- What is an Erlang calculator?
- It is a staffing tool that converts forecast contact volume, average handle time and a service level target into the number of agents required per interval. It exists because staffing does not scale linearly with volume: queueing means the last few seconds of answer speed cost disproportionately more agents than the first.
- What inputs does it need?
- Expected contact volume for the interval, average handle time, the service level target expressed as a percentage answered within a number of seconds, and the interval length. Shrinkage is applied after the calculation to turn required agents on the phone into rostered heads.
- Why does the output need shrinkage added?
- Because the model returns the number of people who must be handling contacts, not the number who must be rostered. Breaks, training, meetings, absence and other paid non-productive time sit on top. Treating the raw output as a headcount is the most common way an Erlang-based plan comes up short.
- Should the same model be used for every channel?
- No. Use separate models where handle times differ significantly, because voice, chat and email behave differently in a queue. Chat in particular allows concurrency, which the basic voice assumption does not capture.
- What are the common mistakes?
- Using outdated handle-time data, ignoring seasonality or a change in channel mix, and treating one forecast as certain when volumes are volatile. Small changes to the handle-time assumption produce large staffing effects, so the inputs deserve more review than the model does.
- How do you know whether to trust the output?
- Compare planned against actual staffing over time and track the deviation, alongside service level, abandonment, answer speed and forecast accuracy. Documenting the service level target and inputs used in each cycle is what makes a later variance explainable rather than mysterious.