Forecasting and Scheduling
This topic provides information that you should consider when planning your forecasting and scheduling requirements for your contact center. See:
You can create forecasts based on various kinds of data. Ideally, you already have a substantial quantity of good-quality historical data on contact center interactions that you can import into the WFM database. If you have historical data, you can use either of two forecasting algorithms depending on the amount of quality historical data available. The Expert Average Engine requires a full week of historical data with no missing timesteps. To use the Universal Modeling Engine, you must have at least a full year of historical data to create forecasts.
If historical data is unavailable or of poor quality, you can create forecasts based on templates. Templates reflect estimated interaction levels for different days and times and can be constructed for each activity.
When you do not have enough historical data to use the Expert Average Engine or the Universal Modeling Engine, you can combine the historical data with overlap templates, which fill in gaps in the historical data.
Creating optimal forecasts depends not only on whether historical data is available, but also on usual workflow. Contact centers with very regular interaction volumes require different forecasting considerations than contact centers that experience frequent or marked variations of interaction levels.
If your site activity load is highly predictable, you can apply a specific interaction volume or AHT to each time interval in the scenario.
Forecasting also incorporates figures such as staffing overheads, service objectives, and occupancy into the staffing calculations, allowing precise regulation of forecasting levels. You can create a variety of forecast scenarios, by using different service objectives or staffing parameters to help you create realistic contact center strategies for varying circumstances. After you decide which scenario best fits your environment, you publish it to the WFM database, where it becomes a part of your Master Forecast, upon which schedules are built.
Factors and Events
WFM can track events that may affect interaction volume. These events are based on factors, which are event types upon which events are built. When planning your forecasting, consider what factors and events might affect forecasts so you can configure them before creating forecasts. A sales promotion or marketing campaign, for example, may cause a predictable peak in interaction volume. Such events are entered in WFM Web and used by the advanced WFM algorithms. If an event recurs, the forecasting algorithms learn the impact of that event and account for its impact in future forecasts.
WFM schedules each agent individually, building schedules that allow for intra-day overhead. Therefore, you do not need to inflate staffing requirements to accommodate overhead. The only overhead additions that you need to account for are intangibles, such as starting up agent desktop applications, bathroom breaks, and so forth, and unplanned overhead, such as training or meetings that are not yet scheduled (or sick days, which, presumably, are always unplanned).
Because each site is different, some planners might choose to fully configure meetings and training. Others might opt to build these into the schedule after it is generated. WFM supports both strategies. However, a good rule of thumb says that if the meeting or training must occur at a specific time, it should be configured beforehand. Otherwise, you can add meetings and training after building the schedule.
Creating Blank Schedules
Agent-based scheduling might not always be appropriate for your contact center. If not, you can also create schedules using profile agents. Profile agents are user-defined, hypothetical agents, based on contract data. Using profile agents results in blank schedules that contain an appropriate number and assortment of schedule slots for the agents to be hired.
Managing Schedule Bidding
Supervisors can create a profile schedule which authorized Agents then bid against, for the schedule slots that they prefer. The Supervisor can automate the resolution of conflicting bids according to stated Agent preferences as well as their Seniority and Rank, and then tweak it manually before publishing the official schedule. Such a schedule can be designed to repeat over an entire quarter.