# How to Change Scanned Document Into Word: Which Route

- Author: Schema Law (https://www.schemalaw.com/)
- Published: 2023-02-09T12:01:00Z
- Updated: 2026-10-11T19:22:20Z
- Category: Converting scans
- Canonical: https://www.schemalaw.com/blog/how-to-change-scanned-document-into-word

> How to change a scanned document into Word: five routes compared by the file you hold and the job the Word file must do, with the cleanup each one leaves.

To change a scanned document into Word, pick the route by two things: the kind of file you hold and what the Word file is for. A clean scanned PDF you only need to quote can go through Word itself. A long contract you will redraft deserves a dedicated OCR converter. A few lines are quicker to retype.

Every route ends in a DOCX, which is why they look interchangeable. They are not. Each accepts different files, hands back a different kind of Word document and leaves a different amount of repair. This post sets the five routes next to each other so you can choose once, and points to the post that walks through each procedure.

## Five routes from a scanned document to Word, compared side by side

There are five practical routes: open the scan in Word, use a free online converter, let a phone scanning app do the recognition, upload to a dedicated OCR converter, or retype. The first four all run text recognition on a picture of the page. They differ in what they accept and in how much they leave you to fix.

| Route | What it starts from | What you get | Main cost |
| --- | --- | --- | --- |
| Word's own PDF conversion | A scanned PDF | A Word copy that tries to look like the page | Layout repair on anything beyond plain paragraphs |
| Free online converter | A small PDF or image | Varies by service | Size and page caps, and a decision about where the file goes |
| Phone scanning app | The paper itself, or a photo | A PDF or image, sometimes with recognized text | Photo quality sets the ceiling; a second step to reach DOCX |
| Dedicated OCR converter | A PDF, PNG or JPEG | A DOCX built for editing | A page allowance or a subscription |
| Retyping | Anything you can read | Exactly what you type | Your time, and your own typing errors |

Two things in that table are worth a closer look.

First, the routes are not equally available to every file. Word converts a PDF. Hand it a JPG and, by Microsoft's own instructions for [scanned images in Word](https://support.microsoft.com/en-us/office/insert-scanned-text-or-images-into-word-b4ae150e-319f-4e18-b27b-418f1d690823), it goes in as a picture. A converter with a 2 MB ceiling will refuse most multi-page scans. So the file in your hand removes some routes before you weigh anything else.

Second, "a Word file" means different things. One route tries to reproduce the look of the scanned page. Another writes plain headings, paragraphs, lists and tables and lets the look go. For quoting, either will do. For a document you will take apart and rebuild, the difference decides how your afternoon goes.

## Which route to Word fits the file you are holding?

Start with the file type. A scanned PDF can take any of the routes. A JPEG or PNG cannot be converted by opening it in Word, so it needs a converter that accepts images. A TIFF fax or an iPhone HEIC photo usually has to change format first. Paper has to be captured before anything else happens.

| What you hold | Routes open to it | First thing to do |
| --- | --- | --- |
| Scanned PDF, clean print | All five | Check whether it already has text by trying to select a word |
| Scanned PDF, poor copy or fax | Dedicated converter, retyping | Find a better copy if one exists |
| JPEG or PNG of a page | Online or dedicated converter, retyping | Confirm the whole page is in frame and readable when zoomed in |
| TIFF from a fax server | Depends on the tool | Convert the format, or use a tool that takes TIFF |
| Phone photo (often HEIC) | Depends on the tool | Convert to JPEG, or retake it with a scanning app |
| Paper page | Any, after capture | Scan to PDF, or photograph it if you cannot take it away |

**Scanned PDFs.** Before choosing a route at all, try to select a sentence. If words highlight and paste correctly, the PDF already carries text, and you may not need recognition. The post on [why you can't copy text from a PDF](https://www.schemalaw.com/blog/why-cant-i-copy-text-from-a-pdf) shows how to tell in under a minute.

**Image files.** Microsoft's page on [inserting scanned text or images into Word](https://support.microsoft.com/en-us/office/insert-scanned-text-or-images-into-word-b4ae150e-319f-4e18-b27b-418f1d690823) treats an image as something you add through the Insert tab and the Pictures command, naming .JPG, .PNG and .GIF. That gives you a picture inside a document, with no text to edit. For words, an image needs a converter that reads pictures. Schema Law accepts PNG and JPEG as well as PDF, up to 20 MB a file.

**Faxes and phone photos.** These fail on format before they fail on quality. Fax servers often save TIFF, phone photos often arrive as HEIC, and many converters take neither. The post on [HEIC to JPG](https://www.schemalaw.com/blog/heic-to-jpg) deals with the phone case. The post on [image file types as OCR input](https://www.schemalaw.com/blog/scanned-image-to-text) covers what to do with each. A fax also arrives at low resolution, which narrows the routes further; see the guide on [how to OCR a faxed document](https://www.schemalaw.com/blog/how-to-ocr-a-faxed-document).

