How to Edit a Scanned Document in Word Without Retyping
By Schema LawPublished Updated
To edit a scanned document in Word, open the scanned PDF from inside Word with File, Open and accept the message that Word will convert it. You then edit the converted copy as you would any Word document. The scan itself never changes, and the copy needs layout repair and proofreading before you rely on it.
Most of the work comes after the conversion. This post covers what Word changes on the page, how to repair the usual damage with Word's own tools, how to keep a record of your edits, and when to give up on this route.
Can you edit a scanned document in Word at all?
Only after the scan has become text. A scanned page is a picture, and Word cannot put a cursor inside a picture. Microsoft's method is to open the scanned PDF in Word, which converts it into an editable document. From then on you are editing Word's reading of the page, not the page.
Microsoft sets this out on its page about how to scan and edit a document: after you scan a document, you can open it in Word to edit it. The same page says the conversion works best with documents that are mostly text.
Keep three objects apart in your head, because the answer to most questions depends on which one you mean.
| Object | What it is | Can you edit it in Word? |
|---|---|---|
| The paper | The original | No |
| The scanned PDF | Pictures of the pages | No. Microsoft says the original PDF won't be changed at all |
| The converted Word file | Word's attempt to rebuild the pages as text | Yes |
If you have not scanned the paper yet, or you are starting from a phone, the post on how to scan a document into Word covers that half of the job and explains how Word's conversion works.
Opening the scanned PDF in Word so the text becomes editable
Go to File, Open in Word, find the scanned PDF and open it. Word tells you it is going to make a copy of the PDF and convert its contents into a format Word can display. Select OK, and the editable copy opens.
Those are the steps on Microsoft's Edit a PDF page, which lists Word for Microsoft 365, Word 2024, 2021, 2019 and 2016. Microsoft adds that you might have to select Browse to find the PDF in a folder.
Set yourself up before you type a word
The first ten minutes decide how painful the editing will be. Do these in order.
- Save the converted copy twice. Save once as
Agreement-converted-baseline.docxand close it. Save again asAgreement-working.docx. You will edit only the working file. The baseline is what you compare against later. - Put the scan on screen next to Word. Open the PDF in your PDF reader on one half of the monitor and the working file on the other.
- Turn on formatting marks. Microsoft's page on showing or hiding formatting marks says the Show/Hide button turns hidden characters such as spaces, paragraph markers and tab marks on and off. With the marks visible you can see how Word built each line.
- Page through the whole file once without fixing anything. Note which pages look right, which have broken tables, and whether any page is a single picture.
Saving when you are done
Microsoft's page says that after your edits you save the result as a Word document or a PDF, choosing the type from the Save as dropdown. For a PDF it gives Export, Create PDF/XPS Document, Create PDF/XPS, then Publish. Give that new PDF a name that cannot be confused with the scan. A file called Agreement.pdf sitting next to Agreement-scan.pdf is how the wrong one ends up attached to an email.
What does Word's PDF conversion change on the page?
Word changes the layout. Microsoft says the converted document might not have a perfect page-to-page correspondence with the original, and that lines and pages may break at different locations. It also names specific elements that convert badly, footnotes and some tables among them.
The fullest account is on Microsoft's page about opening PDFs in Word. It says that when you open a PDF in Word, it might not look exactly the way it looked as a PDF, and that Word sometimes fails to detect an element, so the Word version doesn't match the original.
| What Microsoft says may change | Where you will meet it in legal papers |
|---|---|
| Lines and pages break at different locations | Page and line citations no longer match the scan; a 12-page lease becomes 13 pages |
| A footnote Word doesn't recognize is treated as regular text and might not sit at the bottom of the page | Briefs, opinions, expert reports |
| Footnotes that span more than one page, and endnotes, don't convert well | Long briefs and law review material |
| Tables with cell spacing don't convert well | Fee schedules, rent rolls, payment tables |
| Page colors, page borders and frames don't convert well | Certificates, forms, bordered notices |
| Font effects are represented by graphics | Letterhead and logos |
| A page that is mostly charts or graphics might show up as one image whose text can't be edited | Plats, diagrams, photo exhibits |
Two consequences deserve a sentence each.
Never cite from the Word file. Page 7 of the converted document may hold what is on pages 6 and 7 of the scan. Any reference to a page, line or paragraph position has to come from the scanned PDF.
The Word file is not the document. It is a new document that resembles the scan. If the question is ever "what did the signed agreement say", the answer is in the scan or the paper. The post on whether OCR keeps formatting explains which kinds of structure tend to survive any conversion and which do not.
On the working copy, look up every party, the damages figure, and a section number in the contract or the pleading. When the scan is an affidavit, include the date by the signature. Read every hit on the untouched PDF. A character the conversion guessed wrong will sit in the text as if you had typed it.
Repairing text boxes, line breaks and numbering Word creates from a scan
Fix structure before wording. Use Show/Hide to see what Word built, move text out of any boxes into the body, remove paragraph marks that break lines in the wrong places, and rebuild numbering with Word's list commands. Then proofread against the scan.
Microsoft does not publish a list of what the conversion produces from a scanned page, so treat what follows as things to look for. Each repair uses a documented Word command.
Text sitting in boxes
Click in a paragraph and try to select from it into the next one. If the selection stops at an invisible wall, the text may be inside a text box. Microsoft's page on adding, copying or removing a text box says that to delete a box you select its border and press Delete, and warns that if the pointer is inside the box, Delete removes the text and leaves the box.
To get the words out:
- Click inside the box, select its text and copy it.
- Click in the body of the document where the text belongs and paste.
