How the calculation works
- Workload in Erlangs. Contacts per hour times handle time in hours. 120 chats an hour at 6 minutes each is 120 × 0.1 = 12 Erlangs: on average, 12 agents are busy at any moment.
- Chance of waiting. For each number of agents above the workload, Erlang C gives the probability that a new contact has to wait.
- Service level. The share answered within your target time: 1 - P(wait) × e^(-(agents - workload) × target time / handle time).
- Agents on duty. The smallest number of agents that meets your service level target.
- Agents to schedule. Agents on duty divided by (1 - shrinkage), rounded up.
Reading the table
The table shows the service level, average wait and occupancy for one or two agents fewer and more than the answer. Each added agent improves the service level by less than the one before, which helps when you weigh cost against speed.
What Erlang C assumes
- Contacts arrive at random, and the hour you enter is representative. Use your busiest hour, not a daily average.
- Nobody hangs up while waiting. In practice some customers abandon, so real waits are a little shorter than the formula predicts.
- Every agent handles every type of contact. If you route by skill or language, run the calculator for each group.