**Paper.** With the original on your desk you control the one thing that helps every route: the quality of the capture. A feeder or flatbed scan gives a converter more to work with than a photo taken at an angle under a ceiling light.

## Which route fits the job the Word file has to do?

Match the route to how much of the Word file must be right and how heavily you will rework it. Quoting one passage needs correct words and no layout. Light edits need text that rewraps. Redrafting needs real structure. A redline needs a verified text before any comparison means anything.

| The job | What matters | Route that usually fits |
| --- | --- | --- |
| Quote a paragraph in a brief or letter | The words of that paragraph, checked | Whichever is nearest; retype if it is a few lines |
| Light edits: a date, a name, a clause | Text that flows when changed | Word's conversion for clean prose; a dedicated converter if there are tables |
| Redraft from an old agreement | Headings, lists and tables as real Word objects | Dedicated OCR converter |
| Redline against another version | Text proved against the scan first | Dedicated converter, then verification, then compare |

### Quoting

For a quotation, the conversion is a convenience and nothing more. You will read the passage against the page image before it goes into a filing, so the route barely matters. What does matter is citing the scan and its page numbers, not the converted file. The post on [quoting text from a scanned PDF](https://www.schemalaw.com/blog/how-to-quote-text-from-a-scanned-pdf) covers that discipline.

### Light edits

Changing a party's address in a two-page letter is the job Word's own conversion suits best. Microsoft's page on [opening PDFs in Word](https://support.microsoft.com/en-us/office/opening-pdfs-in-word-1d1d2acc-afa0-46ef-891d-b76bcd83d9c8) says the conversion works best with files that are mostly text, and gives legal documents as one example. A letter or a short order is that kind of file.

### Redrafting

If the scan is the starting point for a new agreement, you will delete clauses, add others and reapply your firm's template. Layout copied from the scan gets in the way of all three. Take the route that gives you structure and no decoration.

### A redline

A comparison between a converted file and another draft reports every misread character as a change. Fix the conversion first, freeze it, and then compare. The post on [comparing two scanned contracts](https://www.schemalaw.com/blog/how-to-compare-two-scanned-contracts) explains the order of work.

## The cleanup bill each route to Word leaves behind

Cleanup is the real price of each route. Word's conversion leaves layout to repair. Free converters often drop lists and tables. A phone capture adds misreads from shadows and angle. A dedicated converter leaves page-end joins and styling. Retyping leaves only your own typos. All of them leave proofreading.

**Word's own conversion.** Microsoft is plain about the weak spots. Its page on opening PDFs in Word lists elements that do not convert well, among them tables with cell spacing, footnotes that span more than one page, endnotes, frames, and page colors and borders. It adds that when a PDF "contains mostly charts or other graphics, the whole page might show up as an image", and then the text cannot be edited. For a lawyer, that list reads as rent schedules, briefs with long footnotes and bordered forms.

**A free online converter.** The limits vary by service, so read the service's own help page before you count on it. Google's is a fair example because Google documents it. The Drive help page on [converting PDF and photo files to text](https://support.google.com/drive/answer/176692) says the file "should be 2 MB or smaller" and that lists, tables, columns, footnotes and endnotes "are not likely to be detected". What opens is a Google Doc. A numbered contract with a fee table loses exactly the parts that take longest to rebuild.

**A phone scanning app.** The app is a capture step, and its cleanup cost is mostly inherited. A page photographed flat and evenly lit converts much like a scan. A page with a curled corner, a thumb in frame or a shadow across the last paragraph produces errors no later step removes. Count on a second tool to reach DOCX, and count the retakes.

**A dedicated OCR converter.** Expect a document meant for editing, and expect it to look plain. Schema Law's Word export carries headings, paragraphs, bulleted and numbered lists, tables and bold type, and starts a new Word page for each page of the scan. It does not carry the page images, so signatures, stamps and letterhead are not in it. Your cleanup is joining paragraphs that crossed a page end, removing stray running headers and applying your template.

**Retyping.** There is no layout repair because you built the layout. The cost is time and a different kind of error: a skipped line, a transposed figure, a word your fingers supplied from habit.

One cost is common to all five. Nothing in a converted file marks which words were read wrongly, and nothing in a retyped file marks what you mistyped. Budget a read of names, dates, amounts and section numbers against the scan whichever route you take. The post on [why OCR output is full of errors](https://www.schemalaw.com/blog/why-is-ocr-output-full-of-errors) sets out how to proofread in proportion to the task.

## When does retyping a scanned document beat converting it?

Retyping wins when the passage is short, when the page is mostly handwriting, or when the copy is so degraded that correcting a conversion would take longer than typing fresh. As a rough test, if you could type it in less time than you would spend checking a conversion, type it.

Three cases come up often.

