Research brief · Published
Manual payroll adjustment register review
A review design for testing whether every manual payroll adjustment is numbered, supported, approved by a second person, and reconciled to the payroll register after the run.

Research finding
Build the population from manual payroll edits, not from the exception report alone.
Methodology
This brief triangulates the headline measure against official Philippine government, regulatory, development, and labor sources. It translates evidence into an operating control and separates context from recommendations.
| Measure | Interpretation |
|---|---|
| Six evidence fields checked per adjustment. | Context signal for planning, not a promise about an individual worker or provider. |
| 4 source records | Primary source links are listed and numbered below for review. |
Key takeaways
- Build the population from manual payroll edits, not from the exception report alone.
- Test whether an entry existed before the payroll record changed.
- Check that support covers the amount, the employee, and the pay period.
- Confirm that the preparer and approver are different named people.
- Reconcile every posted adjustment to a register entry and close open items with dates.
Question and scope
This review asks whether manual payroll adjustments are authorized, supported, and reconciled before and after the payroll run. The scope is every manual change to hours, rates, deductions, taxes, or net pay outside the normal import. The unit is one adjustment entry for one employee or group in one pay period. The design tests the control around the change, not the accounting treatment. Record the review period, the payroll systems covered, and the person who owns the register.
Population and unit
Build the population from the adjustment register, the payroll audit log, and the correction tickets for the review period. Include entries that were reversed or rejected as well as those that posted. The unit is one adjustment number. A population built only from the register will miss unregistered edits, so the audit log must be part of the frame. List the audit log source and the extraction date so an unregistered edit can be traced.
Register design
Check that the register assigns a unique number before the edit and captures the pay period, employee or group, field changed, old value, new value, reason code, support reference, preparer, approver, and posting result. Confirm that the number cannot be reused. A register without a required field set produces incomplete entries that cannot be analyzed. A register that is separate from the payroll system should still be backed up and access controlled.
Entry completeness
Sample entries and confirm that every required field is populated. Look for free-text reasons that do not match a code and for support references that point to a general folder rather than a document. Check that the old and new values are both recorded. An entry that shows only the new amount cannot prove what changed. Mark each missing field as a finding rather than filling it in from memory during the review.
Reason coding
Review the reason code list and the distribution of codes in the sample. A large share of entries in a catch-all code weakens the analysis. Check whether repeated codes point to a source-system problem. Confirm that the code definitions are available to preparers and reviewers, not held by one administrator. Ask two preparers to explain the same code and compare their answers with the written definition.
Supporting evidence
Confirm that each entry has a document that supports the amount and the dates, such as a signed timesheet, an approval email, a corrected offer letter, or a benefit election record. Check that the document names the employee and the period. A document that supports a different amount is not sufficient evidence. Keep the support document link stable so it still resolves when the reviewer opens the entry later.
Separation of duties
Compare the preparer and approver on each entry and confirm that they are different named people. Check the payroll system roles as well as the register, because a shared login can hide the same person acting twice. Record any case where one person prepared, approved, and released the change. That is a control gap even when the amount is correct. Record the system role and the person behind it, not only the name typed in the register.
Approval thresholds
Review the value and risk thresholds that require a higher approval. Confirm that the threshold is published and applied consistently to retroactive pay, deduction reversals, and large corrections. Test an entry just below and just above the threshold. A threshold that exists only in practice is not a control. Test a low-value entry and a high-value entry to confirm the threshold is applied by amount, not by habit.
Cutoff control
Check the time between the adjustment request and the payroll release. Confirm that manual entries have a cutoff earlier than the release time. Look for entries approved after the cutoff and released anyway. A last-minute approval is often a signature without a review. Compare the request time, approval time, and release time for each sampled entry in the affected pay period.
Post-run reconciliation
Compare the count and value of register entries with the adjustments visible in the payroll register after the run. Match each entry by number and confirm the employee, period, field, and amount. Investigate posted changes with no entry and entries that did not post. The reconciliation is the proof that the register describes the live payroll. Save the reconciliation with the run packet and note who performed the match.
