Redact
How to Remove ID Numbers, Phone Numbers and Other Sensitive Data
The hard part of sending a document containing an ID number or phone number is not covering it — it is making sure it is genuinely gone. Drawing an opaque rectangle over text hides it visually while leaving it fully intact underneath.
Covering is not deleting
Every PDF reader ships an annotation or drawing tool that can place an opaque black rectangle over text. It looks fine, but the text layer is never modified — it is merely hidden.
Select that text with the mouse and paste it into a text box, and the ID number comes out in full. Search finds it. Upload the file to a conversion service and their server reads it too.
Real redaction has to regenerate the page: render it to an image, paint the sensitive region out, and write the new image into a fresh PDF. The original characters then do not exist in the output at all.
What UniPagePDF's redaction does
Drag over the area you want removed, pick a colour (black, red, or blue), optionally add a reason label, then export. Each page goes through a full round trip:
- the page is rendered to a high-resolution bitmap
- the marked regions are filled with solid colour on that bitmap
- a new image is produced and written into a new PDF
Because the text never reaches the new file, it is not searchable, not copyable, and not recoverable. The trade-off: pages are re-encoded, the text layer is lost (which is inherent to redaction), and the file size changes.
Batch find: no manual marking page by page
Redacting a hundred pages by hand is impractical. The tool includes batch find: enter a string or a regular expression and it scans the whole document and marks every match. Four presets are built in:
| Preset | Matches |
|---|---|
| SSN | xxx-xx-xxxx social security numbers |
| Phone | 11-digit mobile numbers and xxx-xxx-xxxx |
| Standard email addresses | |
| ID card | 18-character national ID numbers, including a trailing X |
Results report a count and the page numbers involved, and jump to the first match for review. Still inspect each mark afterwards — automatic matching can miss text that was split across fragments or uses unusual font encoding, and can flag innocent duplicate numbers.
Scanned documents cannot use batch find. A scan has no text layer at all; you must run OCR first to create one. That order cannot be reversed.
Watermark vs redaction
| Watermark | Redaction | |
|---|---|---|
| Content deleted? | No — overlay only | Yes — pages regenerated |
| Text layer | Preserved | Lost (whole file) |
| Use for | Marking CONFIDENTIAL / DRAFT / version | Removing real data before sharing |
Choosing a mark colour
- Black (default) — total concealment, the common case
- Red — marks the section as void or requiring attention
- Blue — marks it redacted while the underlying content is still open to discussion
You can attach a reason (for example "privacy") and choose whether to display it inside the mark. Useful for internal review, where the recipient can see why something is covered. Never put sensitive content in the reason text itself — it stays visible.
The whole file loses its text layer
Because every page is re-encoded as an image, text becomes unsearchable across the entire document — including pages you never marked. This is an unavoidable consequence of redaction.
If you need both an external redacted copy and an internal complete one, keep the original and generate a separate redacted version rather than working on the original in place.