Occupancy: the metric that quietly burns out your agents
Introduction
Occupancy is one of the most important numbers in workforce planning and one of the most misunderstood. Managed well, it tells you whether your staffing is efficient. Managed badly — chased as a target in its own right — it becomes the quiet engine of burnout, attrition and service-level volatility. This article explains what occupancy actually measures, where the safe range sits, and why pushing it too high costs far more than it saves.
What occupancy actually is
Occupancy is the proportion of an agent’s logged-in, available time that is spent actually handling contacts — talking, holding, and wrapping — rather than waiting for the next one to arrive. If an agent is logged in for an hour and spends 50 minutes on contacts and 10 minutes waiting, occupancy is around 83%. Crucially, it is not the same as utilisation or productivity, which usually fold in shrinkage and non-contact activities. Occupancy is specifically about how much of the “ready” time is filled with work.
Why it’s a consequence, not a lever
Here is the point most operations miss: occupancy is an output of your staffing and your service-level target, not an input you set. For a given offered load, the maths of queueing fixes the relationship. If you staff to hit 80% answered in 20 seconds, occupancy lands wherever the maths puts it — and that figure rises as the operation gets bigger, because large queues are inherently more efficient (the pooling effect). A 20-seat team hitting service level might run at 80% occupancy; a 500-seat operation hitting the same target might run at 90% on identical assumptions. That’s why a single occupancy target applied across teams of different sizes is meaningless.
The trade-off with service level
Occupancy and service level pull against each other. To give callers a fast answer you need agents waiting and ready, which means some idle time — lower occupancy. Squeeze that idle time out to lift occupancy and you remove the very buffer that absorbs random arrival spikes, so service level becomes volatile and abandonment climbs. The slack that looks like waste on an efficiency dashboard is the thing keeping your service level stable. There is real idle time worth recovering through better scheduling and shrinkage control, but beyond a point you are trading customer experience for a vanity number.
The human cost above the line
Sustained high occupancy is a wellbeing problem before it is a service one. Between-contact seconds are how agents reset after a difficult call, take a breath, finish a thought. Strip them away and you get rising error rates, longer-term sickness, disengagement and attrition — and attrition is enormously expensive. Worse, it’s self-reinforcing: every leaver pushes occupancy higher for those who remain, which drives the next round of leavers. Plenty of operations have run a team into the ground chasing a 95% occupancy target without ever connecting it to the attrition line right next to it.
What to do with it
Treat occupancy as a health check, not a target. Watch it by team and interval, set a sensible ceiling for sustained periods (many operations get nervous above the high-80s for voice, and lower for emotionally heavy work), and when it runs hot, read it as a signal that you are under-staffed or your shrinkage assumptions are optimistic — not as proof your people could simply try harder. Recover genuine idle time through scheduling that matches the curve and through honest shrinkage planning, but protect the breathing room that keeps both your service level and your people stable.
Related: see how occupancy falls out of the staffing maths in the Erlang C calculator, and why large queues run more efficiently.