- **A few lines.** One clause for a letter, a caption, an address block. Uploading, waiting, exporting and checking takes longer than thirty seconds of typing.
- **Handwriting.** Microsoft's page on inserting scanned text says handwritten text is "seldom recognized", and that typed or printed text gives the best results. Other tools vary, and you should test your own pages before relying on any of them. The post on [OCR for handwritten notes](https://www.schemalaw.com/blog/ocr-handwritten-notes) describes what recognition can and cannot do with handwriting.
- **A copy of a copy of a fax.** When half the characters are broken, a conversion gives you a page where every word needs a second look. A typist reading for sense does better.

Retyping has a ceiling. Past a page or two, typing errors pile up and the time grows in a straight line, while a conversion's time barely changes with length. It is also a poor choice for tables of figures, where a human typist transposes digits as readily as software misreads them.

A middle path often works best. Convert the typed body of the document, and type in the handwritten interlineations and margin notes yourself, in square brackets so a reader knows they were transcribed by hand.

## Each route's procedure in two sentences, and the post that owns it

Each route is a short procedure, and each has a post on this blog that covers it in full. What follows is only enough to recognize the route and find the right guide.

- **Word's own PDF conversion.** Open the scanned PDF from inside Word through the File menu; Microsoft's page on opening PDFs in Word says Word makes a copy and converts it, so you always have the original PDF. The steps are in [how to scan a document into Word](https://www.schemalaw.com/blog/how-to-scan-a-document-into-word), and the repairs afterwards are in [how to edit a scanned document in Word](https://www.schemalaw.com/blog/how-to-edit-a-scanned-document-in-word).
- **A free online converter.** Upload the file in a browser, wait, and download the result, staying inside the service's size and page caps. The post on [what free scanned PDF to Word tiers limit](https://www.schemalaw.com/blog/free-scanned-pdf-to-word) compares those caps.
- **A phone scanning app.** Capture each page flat and square, save the set as one PDF, and send that PDF through one of the conversion routes. Technique is covered in [how to OCR a photo of a document](https://www.schemalaw.com/blog/how-to-ocr-a-photo-of-a-document).
- **A dedicated OCR converter.** Upload the PDF or image, look over the parsed pages, and export Word. The guide to [converting a scanned PDF to Word](https://www.schemalaw.com/blog/how-to-convert-a-scanned-pdf-to-word) owns the full procedure and the review checklist, and [OCR JPG to Word](https://www.schemalaw.com/blog/ocr-jpg-to-word) covers the same job for a picture file.
- **Starting from paper.** Prepare the original, scan it to a single PDF, and check the scan before the paper leaves your desk. The post on [scanning text to Word](https://www.schemalaw.com/blog/scan-text-to-word) covers devices and preparation.

Where the document belongs to a client, add one step before any route that uploads: find out what the service keeps and what it does with it. The post on [whether free online OCR is safe for legal documents](https://www.schemalaw.com/blog/free-online-ocr-safe-for-legal-documents) lists the questions.

## A three-page trial before you commit a long scan to one route

Before sending a long document down any route, run its three hardest pages through and time the repair. Pick the page with a table, the page with the worst print and the signature page. If those come back usable, the rest will, and if they do not, you have lost minutes and not a day.

Here is how that plays out. Tomas, a paralegal, holds a 28-page scanned supply agreement between two fictional companies, Calder Marine Supply and Whitlow Fabrication. The partner wants a restated agreement, so the job is redrafting.

1. **Rule out routes by file.** It is a scanned PDF of 9 MB. That is too large for a converter capped at 2 MB, and too long to retype.
2. **Rule out routes by job.** Redrafting needs numbered clauses and a pricing schedule as real Word content. Tomas shortlists Word's conversion and a dedicated converter.
3. **Extract three pages.** He saves pages 4 (dense definitions), 19 (the pricing schedule) and 28 (signatures) as a small PDF.
4. **Run both and time the repair.** He opens the extract in Word and notes what the schedule looks like. He uploads the same extract to Schema Law, where the Free plan allows one document a day of up to three pages with Word export included, and compares.
5. **Multiply.** Whatever the repair time per page was, he multiplies by 28 and puts that number in front of the partner with his recommendation.

The trial answers the question that no comparison table can: how your document, with its particular print and its particular tables, behaves on each route. Results on someone else's lease tell you little about your supply agreement.

If the dedicated route wins and the document is longer than a free allowance covers, the [pricing page](https://www.schemalaw.com/pricing) lists the page limits for each plan. Every plan exports Word along with the other seven formats, so the choice of plan comes down to page volume and the support that goes with it.

Keep the scan whatever you choose. The Word file is a working copy that says what the scan says, as far as someone has checked. The scan is still the document.

## Try it on a real document

Upload a scanned PDF and export it as Word, a searchable PDF or plain text on the free plan. Your documents are never used to train an AI model. [Start free](https://www.schemalaw.com/api/auth/google)