Agent Occupancy Rate
Definition
Agent Occupancy Rate
Agent occupancy rate is the share of an agent’s logged-in time spent on live work: talk, hold, and wrap-up. It shows how hard each staffed hour is used, and paid breaks stay out of the divisor. Most teams just call it occupancy.
Be straight about one thing up front. This is the same metric OA already covers under agent occupancy, and under the occupancy rate entry listed below. Different names, one formula.
What this page adds is the practical side: the bands managers actually steer by, the trade-off against service level, and the line between occupancy and utilisation that trips up a lot of new team leads.
Key takeaways
- Occupancy is handling time divided by logged-in available time, with scheduled breaks stripped out of the divisor.
- Roughly 75–85% is the usual healthy window for a voice team.
- Push past about 90% and burnout, sick days and attrition follow.
- Occupancy and service level pull against each other, so you cannot max both at once.
How it works
Occupancy answers one question: of the time an agent was signed on and available, how much went into actual contacts? Add talk time, hold time and after-call work, divide by logged-in time less scheduled breaks, then multiply by 100.
The top line is easy to picture. It is every second the agent is tied to a contact, including the wrap-up that happens after the caller has already hung up.
That wrap-up counts. It is why after-call work belongs inside the numerator, and why teams that exclude it report an occupancy figure far lower than the one their agents are living.
The divisor is where teams argue. Paid rest breaks are hours worked under 29 CFR 785.18, which covers rest periods running from 5 minutes to about 20 minutes.
So a paid break sits inside paid hours but outside the occupancy divisor. The agent is on the clock, but they are not available to take a contact, and occupancy only measures available time.
Managers steer by bands, not by a single target. Here is how the usual ranges read on a voice floor:
| Occupancy band | What it usually signals | Typical action |
|---|---|---|
| Below 70% | overstaffed for the volume | trim the roster or add offline work |
| 70–75% | slack sitting in the schedule | tighten the interval plan |
| 75–85% | the healthy working window | hold steady and monitor |
| 85–90% | pressure building on the floor | watch wrap times and aux abuse |
| Above 90% | burnout territory | add heads or shed volume now |
Those bands are guides, not law. A short-contact sales queue tolerates more pressure than a complex technical desk where every call needs thinking time and careful notes.
Here is the part that catches people out — occupancy and answer speed pull in opposite directions. To answer fast you need agents sitting idle, waiting, and idle time is exactly what drags occupancy down.
Chase 95% occupancy and your queue times will blow out. Chase an instant answer on every contact and you will pay for a lot of empty seats. The job is picking the point between them.
That trade-off is what staffing maths exists for. A workforce management team models arrival patterns, then sets headcount so the queue clears at target speed without parking anyone at 95% all shift.
Utilisation is the wider cousin, and the two get mixed up constantly. Utilisation keeps paid non-productive time inside its divisor, so training, coaching and huddles all count against it.
Occupancy ignores that time completely. Same shift, same agent, two different numbers — utilisation always reads lower, and quoting one when your boss wanted the other is a fast way to lose an argument.
Read occupancy by interval, not by day. A daily average of 80% can hide a brutal 96% lunchtime peak and a dead 60% evening, and the agents will remember the peak.
Watch the aux codes too. When occupancy runs hot, agents protect themselves by parking in an unavailable state, so your reported figure falls while the real pressure stays — the number improves, the floor does not.
Examples
Occupancy reads differently by channel, by shift and by employer. A small voice team at midnight looks overstaffed on paper. A back-office queue can run hot without hurting anyone. Context decides whether a number is good or bad.
A US voice floor watching cost per hour. The US Bureau of Labor Statistics put the median hourly wage for customer service representatives at $20.59 in May 2024.
At that rate, idle time carries a visible price tag, so finance tends to push occupancy up. The same source projects employment falling 5 percent from 2024 to 2034, with about 341,700 openings each year on average.
Churn that heavy changes the argument. On a floor already replacing people every year, running at 92% turns a morale problem into a recruitment bill — and the recruitment bill usually beats the idle-time saving.
A public-sector service desk. The US federal customer experience programme was set up under Executive Order 14058 in 2021, which directed 17 agencies to take 36 specific actions on service delivery.
Programmes judged that way are measured on what the citizen experienced, not on how busy the desk looked. OMB Circular A-11, Part 6, Section 280 carries the annual guidance. Occupancy stays an internal dial there, never the headline.
A blended offshore chat team. Chat agents handle two or three conversations at once, so raw occupancy can read above 100% when the counter adds concurrent sessions together.
Cap the numerator at wall-clock time before anyone reports it. Otherwise the metric stops meaning anything, and a team that is genuinely stretched looks identical to one that is coasting.
A seasonal promotion peak. Volume jumps for a fortnight while the roster stays flat, and occupancy climbs with it. A few days above 90% is survivable. Three weeks is not, and your attrition report will say so a quarter later.
Each of those floors runs the same formula and lands somewhere different. That is the point — occupancy is a diagnostic you read against context, not a target you set once, publish on a wallboard and defend forever.
Related terms
Occupancy sits inside a small family of staffing metrics, and mixing them up is the most common reporting error on a new floor. These five turn up in the same weekly pack, so learn the differences early.
- Occupancy Rate: the same measure under its shorter and far more common name.
- Agent Utilization: a wider ratio that keeps paid non-productive time inside the divisor.
- Shrinkage: the share of paid hours lost to breaks, training, leave and meetings.
- Average Handle Time (AHT): the mean length of a single contact, wrap included.
- Erlang C: the queuing formula that ties staffing levels, occupancy and answer speed together.
FAQ
What is a good agent occupancy rate?
Roughly 75–85% is the window most voice teams aim for. Above about 90%, agents get no recovery time between contacts. Below about 70%, you are paying for idle capacity.
How do you calculate agent occupancy rate?
Add talk time, hold time and after-call work, then divide by logged-in available time with scheduled breaks removed. Multiply by 100 to get a percentage.
Why can you not run high occupancy and high service level at once?
Fast answer speed needs spare agents waiting for the next contact. That waiting is idle time, and idle time is exactly what pushes occupancy down.
What is the difference between occupancy and utilisation?
Utilisation counts paid non-productive time, such as training and meetings, inside its divisor. Occupancy leaves that time out, so it reads higher for the same shift.
Does agent occupancy rate include breaks?
No, scheduled breaks come out of the divisor even though they are paid time.
If you are building or benchmarking an offshore floor, the Outsource Accelerator hubs are the place to compare how providers actually staff it.
Independent
