Call center staffing planner: agents for every interval of the day
Paste or type the calls you expect in each half hour or hour of a day. The planner runs Erlang C on every interval and returns the agents you need on the phones, the people to schedule after shrinkage, the service level and speed of answer you can expect, and the day's peak and total agent-hours.
Respuesta rápida
To staff a whole day, run Erlang C once per interval. For each half hour, traffic in Erlangs is calls × average handle time ÷ 1,800 seconds; Erlang C finds the fewest agents that meet your service level target, and dividing by (1 − shrinkage) gives the people to schedule. Adding the intervals gives the day's agent-hours. In the illustrative example day, 2,257 calls at 240 seconds and 80/20 need a peak of 20 agents (29 scheduled at 30% shrinkage) and 220 agent-hours on the phones.
Call volumes for the day
Handle time, targets and shrinkage
Peak agents on the phones
Peak people to schedule with shrinkage
Hours are interval requirements added up. Real shifts come in whole blocks, so the hours you actually roster are usually higher.
Staffing curve for the day
Point at or tap a bar to see its figures.
Staffing by interval
Set by: Target means the service level target decides the agent count; Cap means your occupancy cap decides it; No calls means no calls were entered for the interval; Too large means the target needs an unusually large team, so check the inputs.
| Interval start | Llamadas | Traffic (Erlangs) | Agents on the phones | To schedule | Service level (%) | Average speed of answer (s) | Occupancy (%) | Set by |
|---|
How this planner calculates
The planner uses the same Erlang C method as our calculadora de Erlang C, once per interval. Traffic A = calls × average handle time ÷ interval length in seconds. For each agent count N above A it computes the Erlang B blocking probability with the stable recursion B(k) = A·B(k−1) ÷ (k + A·B(k−1)), converts it to the Erlang C probability of waiting, P(wait) = N·B ÷ (N − A·(1 − B)), and derives service level = 1 − P(wait)·e−(N−A)·T÷AHT, average speed of answer = P(wait)·AHT ÷ (N − A) and occupancy = A ÷ N. Each interval gets the smallest N that meets the service level target and, if you set one, the occupancy cap. People to schedule = N ÷ (1 − shrinkage), rounded up. Day service level and speed of answer are averages of the intervals weighted by their calls; average occupancy is workload ÷ agent-hours on the phones.
Results are model estimates based only on the numbers you enter. Erlang C assumes random arrivals, callers who never hang up and agents who can all take every call, and it treats each interval on its own. Nothing you enter leaves your browser.
How the staffing planner works
A single Erlang C calculation answers one question: how many agents does this interval need? A day is many intervals, and call volume changes across them. The planner repeats the calculation for every interval you enter, then adds the results up, so you see the shape of the day instead of one number.
- Enter the volumes. Choose half-hour or hourly intervals and the time of the first one, then paste or type the calls you expect in each interval, in time order. If your forecast is a spreadsheet column, paste it as it is.
- Set the queue parameters. Average handle time, the service level target (for example 80% within 20 seconds), shrinkage and, if you use one, a maximum occupancy.
- Read each interval. For every interval the table shows the traffic in Erlangs, the agents needed on the phones, the people to schedule after shrinkage, and the service level, average speed of answer and occupancy that staffing delivers.
- Read the day. The summary shows the peak agents and when the peak first occurs, the peak people to schedule, total agent-hours on the phones, staff hours to schedule, and the service level and speed of answer for the day as a whole, weighted by calls.
- Take it further. Copy the table as CSV into a spreadsheet to build shifts around the curve or to compare it with what you actually rostered.
For the reasoning behind a single interval, see How many call center agents do I need? or try one interval at a time in the calculadora de Erlang C.
Worked example: the example day
El Load example day button fills in an illustrative day. The volumes are made up to show how the planner behaves; they are not data from any real contact center. The day runs from 00:00 to 23:30 in half hours, with 2,257 calls in total: a handful of calls overnight, a climb from 07:00, a morning peak of 120 calls at 10:30, a softer afternoon and a long evening tail. The other inputs are the planner's defaults: 240 seconds AHT, 80% answered within 20 seconds, 30% shrinkage and no occupancy cap.
