List the personal data the document can contain
Start with the data the document is likely to carry: names, addresses, phone numbers, account references, and identity numbers. Add free-text fields where a customer may have written something sensitive. The list should reflect the business, not a generic form. A support log and a payroll file do not carry the same risk.
Separate data that must be removed from data that may remain. An account number may be needed for the recipient to act, while a home address may not. Write the purpose beside each field. Redaction without a purpose removes too much or too little.
Check the metadata as well as the page. A file can show a clean image while the document properties still hold an author name or a file path. Screenshots can capture a notification with a customer name. The hidden layer is part of the document.
Name who decides when a field is unclear. A privacy lead or the client owner should resolve the edge case, not the specialist alone. Record the decision with the file. The next person should not have to guess why a field was kept.
Check the visible layer and the hidden layer
Read the document as the recipient will see it. Look for names in headers, footers, comments, and tracked changes. Check images and charts for labels that repeat personal data. A clean body can still carry a name in a caption.
Inspect the file properties, the revision history, and any embedded objects. A spreadsheet may keep a hidden sheet with the original data. A PDF may retain the text under a black box. Removal of the visible mark is not removal of the data.
Use the tool's true redaction function when one exists. Drawing a black rectangle over text leaves the text selectable in many formats. Flatten or rasterize only when the business accepts the loss of searchable text. Record which method was used.
Reopen the exported file on a second device before sending it. A copy that looks clean in one viewer may expose layers in another. Check the file size and the page count as well. The final check should test the file that will actually leave.
Use a second reader before release
A second reader catches what the preparer has stopped seeing. Choose someone who did not create the document and give them the field list. Ask them to mark anything that looks like personal data. Their job is to question, not to approve silently.
Keep the second read short and specific. The reader should confirm the field list, the hidden layer, and the final export. Record the name and the time. A second read that takes seconds is unlikely to find a missed name.
Separate the preparer from the sender when the document is sensitive. The person who redacts should not also release the file without a check. This is a control, not a comment on trust. The separation makes the decision visible.
Give the reader authority to stop the release. If they find a problem, the file returns to the preparer with a note. Do not let a deadline override the check. A late document is easier to explain than a disclosed record.
Record the decision with the file
Record what was removed, what was kept, and why. A short note beside the file is enough when it names the fields and the rule. The record should travel with the document, not sit in a separate email. A reviewer should be able to reconstruct the choice.
Note the tool and the version used for redaction. A method that worked in one release may fail after an update. Keep the exported file and the note together. The pair is the evidence that the check happened.
Record any exception that a manager approved. Include the field, the reason, and the recipient group. An exception that is not written down becomes a precedent without a decision. The next reviewer can then judge whether it still applies.
Set a retention period for the redacted file and the original. The original may need to stay for legal or operational reasons, but it should not travel with the shared copy. Store it in the approved location. The note should say where the original lives.
Refresh the check when the template changes
A new template can add a field that the old check never looked at. Review the list when the form, the report, or the export changes. Add the new field and test it with a sample. A check that assumes the old layout will miss the new risk.
Retrain the team when the rule changes. A short example of the new field is more useful than a policy paragraph. Ask each preparer to run one sample through the updated check. Record who completed the training.
Review failed checks monthly for repeat causes. Two misses in the same header field may point to a template problem. Fix the template before asking people to be more careful. The control should remove the chance, not just the error.
Keep the check proportionate to the document. A public brochure needs less scrutiny than a customer record. Write the levels so the team does not over-redact routine material. The goal is a decision that fits the risk.
For the next step, compare this workflow with Set a document retention rule for outsourced support records and Set rules for direct customer contact by an assistant. For access and oversight, use the authoritative planning reference alongside your own policy.
