Philippines staffing guide · Published
Write a rework log for an outsourced quality queue
Rework consumes capacity but rarely appears in a quality score. A short log records the cause, the minutes, and the repeat pattern so the fix targets the process.

The short answer
A rework log works when each correction records the cause, the minutes, and the source of the original instruction. Review causes rather than individuals, connect the log to the quality sample, and change one upstream step before adding more review.
Why rework stays invisible
A quality review usually records whether the output passed. It rarely records what happened after a failure, even though correction consumes the same capacity as new work. The reviewer edits the draft, the coordinator rechecks it, and the queue absorbs the delay.
None of that effort appears in the pass rate. The correction is real work, and the log should treat it as such.
The hidden cost is larger when the same correction repeats. A formatting fix may take four minutes, but the same four minutes across a week is a meaningful share of a role.
A rework log makes the repetition visible without turning the review into a blame exercise. The goal is to find the instruction that produced the error.
Define one rework event
Write a definition that the team can apply without arguing. A rework event starts when an output is returned for a change and ends when the corrected version is accepted.
Count the minutes spent by both the reviewer and the coordinator. Do not count a clarification that happens before the output is submitted, because that is a question rather than a correction.
Keep the definition narrow enough to count consistently. A change requested by the customer is rework only if the original instruction was clear and the output missed it.
A change caused by a new customer requirement is a scope change and belongs in a different log. Separating the two prevents the quality queue from absorbing commercial decisions.
Capture the cause, not the person
Record a cause from a short list rather than a free-text story. Useful causes include a missing source document, an unclear acceptance example, a tool error, a wrong template, and a judgment the coordinator was not authorized to make.
The list should fit on one line beside the case. A cause that requires a paragraph will not be recorded during a busy shift.
Avoid names in the cause field. The log exists to improve the instruction, not to rank people, and a named log invites defensive recording.
If a person needs coaching, that conversation belongs in a separate review. The log can still link to the case and the reviewer so the evidence is available when a pattern appears.
Group the log by cause each week
Review the log weekly and group the events by cause. Count the events and the minutes for each group. A cause with many short events may cost more than a cause with one long event, and the counts make that visible.
Sort by total minutes rather than by the loudest complaint. Bring the grouped counts to the review instead of a narrative of individual cases.
Look for causes that cluster around one client, one template, or one shift. A cluster is a signal that the process, not the individual case, needs attention. Record the cluster and the evidence in the weekly note.
One clear cluster is a better starting point than a long list of unrelated fixes. A cluster that persists for two weeks deserves a named owner and a repair date.
Connect the log to the review sample
Use the review sample to estimate the rework rate, not to prove that every item was perfect. If the sample finds three failures in forty items, the rate is a planning input rather than a verdict on the whole queue.
Compare the rate across weeks and note the sample size beside it. A rate without a sample size is easy to misread.
Check whether rework is rising while the pass rate stays flat. That pattern can mean the sample is missing a class of errors or that corrections are being made without being recorded.
Open a few corrected cases and compare them with the log. The gap between the two is often the most useful finding of the month.
Set a repair rule and close the loop
Choose one upstream repair for each review cycle. The repair might be a better example, a checklist line, a template field, or a clearer approval range. Assign an owner and a date, and state how the effect will be measured.
A repair without a measurement becomes another opinion. Write the expected effect beside the repair so the next review has a comparison.
Close the loop with the coordinator who did the rework. Tell them what changed and why their correction mattered. People record rework more honestly when the log produces a visible improvement rather than another meeting.
Keep the rule simple, review it after a month, and remove fields that nobody uses. A visible change is the strongest reason to keep recording the next correction.
Keep the log simple enough to survive
A rework log fails when the entry takes longer than the correction. Use a form with five fields and a save button that stays open during the review.
Let the reviewer record the cause while the case is in front of them. A short entry completed in the moment is worth more than a detailed one written from memory.
Review the fields after a month and remove any that nobody uses. Add one field only when a decision needs it, such as the customer promise or the rework owner.
Keep the log in the same tool as the queue so the case and the cause stay linked. A separate spreadsheet will drift away from the work.
Questions to copy for the sales call
- "Which correction consumed the most minutes last month?"
- "What cause appears most often in the log?"
- "Which template or instruction produced the repeated error?"
- "What will we change upstream, and how will we measure it?"
Sources
- SBA: Manage your business: Practical guidance on quality, roles, and management routines.
- NIST SP 800-53 Rev. 5: Control framework that treats review and correction as documented activities.
- CISA: Cybersecurity Best Practices: Reference for recording and reviewing operational exceptions.
Common questions
Does every correction need a log entry?
Log corrections that take time or repeat. A single typo fix is not worth an entry, but the same template error three times is a pattern worth counting.
Should the log include the customer name?
Use a case reference instead of a name. The cause and the minutes matter more than the identity, and a reference keeps the log usable in a shared review.
How many causes should the list hold?
Keep five to seven causes that cover most events. Add an other category for the rest and review it monthly to see whether a new cause has become common.
Who owns the weekly review?
The person who can change the instruction or the template. A coordinator can maintain the log, but the repair needs an owner with authority over the process.