Call Center Staffing Calculator

The best staffing calculator is one that uses Erlang C rather than a simple ratio, asks for shrinkage rather than pretending agents are on the phones 100% of the time, and shows you the sensitivity around the answer instead of a single number. This one does all three, free and without signup. For ongoing operations a workforce management system goes further — intraday forecasting, scheduling, adherence — but for sizing a queue or sanity-checking a quote, Erlang C is the industry-standard starting point.

Erlang C is a queueing formula, published by A.K. Erlang in 1917 for telephone exchanges, that gives the probability a caller has to wait given the number of servers and the offered load. It remains the standard model for call center staffing because its assumptions — random arrivals, queuing callers, interchangeable agents — are close enough to reality that its answers are useful, and its errors are predictable: it slightly overstates staff needed when callers abandon, which makes it a safely conservative planning tool.

Divide 3,600 seconds by your average handle time, then multiply by a sustainable occupancy — around 85%. At a 300-second handle time that is roughly ten calls an hour sustained. But be careful with the arithmetic's implication: you cannot staff by dividing total calls by calls-per-agent, because calls arrive randomly, not on schedule. That shortcut ignores queueing entirely and understaffs every queue it touches — which is precisely why Erlang C exists.

The most common convention is 80/20 — 80% of calls answered within 20 seconds — but it is a convention, not a law. Sales lines and emergency queues justify tighter targets because a caller who waits is revenue or safety lost; back-office and low-urgency support lines run acceptably at 80/30 or 70/30 for materially less money. The right question is what an extra second of average wait costs your business, and it has a different answer per queue.

Shrinkage is every hour you pay for that is not available to take calls: breaks, training, coaching, meetings, absence, and system time. It typically runs 25–35% in a well-run operation, which is why an eight-agent requirement is an eleven-or-twelve-agent schedule. Underestimating shrinkage is the single most common reason a mathematically correct staffing plan fails on the floor — the model was right about the queue and wrong about the humans.

Occupancy is the share of an agent's logged-in time spent handling contacts rather than waiting for one. It sounds like efficiency, and above roughly 90% it becomes the opposite: back-to-back calls for a full shift drive burnout, attrition and creeping handle times, which degrade the very inputs the staffing model was built on. Small teams run hotter than large ones at the same service level — that is the queueing math, not a management failure — which is one reason pooling volume with an outsourcer is often cheaper than it looks.

For email and other deferred work, no — those are backlog problems, not queueing problems, and simple capacity arithmetic answers them. For live chat it works with one adjustment: agents handle several chats concurrently, so divide your handle time by a realistic concurrency factor (usually 1.5–2.5, not the 3–4 vendors quote) before entering it. The service level logic is the same; the effective handle time is what changes.