Editing Scanned PDF Text: What You Can and Cannot Change

Editing scanned PDF text is never a matter of clicking into a sentence, because a scanned page is a photograph of paper. You have two routes. Convert the scan to Word and edit the words there, or cover the old words on the picture and place new ones on top. Either way, the result is a new document.

Which route fits depends on how much you are changing and on what the page is. A blank intake form you scanned last week is yours to rework as you like. A countersigned agreement is a record of what somebody signed, and a changed copy of it has to be handled with more care than the editing itself.

Can you edit the text of a scanned PDF directly?

No. A scanned PDF holds one picture per page and no characters, so there is no text to put a cursor in. What PDF programs call editing a scan is one of two things: recognizing the letters first and rebuilding the page from them, or painting over a region of the picture and typing on top.

It helps to see what the file contains. When a copier scans a page, it records a grid of coloured dots. The word "Tenant" on that page is a few hundred dark dots in a pattern you recognize. The file does not know they spell anything. Deleting the "t" at the end is as meaningful to the software as deleting the corner of a coffee stain.

So every tool that offers to change text on a scanned document is doing some preparatory work for you:

• Recognition and rebuild. The program runs optical character recognition over the page, guesses a font, and replaces the picture of each line with real, typed characters in the same position. People call the result an editable PDF. The page now looks slightly different everywhere, even where you changed nothing, because every line has been retyped by software.
• Cover and overtype. The program leaves the picture alone, lets you draw a filled box over a few words, and places a new text box on top. Only the patch changes. The rest of the page is still the scan.
• Convert out. A separate tool reads the page and gives you a Word file. You edit there and, if you need a PDF again, you save one from Word. That PDF is no longer a scan at all.

A searchable PDF does not change this. Its hidden text layer lets you find and copy words, but the visible page is still the image. Editing the hidden layer would change what search finds without changing what a reader sees, which is the opposite of what you want.

One more thing can stand in the way before any of this begins. If the file's author restricted changes, your PDF program will refuse to add even a text box. The post on protected files describes the kinds of protection a PDF can carry and how to tell which one you have met.

Editing in the PDF versus editing a Word conversion of the scan

Edit in the PDF when the change is a few words and the page must keep looking like the scan. Edit a Word conversion when you are rewriting sentences, moving clauses or reusing the language in a new document. The first keeps the appearance and limits what you can do. The second gives you full control and discards the appearance.

Three situations come up often enough to name.

A form you want to reuse. The firm has a paper engagement checklist that exists only as a scan. You want to change the fee paragraph and reissue it. Convert to Word, fix the recognition errors once, and keep the Word file as the master from now on. Patching the scan every year is the slower path.

A typo in a document you are about to send. Your own cover letter was printed, signed and scanned, and the matter number in the heading is wrong. If you have the original word-processing file, correct it there and sign again. That is cleaner than any edit to the scan.

Language from someone else's document. You need the indemnity clause from a scanned supply agreement as the starting point for a new draft. Nothing about the scan should change. Convert it, check the clause, and copy the text into your draft. The guide to getting scan text into Word from the paper page covers the conversion route from the beginning.

Schema Law belongs to the second column only. It reads a PDF, PNG or JPEG into text and tables and exports them in the formats on the export list, Word among them. It does not edit the PDF you uploaded. If you take the conversion route, read the list of what to check in the Word file before editing before you change anything, so that software misreadings and your own edits never get mixed together.

If you want to edit a scanned document online for free, check the page cap of any free converter before you rely on it. Schema Law's free plan allows one document a day, three pages at most, with Word among its exports. The pricing page has the limits for longer files. For a client's document, where the upload goes matters more than what it costs.

Changing a word, a date or a name on a scanned page

To change a word, a date or a name on a scanned page, work on a copy, cover the old text with a box the colour of the paper, type the replacement in a matching font, flatten the page so the patch cannot slide, and compare the result with the untouched original at high zoom.

Take a fictional example. Calloway Marine Supply sends your office a scanned, unsigned purchase order template. The delivery date in the second paragraph reads "14 June 2026" and should read "24 June 2026". Nobody has signed it, the client has asked you to fix it, and they want it back as a PDF that looks the same.

1. Duplicate the file first. Save a copy under a new name before you open any editing tool. The scan you received stays exactly as it arrived.
2. Zoom to 300% or more on the text you are changing. At normal size you cannot judge the paper colour or the edges of the letters.
3. Sample the background. Scanned paper is rarely pure white. Pick the fill colour from the page beside the words, or a white patch will show as a bright rectangle on a cream page.
4. Cover only what changes. Draw the box over "14" and nothing else. The smaller the patch, the less there is to match. Keep clear of the line above and the line below.
5. Add a text box and type the replacement. Choose a typeface of the same family as the original, serif for serif. Set the colour to a dark grey sampled from the nearby letters, since scanned ink is seldom true black.
6. Nudge until the baseline agrees. Line the bottom of your digits up with the bottom of the neighbouring word, then match the height of a capital or a digit beside it.
7. Flatten. A box and a text annotation are separate objects that a reader can move or delete, exposing the "14" underneath. Flattening, or printing to a new PDF, merges them into the page.
8. Check both ways. Look at the page at 100% for anything that catches the eye, then at 400% for the old text peeking out at the patch edge. Search the file for the old value in case a text layer still holds it.

