How the calculation works
- Contacts per hour = contacts in the interval ÷ interval length in hours
- Traffic intensity (Erlangs) = contacts per hour × average handle time (seconds) ÷ 3,600
- Probability of waiting = Erlang C(traffic, agents), computed from the Erlang B recursion to avoid factorial-sized numbers
- Service level = 1 − probability of waiting × e^(−(agents − traffic) × answer-within seconds ÷ average handle time)
- Average speed of answer = probability of waiting × average handle time ÷ (agents − traffic)
- Occupancy = traffic ÷ agents
How the agent count is found
The calculator starts at the traffic intensity rounded up and adds one agent at a time. It stops at the first agent count that meets the service-level target and keeps occupancy at or below your maximum. If no count up to 1,000 agents works, it reports that the queue exceeds its limit; split the workload into smaller queues or shorter intervals.
Service level is the share of contacts answered within your threshold. The defaults, 80% answered within 20 seconds with a maximum occupancy of 85%, are editable starting points, not recommended targets.
Using it for call center and chat staffing
The result is the number of agents who must be available in that interval. Shrinkage is not included: apply it outside the Erlang result to get scheduled headcount, as the staffing calculator does for monthly workload.
For live chat, the model treats each agent as handling one conversation at a time. It does not model concurrency, so chat teams that run several conversations per agent need their own concurrency rules on top of this result.
Assumptions and limits
- Contacts arrive at random, and nobody abandons the queue while waiting.
- All contacts are the same kind of work and go to a single pooled queue that any agent can answer.
- Each interval is calculated on its own, so plan with interval-level arrival data rather than a daily average.
- Real schedules still need shrinkage, concurrency rules and skills on top of the result.