Occupancy

Occupancy

Occupancy (also called agent occupancy or utilization rate) is a workforce management metric that measures the percentage of time agents spend actively handling customer contacts versus waiting for work while in a ready state. It is calculated as the ratio of handle time to total available time and reflects how efficiently staffing levels match incoming demand.

Occupancy is one of the most important — and most frequently misunderstood — metrics in contact center operations. While high occupancy appears efficient, sustained occupancy above 85-90% is associated with agent burnout, increased errors, and higher attrition. Conversely, low occupancy indicates overstaffing relative to demand. Understanding the mathematical relationship between occupancy, service level, and team size is essential for effective workforce management.

Definition and Formula

Occupancy is calculated as:

$\text{Occupancy}=\frac{\text{Total Handle Time}}{\text{Total Available Time}}\times 100\mathrm{\%}$

Where:

• Total Handle Time = talk time + hold time + after-call work (ACW) across all contacts handled during the period
• Total Available Time = handle time + idle time (time in ready state waiting for the next contact)

Occupancy does not include time spent in auxiliary states (meetings, training, breaks, coaching) — those are captured under shrinkage. Occupancy measures only what happens during the time agents are logged in and available.

Example

An agent is available for 60 minutes during an hour. They handle contacts for 48 minutes and wait idle for 12 minutes:

$\text{Occupancy}=\frac{48}{60}=80\mathrm{\%}$

Occupancy vs. Utilization vs. Productive Time

These terms are frequently conflated but measure different things:

Metric	Numerator	Denominator	What It Answers
Occupancy	Handle time	Available time (handle + idle)	"When agents are available, how busy are they?"
Utilization	Handle time	Total paid time (including shrinkage)	"What fraction of paid time is spent on contacts?"
Productive time	Handle time + productive aux	Total paid time	"What fraction of time is spent on any work activity?"

Occupancy is the narrower, more diagnostic metric: it isolates the demand-supply balance during available time, excluding the separate question of how much shrinkage occurs.

The Mathematics of Occupancy

Occupancy is not an independent variable that managers can set to a target — it is a mathematical consequence of three inputs:

1. Contact volume (arrival rate)
2. Average Handle Time (service time per contact)
3. Number of agents available

Given these inputs, the Erlang C model determines both the expected service level and the resulting occupancy. The relationship follows directly from queueing theory:

$\text{Occupancy}=\frac{\text{Traffic Intensity (Erlangs)}}{\text{Number of Agents}}=\frac{\lambda \times \text{AHT}}{N}$

Where λ is the arrival rate and N is the number of agents.

The Inverse Relationship with Service Level

Occupancy and service level move in opposite directions. To achieve a higher service level, more agents must be available relative to demand, which necessarily increases idle time and decreases occupancy. This relationship is non-linear:

Service Level Target	Approximate Occupancy (100-seat center)	Approximate Occupancy (20-seat center)
80/20	86-88%	72-78%
80/30	88-90%	75-80%
90/20	82-84%	65-72%

These are illustrative; actual values depend on AHT, call volume, and interval length.

The Team Size Effect

One of the most important — and least intuitive — properties of occupancy is its dependence on team size. Smaller teams must accept lower occupancy to maintain the same service level as larger teams. This is a direct consequence of queueing mathematics:

• A 10-agent team achieving 80/20 may operate at 60-65% occupancy
• A 50-agent team achieving 80/20 may operate at 82-85% occupancy
• A 200-agent team achieving 80/20 may operate at 90%+ occupancy

This occurs because larger pools benefit from statistical pooling — the law of large numbers smooths demand variability. Operations leaders who set uniform occupancy targets across teams of different sizes create an impossible constraint for small teams and an understaffing risk for large teams.

Optimal Occupancy Ranges

There is no single "correct" occupancy target. However, research and practitioner experience suggest general ranges:

Occupancy Range	Implications
Below 70%	Significant overstaffing; agents idle frequently. Acceptable in small teams or during low-volume intervals.
70-80%	Comfortable pace for agents; good service levels in small-to-medium teams.
80-85%	Efficient for medium-to-large teams; sustainable for most agents.
85-90%	High efficiency; approaching burnout risk for sustained periods. Typical in large, well-managed centers.
Above 90%	Agents have virtually no recovery time between contacts. Associated with increased stress, errors, after-call shortcuts, and attrition.

Burnout and Agent Well-Being

Sustained high occupancy is the leading operational driver of agent burnout. When agents move immediately from one contact to the next without recovery time, they experience:

• Emotional exhaustion from continuous customer interaction
• Reduced empathy and engagement quality
• Increased after-call work time (as agents use ACW as informal breaks)
• Higher absenteeism and voluntary attrition

This creates a negative feedback loop: agents who leave must be replaced, new agents handle fewer contacts during ramp, and remaining agents face even higher occupancy. The employee experience (EX) dimension of the CX/Cost/EX triad directly addresses this tension.

Occupancy in Multi-Skill and Blended Environments

In skill-based routing and blended environments, occupancy measurement becomes more complex:

• Multi-skill agents may have different occupancy levels for different contact types
• Blended agents handling both inbound and outbound work may show high occupancy but low service level if outbound work takes priority
• Chat agents handling concurrent sessions may show >100% "handle time" relative to clock time; occupancy must be redefined for concurrent-contact channels
• Back-office blending uses idle time between contacts for deferred work, effectively targeting higher occupancy while maintaining service level

Occupancy in Back Office and Knowledge Work

In back-office operations, occupancy takes a different form. Because work items sit in queues rather than arriving in real time, agents can maintain near-100% occupancy by continuously pulling items. The constraint shifts from "how many agents to handle arriving calls" to "how many items can N agents complete within the SLA window."

Back-office occupancy targets are typically higher (85-95%) because the work is self-paced, but the same burnout considerations apply to cognitively demanding knowledge work.

Maturity Model Position

Occupancy understanding and management evolves across maturity levels:

• Level 1 (Reactive): Occupancy not tracked, or tracked only as a daily average. High occupancy celebrated as "efficiency."
• Level 2 (Foundational): Occupancy tracked at interval level. Beginning to understand the service level tradeoff.
• Level 3 (Integrated): Occupancy targets differentiated by team size and skill group. Used in schedule optimization constraints.
• Level 4 (Optimized): Dynamic occupancy management; intraday automation adjusts to maintain target ranges. Burnout risk modeled.
• Level 5 (Adaptive): Occupancy optimization across human and AI agent pools. AI agents absorb demand spikes, protecting human agents from sustained high occupancy.

See Also

• Workforce Management — Overview of the WFM discipline
• Service Level — Primary metric that occupancy supports
• Average Handle Time — Key input to occupancy calculation
• Shrinkage — Time outside the occupancy calculation (breaks, meetings, training)
• Erlang C — Mathematical model linking occupancy to service level
• Pooling Theory — Why larger teams achieve higher occupancy at the same service level
• WFM Goals — The CX/Cost/EX triad and the burnout tension
• Adherence and Conformance — Schedule compliance metrics
• Real-Time Operations — Managing occupancy in real time