Call Center Outsourced evidence brief · Desk review · Published
Where does shrinkage measurement diverge between site and vendor?
A reconciliation method for site and vendor shrinkage reports, with shared categories, clocks, and a published bridge.

On this page
- Key stats
- Key takeaways
- Question and scope
- Define shrinkage categories
- Who counts as unavailable
- Paid break versus unproductive time
- Schedule changes and adherence
- Site versus vendor clock
- Reconcile the denominator
- Test a high-shrinkage week
- Report the bridge, not just the total
- Independent replay
- Privacy and access
- Limits and next step
- Decision the measure supports
- Coordinate with payroll
- Set the monthly review agenda
- What would change the conclusion
- Keep the bridge reproducible
- Put this into a support lane
- Related operating guides
- FAQs
- Sources
Key stats
- $21.53 median hourly wage for customer service representatives, May 2025 (BLS)
- 2 to 4 weeks of typical short-term on-the-job training (BLS)
- 8% of the occupation works in business support services (BLS)
Key takeaways
- Start the reconciliation with total paid hours from payroll.
- Use the same paid-break and meal-period rule in payroll and workforce reports.
- Publish a bridge from the site total to the vendor total with every adjustment named.
- Test the bridge in a holiday or outage week before accepting the annual figure.
Question and scope
This brief examines why two reports of the same call center's shrinkage can disagree. The population is one site's scheduled agent hours during a declared four-week window. The unit is a paid hour and its assigned category. Freeze the categories and the payroll source before comparing the site and vendor reports. Record the report version and the extraction time for both sides. The goal is a shared explanation of the hours, not a winner in an argument about percentages.
Define shrinkage categories
Shrinkage usually includes time paid but not available for contacts. The category list may include breaks, meals, meetings, training, coaching, system downtime, unplanned absence, and special projects. Write each definition with an inclusion and an exclusion example. A category named other is not a definition. Test the list against a week of real records before publishing it. If the same record can fit two categories, the definitions are not yet ready for a comparison.
Who counts as unavailable
An agent in a team meeting is unavailable for contacts but may still be productive for the business. An agent waiting for a system to recover is unavailable through no fault of the team. Decide whether the report measures contact availability, schedule adherence, or paid unavailability. The three questions produce different numbers. Label the report with the question it answers. A reader who assumes the wrong question will draw the wrong conclusion even when every number is correct.
Paid break versus unproductive time
A short rest period is generally counted as hours worked under federal call center guidance, while a bona fide meal period may be unpaid when the employee is relieved of duty. The payroll treatment and the workforce report must use the same rule. A report that treats a paid break as shrinkage and a payroll file that treats it as productive time will never reconcile. Document the rule in both systems and test it on a sample of records. The legal treatment and the operational category are related, but they are not the same field.
Schedule changes and adherence
A schedule change made after the roster is published can look like shrinkage in one report and adherence in another. Record the change, the approver, and the effective time. Compare the actual interval with the schedule in force at that moment, not the schedule that was published at the start of the month. Keep the change log with the reconciliation. A last-minute change that moves an agent from a contact queue to a training room should appear as training, not as unexplained absence.
Site versus vendor clock
The site may measure from the badge-in time, while the vendor measures from the first login. The difference includes the time an agent spends walking to the desk and starting the system. That time may be compensable work. State which clock each report uses and reconcile the gap explicitly. Test the difference at the start and end of a shift, not only at the middle. A clock difference that looks small per agent can become a material number across a large floor.
Reconcile the denominator
Start with total paid hours from payroll. Subtract the categories that both reports agree are unavailable. Compare the remaining available hours with the queue's logged time. Any difference should be assigned to a named category or reported as unexplained. A residual that is simply labeled variance is not a reconciliation. Set a tolerance for the unexplained amount and escalate when it is exceeded. The tolerance should be agreed before the first comparison, not after the numbers arrive.
Test a high-shrinkage week
Choose a week with a holiday, a training day, or a system outage. Recompute both reports and compare the category totals. If the difference grows in the adverse week, the reports may be treating the event differently. Document the treatment before drawing a conclusion about performance. Show the event in the bridge as its own line. A holiday that is coded as absence on one side and holiday pay on the other should be visible, not buried in a total.
Report the bridge, not just the total
Publish a bridge that shows how the site total becomes the vendor total. List each adjustment with its source and owner. A single percentage for shrinkage is not enough to explain a gap between two organizations. The bridge is the evidence that the two reports describe the same hours. Keep the bridge format stable across months so the reader can compare one period with the next. A new format every month makes the trend impossible to follow.
Independent replay
Give a second analyst the payroll extract, the roster, the queue log, and the category definitions. Ask for an independent bridge. Compare the adjustments and the residual. Differences in policy should go to the named operations and finance owners, not to the analyst who produced the more convenient number. Record the resolution and update the definition sheet. The replay should be repeatable by someone who did not attend the original meeting.
Privacy and access
Use aggregated hours where possible and keep agent-level records in the approved system. Restrict the review to the minimum fields needed and remove temporary access at close. Report any payroll field that was unavailable or could not be matched to the roster. Explain whether the gap changes a category total or only an individual record. Do not substitute a team average for a missing individual value unless the method is documented and accepted.
Limits and next step
The result depends on the payroll cut-off, the schedule version, the category definitions, and the quality of the queue log. It does not prove that one team is working harder. The next step is a shared definition sheet and a monthly bridge owned by both sides. Name the owners and the date for the first joint review. A bridge that is produced by only one side is a report, not a reconciliation.
Decision the measure supports
Use the reconciled shrinkage to decide where to add coverage, fix a schedule change process, or correct a payroll coding rule. Do not use it to rank individual agents. A gap between site and vendor is a process question before it is a performance question. Record the expected effect of the fix and review it in the next bridge. The measure is working when both sides can explain the same hours without renegotiating the definitions.
Coordinate with payroll
Payroll owns the hours, and workforce planning owns the category. The two teams should agree on the cut-off, the rounding rule, and the treatment of a shift that crosses midnight. A shift that starts on Sunday and ends on Monday can appear in different weeks if the rule is not written. Test the rule on one complex pay period. The reconciliation should be signed by both owners before it is used in a client report.
Set the monthly review agenda
Use the same agenda each month: paid hours, category changes, unexplained residual, and the one action that will reduce the gap. Keep the meeting short and evidence-led. Invite the operations owner who can change the schedule and the finance owner who can change the coding. A review without those two people produces observations but not corrections. The agenda should end with a named owner and a due date for each adjustment.
What would change the conclusion
A new payroll system, a change in the shift structure, or a new client requirement can change the bridge. State those triggers and review them at the start of each quarter. If a trigger occurs, rebuild the bridge from the source rather than adjusting the old one. A bridge that is patched without a new source check becomes a story about last year's process. The team should know when to start fresh and when to keep the current comparison.
Keep the bridge reproducible
Store the payroll extract, the roster version, the queue log, and the category sheet with the finished bridge. Name the person who produced each input and the time it was pulled. A reviewer should be able to rebuild the same totals without asking for a meeting. If an input cannot be retrieved, mark the bridge as partial. A reconciliation that depends on a file someone kept on a laptop is not repeatable, and it will not survive a change of analyst.
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.