You have to share a document with something in it removed.
Drawing a black box over it does not remove it
The most common redaction mistake leaves the original text sitting in the file underneath the mark, fully selectable. This is how documents leak.
Why the black box fails
A PDF stores text and graphics as separate objects. A filled rectangle placed over a paragraph is a new object drawn on top; the paragraph underneath is untouched. Anyone can select the text through the box, copy it, or extract it with a script.
This has produced real disclosures in court filings and published reports. It is not an exotic attack — it is select-all and paste.
What real redaction does
Proper redaction removes the underlying content, then covers the area. The text objects are deleted from the file, not hidden, so there is nothing left to select or extract.
After redacting, verify it yourself: search the document for the word you removed, and try to select the covered area. If either finds anything, the content is still there.
The parts people forget
Metadata. Author, title, and producer fields travel with the file and often carry a real name or an internal path, even when every page is clean.
Attachments and layers. A PDF can carry embedded files and hidden layers that no page displays.
Filenames. "Smith-termination-draft-v3.pdf" discloses plenty before anyone opens it.
If the document is a scan
On an image-only PDF there is no text object to remove, so covering the area and then flattening the page merges the mark permanently into the image.
Be careful in the other direction too: if the scan has had OCR run on it, there is now a text layer, and that layer needs redacting like any other text. A page can look like a photograph and still be fully searchable underneath.
Redact the pattern, not just the instance
A name usually appears more than once, and rarely in the same form. Search for the surname alone, the initials, the email local-part, and the account or case number before deciding a document is clean.
Watch for the places text hides in plain sight: a header or footer repeated on every page, a table of contents that still lists a removed section, a chart label, and the document outline or bookmarks panel.
The same applies to indirect identifiers. Removing a name while leaving a job title, an office location, and a date of joining can identify one person exactly as well as the name did.
Verify on the file you are actually sending
Do the checks on the exported copy, not on the editing session — the editor may show you a view that the saved file does not match.
Open the export fresh, search for each removed term, select the redacted regions, and inspect the document properties. If you can copy anything back out, the redaction did not happen.