Professional Guide
Erlang C and Call-Centre Staffing: How Many Agents Do You Need?
The Erlang C formula for staffing a service desk or call centre to a target answer time, with the assumptions, limits, and how queue abandonment fits in.
If calls arrive randomly and agents are finite, work queues. Erlang C is the standard queueing model used to estimate how many agents are needed so that a target proportion of callers are answered within a target time.
This guide explains the inputs (call volume, average handling time, service target), what the model assumes, and where it falls short — so you can use the PanelRoster Erlang C calculator with judgement.
The inputs to Erlang C
Erlang C takes call arrival rate (calls per period), average handling time (talk plus after-call work), the number of agents, and a service target (for example, 80% of calls answered within 20 seconds). From these it computes the probability a caller waits and the proportion meeting the target.
The PanelRoster Erlang C staffing calculator solves the reverse problem: given volume, handling time and a target, how many agents are needed.
What the model assumes
Erlang C assumes calls arrive randomly (a Poisson process), arrivals wait patiently in a single queue, agents work independently at a constant handling time, and callers never abandon. These assumptions are close to reality for many service desks but never exact.
The practical effect: Erlang C tends to understate staffing when callers abandon (see queue abandonment) and overstate it when arrivals are smoothed or handling time varies a lot. Treat the result as a starting point, then validate with real queue data.
Combining Erlang C with abandonment and capacity
Callers who wait too long hang up. The queue-abandonment-rate calculator quantifies that share, and the capacity and shift-coverage calculators translate required agents into rosters. Together they answer the operational question: how many agents, on which shifts, to hold service levels through the day.
Re-run the model when volume, handling time or targets change, and compare predicted vs actual service levels monthly.
Key takeaways
- Erlang C converts call volume, handling time and a service target into an agent requirement.
- It assumes random arrivals, patient callers and steady handling — all approximations.
- Model abandonment and shift coverage alongside Erlang C for a realistic staffing plan.
- Validate model output against real queue data; re-run when inputs change.
Tools used in this guide
Related guides
References
References are provided for further reading; PanelRoster is not affiliated with the linked resources.