Research note
What share of employee data changes lack a second review?
A measurement design for counting employee data changes that reach the system without an independent review and separating a missing check from a missing approval.
Published

Table of contents
Question and scope
This study asks what share of employee data changes reach the system without an independent second review. It treats the change entry as the unit and the review record as the evidence. The scope is one employment operations team and a fixed review window. The result describes that team and should not be read as an industry benchmark.
A change is eligible when it alters a field that affects pay, tax, benefits, or the official employee record. A name correction or a phone update may be excluded if the policy treats it as low risk. The inclusion rule should be written before the sample is drawn. A broad definition can make the rate look worse without improving the control.
The study also records whether the change required an approval as well as a review. A second review checks the entry, while an approval authorizes the change. A change can have one and lack the other. Report the two conditions separately so the repair targets the right gap.
Definitions and evidence standard
Define an independent review as a check by someone other than the person who entered the change. A self-check before saving is useful but is not independent. The reviewer should be able to see the source, the old value, and the new value. A review that only confirms the change exists is not a substantive check.
Use the change log, the system audit trail, and the approval record as the evidence. If the review is recorded only in a chat message, the entry is treated as unreviewed unless the message is linked to the change. A later recollection is not evidence. The record should show who reviewed and when.
Separate a review before the update from a review after the update. A pre-update review can prevent an error, while a post-update review can only catch it. The study should report the timing because the control value is different. A post-update check may still be useful when the system cannot hold a change.
Sampling and measurement
Collect changes from completed pay periods and freeze the list before review. Include changes from more than one requester and more than one field type. If the period contains a system migration, report it separately. Record the number of changes and the number of employees affected.
Measure the number of eligible changes, the number with an independent review before the update, the number with a review after the update, and the number with no review. Add the number that also lacked an approval. Report the counts by field type and by requester. A rate without these counts hides where the gap sits.
If a change was reviewed twice, count the change once and record the extra review. A second review may be appropriate for a high-risk field. Report the repeat count beside the main rate. The extra review is evidence of a stronger control, not a separate change.
Findings and interpretation
Present the distribution before suggesting a cause. A cluster of unreviewed changes in one field type may mean the field is not on the review checklist. A cluster under one requester may mean that person has a habit of bypassing the log. Read the actual changes behind the largest cluster before naming a reason.
Compare the review rate with the change volume in the same period. A high volume may reduce the time available for review, while a low volume may make each review more thorough. Report the volume with the result. The comparison keeps a busy month from being read as a control failure.
Check whether the unreviewed changes were later corrected. A change without review may still be correct, while a reviewed change may still be wrong. Report the correction count separately. The combination shows whether the missing review is a risk or only a documentation gap.
Limitations
The study measures the recorded review, not whether the reviewer actually looked at the change. A rubber-stamp review can be recorded and still provide no control. The result also cannot show whether an unreviewed change caused a payment error. It describes the record, not the full effect.
The review may happen in a system that does not connect to the change log. A missing reference may look like a missing review even when a check occurred. The sample may also depend on the reviewer's note-taking habit. These limits should be stated with the counts.
Coding a review as independent requires knowing who performed it. A shared account can hide the reviewer's identity. Record the account and the person when known. An ambiguous entry should be treated as a limitation rather than a finding.
Practical next step
Add a reviewer field to the change log and make it separate from the requester and the approver. The field should accept a name rather than a shared team label. A visible field is easier to enforce than a policy sentence. The reviewer should be able to see the source from the log.
Choose the highest-risk field type and require a pre-update review for that type first. A bank detail or a tax change is a better starting point than a phone number. Test the requirement for one month and compare the review rate. A narrow change is easier to adopt than a blanket rule.
Review the exceptions with the team and ask why each unreviewed change occurred. The answer may point to a missing backup, a system limitation, or a habit. Fix one cause and keep the rest visible. The goal is a review that happens before the update, not a longer log.
Comparing field types
Compare pay-critical fields with contact fields. A bank detail or a tax election can change the amount or destination of pay, while a phone number usually cannot. The review requirement should reflect that difference. Reporting all fields together can make the control look weaker than it is.
Look at changes made by the employee through a self-service portal separately from changes made by the payroll team. A portal may have its own validation and approval. A team entry may rely on a manual review. Report the channel with the result so the controls are not confused.
Compare changes that required an approval with those that did not. A change with an approval but no review may still have an independent check by the approver. A change with a review but no approval may lack authority. The combination shows which control is missing.
What the evidence does not show
The study does not show that every change needs a second review. A low-risk field may be adequately controlled by a validation rule. It also does not show that a reviewed change is correct. The review can be shallow or based on the wrong source.
It does not prove that the person who entered the change was careless. The log may not prompt for a reviewer, or the system may not support a hold. The gap is a process signal, not a personal judgment. The repair should start with the form and the system.
Another employer would bring a different field mix and a different review habit, so the rate should not travel. The method can travel if the definitions are kept intact. What the local result offers is a sharper question about which fields need a second pair of eyes. That question is more portable than the percentage.
Using the result in a payroll conversation
Show one reviewed change and one unreviewed change side by side. Ask the payroll owner which one they would rather defend in an audit. The comparison makes the control gap visible without blaming an individual. The owner can then choose the first field to require.
Ask the reviewer which check is hardest to perform and why. The answer may point to a missing source or a system that does not show the old value. Use that answer to improve the review screen or the log. A concrete obstacle is easier to fix than a general instruction.
Close with one field, one review rule, and one date to test it. Record the baseline so the next review can compare the same field. If the change does not improve the rate, investigate before adding another control. The goal is a real check, not a longer audit trail.
Sources
References supplied with this research note.
| Source | Publisher | Context |
|---|---|---|
| U.S. Department of Labor, Fair Labor Standards Act | dol.gov | Primary reference supplied with this article. |
| U.S. Department of Labor, Recordkeeping | dol.gov | Primary reference supplied with this article. |
| Internal Revenue Service, Publication 15 | irs.gov | Primary reference supplied with this article. |
| National Institute of Standards and Technology, Cybersecurity Framework 2.0 | nist.gov | Primary reference supplied with this article. |