The last check matters more than it seems. If the scan had a hidden text layer, the layer still says "14 June 2026" after you patch the picture. Anyone who searches or copies from the file gets the old date. Either run recognition again on the edited page so the layer is rebuilt, or remove the layer.

When a patch is the wrong tool

A patch only works when the replacement is about as wide as what it covers. Swapping "14" for "24" is easy. Replacing "Calloway Marine Supply" with "Calloway Marine Supply and Chandlery Limited" is not, because the rest of the line cannot shift right to make room. If the new text is longer, convert to Word and rebuild the document.

A patch is also the wrong tool for removing confidential information. A box drawn over a name hides it from the eye and may leave it in the file. That job needs a proper redaction function, which removes the content beneath. The post on hidden text explains why a black box drawn over words leaves the words in the file.

Why does edited text on a scanned page look pasted in?

Edited text on a scanned page looks pasted in because the new characters are perfect and the old ones are not. Your replacement is drawn by a font, sharp at every size. The surrounding words are a photograph of ink, softened by printing, copying and scanning. The eye picks out the difference at once.

Several separate mismatches add up to that impression. Knowing them lets you fix the ones that can be fixed and judge when to stop trying.

Programs that recognize the page and rebuild it meet the same problem at a larger scale. The software has to guess a font for every line from the shapes in the picture. Where it guesses well, the page looks a little too clean. Where it guesses badly, lines grow or shrink, wrap at different points and push into each other. That is why a scanned PDF converted to an editable PDF so often looks subtly wrong even before anyone has changed a word.

Ask whether the match needs to be invisible at all. For a working copy or an internal form, a correction that is plainly a correction does no harm, and it may be better that it shows. Where an edit must be undetectable to be acceptable, stop and ask why. If the honest answer is that the reader should believe the page was always this way, the question is no longer about fonts. It is whether the scan should be altered at all.

Should you ever alter the scan of a signed or filed document?

As a matter of record keeping, treat the scan of a signed or filed document as fixed. If its content needs to change, the change belongs in a new instrument: an amendment, a corrected filing, a substitute page that the signer initials.

Whether a particular alteration is permitted is a legal question for the lawyer on the matter, and the rules differ between courts and jurisdictions. What follows is about how files are handled, and it is not legal advice. The handling follows from what a scan is for. The scan of an executed agreement stands in for the paper. Its value is that it reproduces what was signed. The Federal Rules of Evidence, for one, define a duplicate as a counterpart that "accurately reproduces the original", and the post on scanned contracts explains when a scanned copy of a signed contract is as good as the original. Those are federal rules, and state courts have their own. The practical point holds anywhere: a scan that has been patched no longer does the one job a scan has.

Some working distinctions help when a request lands on your desk.

• Adding on top is different from changing underneath. An exhibit sticker, a Bates number, a "COPY" watermark or a highlight is added matter that everyone understands as added. Replacing a figure in the body text is a change to what the document says.
• Blank and unsigned documents are different from executed ones. A scanned form with nothing filled in is a template. Edit it freely, and say in the file name that you did.
• Your own error is different from the other side's. If your signed letter went out with a wrong date, send a corrected letter. Do not quietly replace the scan in the file with a patched one, because the recipient holds the version you actually sent.
• A filed document is fixed by the filing. The court's copy is what was filed. A mistake in it is corrected through whatever procedure that court provides, so check its rules or ask the clerk.

When a supervising lawyer does decide that a scan should be edited, for instance to produce an illustrative version for a meeting, the instruction should be written down and the edited copy labelled so it can never be mistaken for the record.

If the scan is a filed complaint, leave the caption, the allegations and the prayer untouched in the record copy. An illustrative copy for a meeting can highlight a cause of action, and the file name has to say it is illustrative. The plaintiff and defendant names on the court copy stay as filed.

Marking which copy of a scanned PDF was edited

Mark an edited scan in three places: the file name, the face of the page and a note kept with the matter. Each one survives a different accident. A name is lost when someone renames the file, a mark on the page is lost when someone crops it, and a note is lost when the file is emailed alone.

File names

Keep the original's name and add to it. Never save an edited copy over the file you received.

A Word conversion needs its own label too, and for a signed contract the post on converting a signed agreement covers labeling the Word file so nobody mistakes it for the executed agreement.

If your document management system keeps versions, do not rely on version numbers alone. "Version 3" tells nobody that version 1 is the record and version 3 is a working copy.

On the page

For anything that might be printed or forwarded, put the status where it cannot be separated from the content. A small footer line such as "Working copy. Edited from scan received 30 May 2026. Not the executed document." does the job. So does a light watermark. Flatten it along with the edit.

In the matter file

Keep a short log for any document where an edited copy exists. Four fields are enough:

• which file is the original, and where it came from,
• which file is the edited copy,
• what was changed, in plain words ("delivery date, paragraph 2, 14 June to 24 June"),
• who asked for the change and who made it.

Set the original to read-only if your system allows it. When you send the edited copy to anyone, say in the covering message that it is an edited copy and attach or reference the original. Months later, when the two files sit side by side in a folder and nobody remembers the request, the name, the footer and the log are what tell them apart.