Direction and allocation
Check the sign and allocation of each adjustment, not only the net total. A positive and a negative entry can net to zero while hiding two separate errors. Confirm the cost center and the general ledger account. A change that reaches net pay but misses the ledger creates a later reconciliation problem. Check the employee year-to-date totals as a second test of the allocation.
Ledger alignment
Compare the adjustment value with the payroll liability accounts and the funding request. Confirm that a retroactive payment is coded to the correct period and cost center. Check whether the adjustment creates a new liability that must be funded separately. Record the accounting owner who confirmed the treatment. Ask the accounting owner to confirm the period and cost center for any retroactive item.
Open-item closure
Confirm that every item left open at the end of the run has an owner and a due date. Check that the item is visible in the next run and that it closes with evidence. A note that an item will be handled later is not a closure. The reviewer should be able to see the status without reading a mailbox. Review the open-item list at the start of the next run, not only at month end.
Monthly pattern review
Review counts by reason code, department, preparer, and pay period. A cluster in one team points to a timekeeping or approval problem, while a cluster in one field points to a system or training problem. Compare the trend with prior months. The review should end with one or two decisions, not a longer report. Compare the top cause with the prior month and record whether the action reduced it.
Provider handoff
When a provider prepares adjustments, confirm that the employer approver is named and that the provider cannot release a change without approval. Compare the provider's adjustment log with the employer register. Ask how the provider handles an urgent correction after the cutoff. Record any case where the provider acted first and the employer approved later. Keep the provider's log and the employer register side by side for the sample period.
Small-team exceptions
A small team may not have a separate approver for every correction. Record the exception and the compensating check, such as an owner review of the support and the post-run match. The check should happen before release. A second review by the same person is not a separation of duties. Record the exception in the register so the compensating review leaves its own evidence.
Sampling
Sample entries across reason codes, value bands, preparers, and pay periods. Include a reversal, a retroactive payment, and an entry that did not post. Compare the sample with the audit log to find unregistered edits. Record the selection rule and the population count so the work can be repeated. Include one entry from each preparer and one entry that was rejected.
Findings and severity
Classify each gap by the value exposed, the number of pay periods, and whether the change reached employees without approval. An unregistered edit that posted is more severe than a missing field caught before release. Record the compensating check and the owner who accepted the risk. State the value at risk and the number of employees affected for each finding.
Limitations
The review depends on the audit log and the register being complete, which is the very thing under test. Some systems overwrite values without preserving the old amount. The design cannot prove intent or detect an edit made outside both records. State the limits and use the payroll register as a second source where possible. Note which records were unavailable and how that limits the conclusion.
Conclusion
Use the review to confirm that every manual change has a number, a reason, support, a second approver, and a post-run match. Repair the register design or the system roles before the next run. Track repeat causes and rerun the sample after any provider or portal change. The control is the reconciliation, not the form. Assign the next review date and the person who will own the register change.
Sources
Prepare a manual adjustment control pack
A Philippines-based payroll support specialist can issue register numbers, collect support, prepare the post-run match, and report open items for owner review. The specialist can also maintain the reason-code list and the monthly pattern summary. Your payroll owner still sets approval thresholds, approves high-value adjustments, and signs the reconciliation.
Review payroll operations supportFAQs
Is every manual adjustment a control failure?
No. A supported correction can be a legitimate part of payroll. The failure is an adjustment without authorization, support, or review. The review tests the control around the change, not the existence of manual work.
Can a small company use one person for preparation and approval?
It can document a compensating review by the owner or a second manager when separation is impractical. The exception and the check must be written down. A shared login or a second signature from the same person does not create independence.
What should be reconciled after the run?
Match every register entry to the posted payroll line by number, employee, period, field, and amount. Confirm the ledger coding and the funding. Investigate unregistered posted changes and registered entries that did not post.
For adjacent operating context, see Payroll Preparation and the payroll operations guide library.