- Choose Keep Text Only from the paste options. Microsoft's page on paste formatting says that option discards formatting and nontext elements and gives the text the style of the paragraph where it lands.
- Select the border of the now-unneeded box and press Delete.
A paragraph mark at the end of every line
With formatting marks on, look at a long paragraph. If a ¶ sits at the end of each line, and not only at the end of the paragraph, the text will not rewrap when you edit it. Add three words and the line runs short while the next one stays put.
For one paragraph, delete the stray marks by hand. For a whole document, use Find and Replace. Microsoft's page on finding and replacing text in Word gives Home, Editing, Replace for Windows, with a More button for further search options. Its macOS instructions describe opening Advanced Find & Replace, selecting Replace, and using the Special menu to put a paragraph mark in the Find what box. A safe sequence:
- Replace every pair of paragraph marks (a real paragraph break) with a marker you will not find in the text, such as
@@@. - Replace every remaining single paragraph mark with a space.
- Replace
@@@with a single paragraph mark.
Run this on the working file only, and only on body text. Select the pages you mean to change first. Signature blocks, addresses, captions and tables depend on their line breaks and will collapse into one line if you include them.
Numbers that are only typed characters
Click on a paragraph number such as "3." or "(b)". If your cursor lands beside it and you can delete it one character at a time, it is plain text and will not renumber when you add or remove a clause. If it is a real list, Microsoft's page on changing the numbering in a numbered list gives the commands: double-click the numbers, right-click the one to change, and choose Set Numbering Value or Restart at 1. The same page warns against changing list numbers by hand, because Word can lose track of the list.
For a contract you plan to revise heavily, decide early. Either leave typed numbers alone and renumber manually at the end, or select the clauses, apply Word's numbering, and delete the typed numbers. Mixing the two produces "3. 3." and cross-references that point nowhere.
The proofreading pass
Layout repair does not catch wrong characters. Microsoft tells you to always proofread the text after opening a scan in Word. Go through the working file against the scan and check, at a minimum:
- Party names and addresses
- Every dollar amount, percentage and date
- Section numbers and cross-references
- "not", "no" and "nor", wherever a missing word reverses the meaning
- Defined terms that differ by one letter, such as "Lessor" and "Lessee"
Tracking your Word edits against the scanned original
Use two Word tools. Turn on Track Changes before you make substantive edits, so each one is marked. When you finish, run Compare with the legal blackline option against the untouched converted copy to produce a separate document showing everything that changed. Neither tool compares your text with the scan image.
Track Changes while you edit
Microsoft's page on tracking changes in Word gives the path: on the Review tab, select Tracking, then Track Changes. With tracking on, deletions are marked with a strikethrough and additions with an underline.
When to switch it on is a judgment call. A reasonable order is:
- Do the structural repairs with tracking off. Thousands of tracked paragraph-mark deletions help nobody.
- Do the proofreading corrections with tracking off as well, and save. That file is your corrected transcription of the scan.
- Turn tracking on. Now make your own changes to the wording.
Anyone who opens the file later can then tell your drafting from the original text.
Legal blackline at the end
Microsoft's page on comparing documents with the legal blackline option says the option compares two documents and displays what changed between them in a third, new document, and that the documents being compared are not changed. The steps:
- On the Review tab, in the Compare group, select Compare.
- Click Compare two versions of a document (legal blackline).
- Under Original document, choose the baseline. Under Revised document, choose your working file.
- Choose what to compare, and whether to show changes at the character level or the word level.
- Select OK.
By default the result opens as a new document. Microsoft cautions that choosing to show changes in the original or revised document instead will modify that document, so leave the default.
What the blackline does not prove
The comparison is between two Word files. If Word misread "$15,000" as "$16,000" during conversion and you did not catch it, the error is in both files and the blackline shows nothing. A clean comparison means "these are my edits to the converted text". It does not mean "this matches the paper". Only a person reading the working file against the scan can establish that, which is why the proofreading pass comes first.
When should you stop editing in Word and convert the scan another way?
Stop when repair is costing more than the conversion saved. If pages arrive as pictures, tables come through as loose lines, or you are correcting several words in every paragraph, the Word route has reached its limit for that document, and a different conversion or a clean retype is the better use of the hour.
Signs that you have reached that point:
- A page is one image. Microsoft says the text on such a page can't be edited. Nothing you do in Word will change that.
- The document is mostly tables. Rebuilding a twelve-column schedule cell by cell is slower and riskier than converting it with a tool that outputs tables.
- The scan is a poor copy. Microsoft says pages that look like a manuscript from a copier won't be formatted well, and that recognition depends on scan quality. Rescan from the paper if you can before trying anything else.
- There is handwriting that matters. Microsoft says handwritten text is seldom recognized.
- You have dozens of files. Opening, converting and repairing one at a time does not scale to a production.
- You need the page to stay the page. If what you want is the original image with searchable text behind it, that is a searchable PDF. This route gives you a rebuilt Word document instead.
A quick rule
Time yourself on the first two pages. If cleanup takes more than about five minutes a page, multiply by the page count and ask whether you would accept that number as a quote from a typist. If not, change the method.
Other routes
The guide to converting a scanned PDF to Word walks through conversion outside Word and how to judge the result. Schema Law is one such tool: you upload the PDF, PNG or JPEG in a browser, it reads the pages into text and tables, and you export Word or another format. The pricing page lists page limits for each plan, and the free plan handles one document a day of up to three pages, which is enough to run the same two-page timing test. Whichever route you pick, the scan remains the record and the proofreading still has to be done.