Occupancy vs Utilization: The Difference and Why It Matters
Occupancy and utilization are the most frequently swapped terms in workforce management, and the swap is not harmless. Teams set "occupancy" targets that are actually utilization targets, beat up agents over a number that schedule design controls, and build staffing plans that assume away the difference. The two metrics share a numerator family and differ in denominator, and in WFM, the denominator is the whole story.
The two formulas
Occupancy measures how busy agents are while they are available for queue work:
- Occupancy % = contact handling time ÷ (contact handling time + available/idle time) × 100
The denominator is logged-in, queue-facing time only: talking, in wrap, or sitting available waiting for the next contact. Breaks, lunches, meetings, training, and coaching are not in it.
Utilization measures how much of paid time goes to handling work:
- Utilization % = contact handling time ÷ total paid (scheduled) time × 100
The denominator is the whole paid day: everything occupancy counts, plus every off-queue activity that shrinkage tracks.
Because utilization's denominator is strictly larger, utilization is always lower than occupancy for the same agent on the same day. The gap between them is, almost exactly, your shrinkage.
A worked example
Take an agent on an 8-hour (480-minute) paid shift:
- 300 minutes handling contacts (talk plus hold plus wrap)
- 60 minutes available, waiting for contacts
- 120 minutes off-queue: two breaks, lunch, a team meeting, a coaching session
Occupancy = 300 ÷ (300 + 60) = 83.3 percent. Utilization = 300 ÷ 480 = 62.5 percent. Same agent, same day, twenty-point spread, and both numbers are healthy. Notice the off-queue 120 minutes is exactly 25 percent of paid time: this agent's day embodies 25 percent shrinkage, and the relationship holds in general:
- Utilization ≈ occupancy × (1 − shrinkage)
Run the check: 0.833 × 0.75 = 62.5 percent. If you know any two of occupancy, utilization, and shrinkage, the third is determined. That identity is also a great data-quality probe: when reported numbers violate it badly, some system is mislabeling states. (Our free shrinkage calculator makes the off-queue side of this equation concrete.)
Why the distinction changes decisions
Occupancy is a queue-health and burnout metric. It is driven by the relationship between offered load and staffed agents in each interval, which makes it mostly a planning outcome, not an agent behavior. Sustained occupancy in the low-to-mid 80s for voice is the conventional healthy zone. Sustained occupancy above 90 percent means agents finish a contact and the next one is already waiting, every time, for hours: AHT creeps up (humans pace themselves), errors rise, and attrition follows. Chronically low occupancy, under 70 percent, means overstaffing or fragmented intervals, and it carries its own morale cost: agents marinating in idle time between random contacts.
The critical management point: an agent cannot fix their occupancy. If the interval is overstaffed, occupancy is low no matter how hard anyone works. Coaching an agent over occupancy is coaching them over your own staffing plan.
Utilization is a cost and capacity metric. It answers the CFO question, of every paid hour, how much produced handled contacts?, and it is moved by different levers: shrinkage discipline (meeting load, training waves, absence) and schedule design, far more than by anything that happens on the queue. A center with 83 percent occupancy and 30 percent shrinkage runs about 58 percent utilization; trimming shrinkage to 26 percent lifts utilization past 61 percent with zero change in queue pressure. That is the honest way to improve utilization. The dishonest way, pushing occupancy toward the ceiling, buys the same points by burning the floor.
The classic confusion failures
- Setting a 90 percent "occupancy" target that is really utilization. To hit 90 percent utilization at 30 percent shrinkage, occupancy would have to be 90 ÷ 0.70 ≈ 129 percent, mathematically impossible. Teams that set such targets discover their floor "fails" forever, then either quietly redefine the metric or grind agents against an unreachable number.
- Benchmarking against the wrong metric. Published voice benchmarks in the low 80s refer to occupancy. Comparing your 62 percent utilization against an 83 percent occupancy benchmark manufactures a phantom 20-point crisis.
- Staffing with occupancy but budgeting with utilization, silently. Finance models headcount from utilization; WFM staffs intervals with occupancy targets. Both are correct in their domain, but when the two functions trade numbers without labeling denominators, headcount plans drift by exactly the shrinkage factor, which is how a budget ends up 30 percent light.
- Treating either as an agent KPI. Occupancy is a planning outcome; utilization is mostly a shrinkage outcome. The agent-controllable metric in this family is adherence, whether scheduled time was spent as scheduled, which is why adherence, not occupancy, belongs on the agent scorecard.
Healthy ranges, with reasoning
For voice queues, plan occupancy between 80 and 90 percent, and treat 85 as the standard planning input: below 80 you are usually paying for idle time, above 90 you are borrowing from AHT and attrition. Asynchronous channels (chat with concurrency, email, back office) can run higher because waiting work buffers smoothly. Utilization follows from occupancy and shrinkage rather than deserving its own independent target: at healthy occupancy (85 percent) and honest shrinkage (25 to 35 percent), utilization lands in the 55 to 64 percent range. A "low" utilization number is not a problem statement by itself; decompose it first. Low because occupancy is low? Staffing-to-load problem. Low because shrinkage is high? Calendar and absence problem. The fixes share no overlap.
This decomposition is also why occupancy belongs in the staffing math itself, as a guardrail. When QueuePilot's Forecast Lab converts a forecast into required agents, the occupancy cap is an explicit input: the requirement is never allowed to assume agents busier than the target, and planned versus actual occupancy is tracked per interval so the early burnout signal, weeks of intervals running above 90, is visible while it is still cheap to fix.
The two-line summary
Occupancy: handling time over queue-available time, the burnout and queue-health dial, owned by planners. Utilization: handling time over paid time, the cost dial, owned mostly by the shrinkage calendar. Label your denominators, target each metric with its own lever, and never put either one on an agent's scorecard.