a forecast of 268 calls for 10 August be appropriate? Why do you expect inbound call volume to reduce by this amount, and that will contribute to a fall in inbound call volumes for the whole week, so what is your rationale this? It probably will depend on the historical weeks’ data that you’ve chosen for the basis of creating this forecast. Next, ask yourself how has the distribution of the call volume over the course of the day contributed to the reduction in the total volume?
You’ll also need to refer back to actual call volumes for the analysis. If your forecast for 3 August was 1306 calls for the week, what was the actual call volume offered to your centre? How far above or below the forecast was the actual, or was it pretty well spot on?
From here on there’s a lot more to do before you could accept the forecast and schedule staff for the week, but in the first instance, gathering and presenting the data for these comparisons, creating the daily distribution numbers and charts is an essential task, and undertaking the comparison analysis before you move on of both forecast v forecast and forecast v historical data are essential steps. Adjusting your forecast after this initial analysis is part of the process, and combines your process with your intuition and experience. Good forecasting is a combination of all three.
For further help on Forecasting or any aspect of your workforce optimisation needs, contact us we’ll be pleased to help or just have a chat about this with you!
Giles Potter is Director of Great Outcomes Ltd.
Giles Potter: @gdcpotter
#WFM #CCTR #workforcemanagement #Forecasting #accuracy #schedule