Service Desk Staffing Calculator (Erlang C)
Work out how many agents your service desk needs to hit its service level, using the Erlang C model. See how waits change agent by agent and plan cover hour by hour.
- Erlang C model
- Shrinkage and occupancy
- Hour-by-hour plan
For a desk taking 120 calls in its busiest hour at 6 minutes each, Erlang C needs 15 agents on the queue to answer 80% within 60 seconds, at 80% occupancy. With 30% shrinkage for leave, training and breaks, roster 22. One agent fewer drops the service level to 65.5%. Size for the real peak, such as Sunday morning on a Sunday–Thursday desk, not the daily average.
1. Your peak hour
Enter the busiest hour's demand and your service level target.
2. Plan the whole day
Spread a day's contacts across the hours using a typical arrival pattern, with the handle time and targets above.
Getting staffing numbers you can defend
- Use real arrival data. Take contacts per hour from your telephony or ITSM reports, or the busy-hours heatmap in the service desk dashboard. Plan for a typical busy week, not the average.
- Measure handle time end to end. Include talk or chat time and after-call work such as updating the ticket. Leaving out wrap-up time is the most common reason rotas come up short.
- Be honest about shrinkage. Add up leave, sickness, training, team meetings and project time from last year. If the rota assumes 20% and reality is 35%, the SLA breaks every afternoon.
- Staff the peaks, not the day. A daily average hides a Monday 09:00 peak that can run at twice the hourly average.
- Revisit after every change. A new self-service portal, a migration or a store opening changes demand. Re-run the numbers before the rota, not after the complaints.
Method & ownership
Agents are sized with the Erlang C queueing model: workload in Erlangs = contacts per hour × handle time, service level = 1 − P(wait) × e^(−(agents − workload) × answer time ÷ handle time), then rostered headcount adds shrinkage. Erlang C assumes callers wait rather than abandon, so treat results as a planning baseline.
The formulas used here are standard, published methods. The implementation, interface, wording, findings logic and benchmark figures are original work by ITSM Compass. Results are estimates, not professional advice. See the terms of use.
© 2026 ITSM Compass. All rights reserved.
Questions
What is Erlang C?
Erlang C is the queueing formula contact centres and service desks use to work out how many agents they need. Given how many contacts arrive, how long each takes and how quickly you want to answer, it predicts the service level and average wait for any number of agents.
What is a good service level for an IT service desk?
80% of calls answered within 20 to 60 seconds is the most common phone target. Many internal IT service desks use 80% within 60 seconds. For chat, 80% within 30 to 60 seconds is typical. Pick what your users need and what you can fund, and write it into the SLA.
What is shrinkage?
Shrinkage is the share of paid time agents aren't available to handle contacts: leave, sickness, training, meetings, breaks and project work. 30% to 35% is common. If you need 10 agents taking contacts and shrinkage is 30%, you need to roster 15.
Why cap occupancy?
Occupancy is the share of time agents spend handling contacts. Above about 85% to 90% for long periods, people burn out, quality drops and absence rises, which pushes shrinkage up. Capping occupancy adds agents at high volumes, even when the service level alone would allow fewer.
Does Erlang C work for tickets and email?
Erlang C assumes contacts wait in a live queue, so it fits phone and chat best. For email and portal tickets, work out hours of work per day (tickets × handle time) and divide by productive hours per agent, then check the backlog against your response targets.
Rota looks right but SLAs still slip?
A fractional service delivery manager can rebuild the capacity plan, the rota and the reporting behind it.