Call Center Outsourced evidence brief · Desk review · Published
How should a mixed-skill queue report occupancy without hiding idle capacity?
A measurement protocol for reporting occupancy across mixed-skill queues without letting a blended average hide idle capacity.

On this page
- Key stats
- Key takeaways
- Question and scope
- Why blended occupancy misleads
- Define the unit and the clock
- Separate productive and available time
- Allocate idle capacity by skill
- Handle cross-skilled agents
- Set a reporting threshold
- Build the reconciliation table
- Test with an adverse week
- Independent replay
- Privacy and access
- Limits and next step
- Decision the measure supports
- Coordinate with quality and training
- Publish the monthly view
- What would change the conclusion
- Put this into a support lane
- Related operating guides
- FAQs
- Sources
Key stats
- 2.67 million U.S. customer service representative jobs in 2025 (BLS)
- 8% of the occupation works in business support services (BLS)
- 289,500 projected annual openings, 2025 to 2035 (BLS)
Key takeaways
- Report occupancy by skill before showing any blended site figure.
- Allocate cross-skilled available time with a written rule and show the residual.
- Suppress figures below a minimum volume instead of reporting unstable percentages.
- Test the measure on a holiday or outage week before using it for a decision.
Question and scope
This brief asks how a mixed-skill queue should report occupancy when agents handle more than one contact type. The population is one contact center site during a declared four-week window. The unit is a reporting interval, not an individual agent or a single call. Freeze the interval length and the skill definitions before looking at the results. Record the source of every input so a later analyst can reproduce the calculation. The question is not which team works hardest, but which queue has usable capacity.
Why blended occupancy misleads
A blended occupancy figure divides total talk time by total paid time across every skill. That calculation can look healthy while one skill waits for an available agent and another skill sits underused. The average hides the queue a customer actually experiences. It also hides the moment when a cross-skilled agent is nominally available but assigned to the wrong queue. Report occupancy by skill and show the blend only as a secondary view. A blended number is useful for a site summary, never for a staffing decision on its own.
Define the unit and the clock
State whether the interval is fifteen minutes, thirty minutes, or an hour. Use the same clock for offered contacts, handled contacts, and paid time. A measure built from a queue report that starts on the hour and a payroll file that starts on the half hour will never reconcile. Document the source and the extraction time for each input. Write down the time zone and the daylight-saving treatment. A one-hour shift in the clock can look like a capacity change when it is only a reporting artifact.
Separate productive and available time
Productive time includes talk, hold, after-call work, and approved ancillary tasks. Available time is paid time when the agent could accept a contact but did not. Keep breaks, meetings, training, and system outages in separate categories. A single unproductive bucket makes it impossible to see whether idle capacity was usable or required. Split planned and unplanned events so the reader can tell a roster choice from an outage. The category definitions should be stable across the comparison period.
Allocate idle capacity by skill
An agent who is available for billing and retention is not equally available to both queues at every moment. Allocate available time by the routing rules in force during the interval. If the rules change at noon, split the interval or document the change. A simple proportional split is acceptable only when the routing is genuinely proportional. Where the router gives priority to one skill, assign the available time to that skill and show the secondary capacity separately. The allocation rule belongs in the report, not in an analyst's notebook.
Handle cross-skilled agents
Cross-skilled agents create a counting problem because their paid time can support more than one queue. Assign each interval to the skill the agent actually served, or use a documented allocation rule. Do not count the same available minute in two queues and then present the sum as a site total. The reconciliation table should show the allocation and the residual. A residual is acceptable when it is labeled and explained. An unexplained residual means the report is not yet ready for a capacity decision.
Set a reporting threshold
Define a minimum volume below which a skill-level occupancy figure is not reported. A queue with three contacts in an interval can produce a dramatic percentage that says nothing useful. Use a threshold based on contacts and paid hours, and mark the excluded intervals rather than dropping them silently. The threshold should be set before the results are visible. Publish the excluded count beside the included count so the reader can see how much of the period was omitted. A suppressed interval is not a zero and should never be charted as one.
Build the reconciliation table
The table should show paid hours, productive hours, available hours, handled contacts, and occupancy for each skill. Add a column for excluded time and one for the allocation rule. The site total should equal the sum of the skill rows after the allocation, and the difference should be explained rather than averaged away. Include a row for time that cannot be assigned to a skill. The table should tie to payroll and to the queue report on the same page. If the three sources disagree, show the bridge before showing the percentage.
Test with an adverse week
Choose a week with a holiday, a system outage, or a training day. Recompute the occupancy and compare it with the original report. If the adverse week changes the ranking of the skills, the measure is not stable enough for a performance claim. Report both the normal and adverse results. Explain which category absorbed the unusual time and who approved the treatment. A measure that only works in a quiet week will mislead the first time the queue is stressed.
Independent replay
Give a second analyst the same routing rules, source extracts, and definitions. Ask for an independent calculation without seeing the first result. Compare the skill totals and the allocation. Differences in the allocation rule should be resolved by the named operations owner, not by choosing the more flattering number. Record the resolution and update the definition sheet. The replay is evidence that the measure can survive a change of analyst, which is the first test of a durable operating metric.
Privacy and access
Use aggregated interval data and avoid agent-level identifiers unless the review requires them. Restrict the working file to the analysts and the operations owner, and remove temporary access when the review closes. Report any field that was unavailable or redacted. Explain whether the redaction changes the ability to reconcile the total. Keep the extraction query with the report so the next review can use the same population. A measure that depends on a personal spreadsheet is difficult to audit and easy to misread.
Limits and next step
The result depends on the routing rules, the interval length, the allocation method, and the quality of the source extracts. It does not prove that one skill is better managed than another. The next step is a written decision about which skill needs capacity, coaching, or a routing change. Name the owner and the date for the follow-up. Set a review point after the change so the team can see whether the capacity moved as intended. A measurement without a decision is only a longer report.
Decision the measure supports
Use the skill-level occupancy to decide where to add hours, move cross-skilled agents, or change the routing priority. Do not use it to rank individual agents. A queue-level measure should lead to a queue-level decision, and the decision should name the owner and the date. Record the expected effect before the change so the next report can test it. Keep the blended figure for the executive summary but never for the staffing call. The measure earns trust when the decision and the result are visible together.
Coordinate with quality and training
Occupancy is not the only reason a queue underperforms. A skill with low occupancy may still have long waits because the available agents lack the training to handle the contact. Share the skill-level view with quality and training before changing the roster. Ask whether the queue is short of people or short of people who can resolve the work. The answer may be a coaching plan rather than an extra shift. A capacity decision made without that conversation can add hours to a queue that still cannot answer the question.
Publish the monthly view
Publish a one-page monthly view that shows occupancy by skill, the excluded intervals, and the allocation rule. Keep the format stable so the reader can compare one month with the next. Add a short note for any change in routing or skill definitions. The note is important because a change in the rule can move the percentage without any change in staffing. A monthly view that hides the rule change will be read as a performance change, and the argument will start from the wrong premise.
What would change the conclusion
State the conditions that would change the recommendation. A new routing priority, a change in contact mix, or a sustained rise in repeat contacts could make the current allocation obsolete. Review those conditions at the next capacity meeting. A measure that is never revisited becomes a comfortable story about the past. The team should know which signals would cause it to reopen the decision, and those signals should be visible before the meeting begins.
Put this into a support lane
Bring one queue, its reporting extract, and the manager who owns the operating decision.
Review the queueRelated operating guides
FAQs
Does this study rank agents?
No. It describes a measurement design and the decisions the resulting evidence can support.
What counts as closure?
A reviewed result linked to the source, the definition, and the named owner who accepted it.