| Interval | Llamadas | Erlangs | Agents on the phones | To schedule | Answered within 20 s | Ocupación |
|---|---|---|---|---|---|---|
| 03:00 | 2 | 0.27 | 2 | 3 | 97.3% | 13.3% |
| 07:30 | 32 | 4.27 | 7 | 10 | 86.0% | 61.0% |
| 09:00 | 96 | 12.80 | 17 | 25 | 86.1% | 75.3% |
| 10:30 | 120 | 16.00 | 20 | 29 | 81.7% | 80.0% |
| 14:00 | 106 | 14.13 | 18 | 26 | 82.1% | 78.5% |
| 18:00 | 50 | 6.67 | 10 | 15 | 86.8% | 66.7% |
| 21:00 | 18 | 2.40 | 5 | 8 | 90.9% | 48.0% |
Across the whole day the planner reports:
- Peak: 20 agents on the phones, first needed at 10:00 and held until the 11:00 interval, which means 29 people scheduled at 30% shrinkage. The 10:30 interval is the same queue as the worked example in the Erlang C calculator: 120 calls at 240 seconds is 16 Erlangs, and 20 agents answer 81.7% within 20 seconds.
- Hours: the day's workload is 150.5 Erlang-hours, but covering every interval to target takes 220 agent-hours on the phones and 324 staff hours to schedule. Average occupancy is 68.4%.
- Day service level: an expected 85.6% of calls answered within 20 seconds and an average speed of answer of 12.8 seconds, weighted by calls. The day comes out above the 80% target because agents come in whole numbers: in most intervals the smallest team that meets 80% delivers more than 80%.
- Quiet hours: overnight, 2 or 3 calls per half hour still need 2 agents on the phones. One agent would answer only about 75% of the 03:00 calls within 20 seconds.
Switch the example to hourly intervals and the planner adds each pair of half hours together. The peak is still 20 agents, in the 10:00 hour, and the day needs 217 agent-hours instead of 220. On this smooth example the difference is small; on a day with sharp peaks inside the hour, hourly intervals hide more.
From interval requirements to shifts
The planner gives you a requirement curve: how many people you need in each interval. It does not build shifts. Shifts start at fixed times and last several hours, so a roster that covers every interval almost always carries more staff hours than the sum the planner shows. Use the curve to see where the peaks are, when the day ramps up and down, and how many people must be in by the busiest interval. Then fit shift start times and lengths around it, and compare the rostered hours with the planner's staff hours to see how much the shift pattern costs. Our call center workforce management guide covers the wider forecasting and scheduling process.
- Shrinkage is a single figure here. In practice breaks and meetings cluster mid-shift and training clusters on quiet days. If you know some intervals carry more planned shrinkage, run them separately with a higher figure. See what call center shrinkage is.
- Handle time is a single figure here. If AHT differs between morning and evening, or between queues, plan each part with its own value. Lowering it reduces agents in every interval; how to reduce average handle time covers the levers.
- Plan per queue. Agents who take different languages or skills are not interchangeable. Plan each queue on its own and adjust for agents who cover more than one.
Assumptions and limits
The planner inherits the assumptions of Erlang C, and adds one of its own by treating each interval separately:
- Nobody hangs up. Erlang C ignores abandonment: it assumes every caller waits until answered. Real callers abandon, which removes work, so where abandonment is significant Erlang C tends to recommend slightly more agents than you need. Erlang A vs Erlang C explains how a model with caller patience changes the answer.
- Each interval stands alone. Erlang C describes a queue that has settled at a steady rate. A backlog left at the end of a busy interval spills into the next one, which the per-interval calculation does not see. Expect the first interval after a sharp rise to be harder than the table suggests.
- Calls arrive at random at a steady rate within each interval. A marketing email or a service outage breaks this. Shorter intervals reduce the error.
- Every agent can take every call, one at a time, in order. Skills, priorities, callbacks and blended outbound work change the real picture.
- Inbound only. The model describes callers waiting in a queue. Outbound predictive dialing has a different kind of abandonment, set by dialing pace; the estimador de tasa de abandono models that trade-off.
None of this makes the result useless. Use the planner as the first answer, then compare its predictions with the service level and abandonment you actually measure per interval, and adjust your inputs where reality differs.
Where the inputs should come from
The plan is only as good as the volumes and handle time you give it. Use calls per interval and AHT per queue from your contact center platform, from the same weekday in recent weeks, rather than monthly averages. In DialerBee, each inbound queue has its own SLA target and threshold, queue depth and queue stats are live on the wallboard, and queue performance and inbound call reports cover the period afterwards. To turn the staff hours into a monthly budget, use the call center cost calculator, and see precios para ver los planes por puesto.
See the queue data behind the forecast
We will walk through inbound queues, SLA targets, the wallboard and reports on a demo tenant.
Preguntas frecuentes