Occupancy

How occupancy works

Occupancy is an output of staffing decisions, not a lever you pull directly. For a given interval, call volume and average handle time set the workload. The number of agents you put in seats sets the staffed time. Occupancy is simply workload divided by staffed time, so it rises when volume spikes, when handle time creeps up or when you are short-staffed.

That is why workforce planners treat it as a check on the plan rather than a target to push. When you size an interval with Erlang C to hit a service level, the model tells you what occupancy you will get as a side effect. Small queues run at lower occupancy because random arrival patterns leave more gaps between calls. Large queues pool the randomness, so a 300-agent queue can run at 90% while still answering most calls in 20 seconds. A 12-agent queue at the same service level might land around 70%.

Measure it per interval and per queue, then roll it up. A daily average of 82% can hide a Monday morning at 96% and a Thursday afternoon at 60%, and those two intervals tell very different stories about your schedule.

Healthy occupancy ranges

A common benchmark for inbound voice is 75% to 88%. Below about 70%, agents spend a lot of time waiting and you are probably overstaffed for the interval, or your schedule does not follow the arrival curve. Above about 90% for sustained periods, there is almost no recovery time between calls.

Ranges shift by channel and work type:

• Inbound voice, simple transactions: 80% to 88% is typical for large queues.
• Inbound voice, complex or emotional calls (collections, insurance claims, healthcare): 75% to 82% is a safer ceiling.
• Chat with concurrency: occupancy is usually measured per session slot, and running agents above 85% across two or three concurrent chats tends to show up quickly in quality scores.
• Back-office and email: work can be queued, so 90% or higher is normal and not a burnout signal by itself.

We would not set a single occupancy goal across all of these. Set the service level per queue, staff to it, and watch occupancy as a guardrail.

Occupancy vs utilization

These two get mixed up constantly, and the difference matters when you are reading a vendor report or a client scorecard. Occupancy uses available time as the denominator: how busy agents were while they were on the phones. Utilization uses total paid or scheduled time as the denominator: how much of the shift went to handling contacts at all.

Take an agent paid for 8 hours who spends 6 hours logged in and available, of which 5 hours are handling calls. Occupancy is 5 ÷ 6 = 83%. Utilization is 5 ÷ 8 = 63%. The 2-hour gap is breaks, lunch, a team huddle and 30 minutes of coaching, which is your shrinkage. Some BPO contracts call utilization what is really occupancy, so always ask which denominator a number uses before comparing sites.

Why occupancy matters for hiring and attrition

Sustained high occupancy is one of the most reliable leading indicators of agent turnover we know of. When agents take call after call with no breathing room, handle times go up, quality drops, sick days rise, and within a few months the attrition curve bends upward. That feeds directly into employee burnout and into a recruiting team that never catches up.

For high-volume hiring this creates a trap. A center that is 15 seats short runs hot, loses more people, and falls further behind. Recruiters are asked to fill a bigger class next month, the new hires land in nesting during peak occupancy, and early tenure attrition spikes again. Breaking the cycle usually means hiring ahead of the attrition you know is coming, not after it shows up in the headcount report.

Low occupancy has a cost too. A 250-seat center running at 70% instead of 82% is carrying roughly 35 more agents than it needs. At a loaded cost of $45,000 per agent, that is over $1.5 million a year.

Common mistakes

• Setting occupancy as an agent-level goal. Agents cannot control how many calls arrive. Telling them to raise occupancy just encourages them to avoid not-ready states they genuinely need.
• Pushing occupancy to cut costs. Running a queue at 93% looks efficient on a spreadsheet and ends up costing more in overtime, abandoned calls and replacement hiring.
• Mixing denominators. Comparing one site's occupancy with another's utilization makes the better-run site look worse.
• Ignoring after-call work. If ACW is excluded from handle time, occupancy is understated and the floor will feel busier than the report says.
• Reading only daily averages. The damage happens in specific intervals. Look at the heat map.

How to manage occupancy well

Start with the forecast. Most occupancy problems are really interval forecasting problems, where the schedule puts people in seats at the wrong half hours. Then schedule to the arrival curve with staggered start times and split breaks rather than whole-team lunches.

Intraday, a real-time analyst should have clear triggers: if occupancy runs above 90% for two intervals, pull agents off training or offline work; if it falls below 70%, release coaching sessions or voluntary time off. Blended agents who can move between voice and back-office queues are the cheapest buffer you can build.

Finally, track occupancy next to attrition by team. If the teams with the highest occupancy also lose the most people in their first 90 days, you have your business case for more seats.