Forecasting

Every schedule starts with an estimate: how much work is coming? A forecast answers that question precisely enough to turn it into a staffing requirement for every quarter or half hour.

#What gets forecast

Two quantities: volume, meaning the number of contacts, and handling time. Both are estimated per channel and skill, because a chat in German needs different people from a billing call.

#What a forecast learns from

History is the foundation: weekdays, seasonal patterns, typical daily curves. But the past does not know the future. So events are added: public holidays, billing runs, campaigns, product launches, announced maintenance.

Unplanned events such as a network outage cannot be predicted, but they can be absorbed during the day. A good forecast is therefore compared with actuals continuously and adjusted.

#From volume to staffing

From volume, handling time and the service level target, planning calculates how many people each interval needs. The classic method uses Erlang models.

In chat and messaging, agents handle several conversations at once, so simulation gives better results. On top comes an allowance for breaks, training, leave and sickness.

Small errors have outsized effects: if an interval is short by just a few people, waiting times often jump.

#Forecasting in Univerx

Forecasting in Univerx Workforce estimates contact volume per channel and skill from history and events. Planned campaigns enter as events, so callbacks do not catch staffing by surprise.

Read more in Forecasting with events and in the guide Workforce planning in the contact centre: the basics.