MAXScanner Guide
Scanned PDF to Word: How to Get Editable Text
A scanned PDF holds pictures of pages, so a converter without OCR puts a picture into Word. We ran one scan through three routes and counted the text that reached Word.
The short answer
A scanned PDF stores a photo of each page, not letters, so Word has nothing to edit until OCR (optical character recognition) has read the text. MAXScanner’s PDF to Word does this for you: it notices that the pages are pictures, runs OCR in your browser and writes the recognized text into the Word file. PDF to Text runs the same OCR and lets you correct the text before you download it as a Word file. On our 200 DPI test scan both routes returned all 655 characters without an error.
OCR gives you the words and the paragraphs, not the original layout. Fonts, column positions and table lines have to be redone in Word. If you only need to search or copy the text and the page must keep its look, make a searchable PDF instead.
How to tell whether your PDF is scanned
You can check in a few seconds, before you choose a converter.
- Try to select one word. If the whole page highlights as one block, or nothing highlights at all, the page is an image.
- Search for a word you can see (Ctrl+F, or Command+F on a Mac). No match means there is no text layer.
- Scanners and phone scanner apps usually save pages as images unless a text recognition (OCR) option was switched on when the PDF was saved.
- One PDF can mix both kinds: pages exported from Word have text, a signed page scanned back in does not. PDF to Text reads the text layer where there is one and runs OCR only on the other pages.
Three routes, one scan: what ended up in Word
We made an A4 notice with 11 pt text and a small timetable, scanned it at 200 DPI and saved it as an image-only PDF of 439 KB. Then we ran it through three routes on MAXScanner’s own pages and opened each Word file.
| Route in MAXScanner | Text in the Word file | Pictures in the Word file | Word file size |
|---|---|---|---|
| PDF to Word on the scan | all 655 characters, 0 errors | none | 3.4 KB |
| OCR PDF, then PDF to Word | all 655 characters, plus 1 stray quote mark that OCR PDF had read | none | 3.4 KB |
| PDF to Text, then Download Word (.docx) | all 655 characters, 0 errors | none | 3.4 KB |
Why converters put a picture into Word
A PDF-to-Word converter without OCR can only move what it finds. On a scan it finds one image per page, so it copies that image into Word. Our own PDF to Word did exactly this until 10 November 2026: it stored the page as a 3.1 MB PNG, seven times the size of the PDF, and Word showed a page you cannot type into. Many online converters still do this unless you switch on their OCR option, which is often a paid feature.
Running OCR PDF first did not help either, for a quieter reason. OCR PDF adds an invisible text layer, so the searchable PDF can be searched and copied, but the converter took over only visible text, and Word again received just the picture. We changed PDF to Word to read that invisible layer and to run OCR itself on pages that have no text, which is why all three routes in the table now end with editable text.
Step by step: scanned PDF to an editable Word file
The quickest route is PDF to Word: add the scan, check the OCR language it suggests and download the .docx. If you want to read and correct the text before it goes into Word, use PDF to Text instead, in these steps.
- Open PDF to Text and add the scanned PDF. Pages that already have text are read directly; the others go through OCR.
- Choose the language of the document before you start. The engine uses one recognition model per language, and you can select more than one.
- Select Extract text, read the result in the text box and correct names, amounts and dates against the scan.
- Select Download Word (.docx). Each paragraph found on the page becomes a paragraph in Word, with the document language set for spell checking.
- Rebuild headings, tables and columns in Word. OCR does not bring back the original fonts or table lines.
Scan resolution decides how many errors you get
OCR reads pixels, so the scan resolution matters most for small print. We drew the same page twice, with 11 pt text and with 8 pt text (the size of footnotes and terms on forms), and scanned each at four resolutions.
| Scan resolution | Pixels of the A4 page | 8 pt text | 11 pt text |
|---|---|---|---|
| 100 DPI | 827 × 1169 | 16 of 788 wrong (2.0 %) | 0 wrong |
| 150 DPI | 1240 × 1754 | 0 wrong | 0 wrong |
| 200 DPI | 1654 × 2339 | 0 wrong | 0 wrong |
| 300 DPI | 2480 × 3508 | 0 wrong | 0 wrong |
Sources
- Tesseract documentation: Improving the quality of the output (checked Nov 10, 2026)
What the resolution test means for your scans
At 100 DPI the 11 pt text was still read perfectly, but 8 pt print came back with digits and similar letters mixed up, the kind of error that changes a time or an amount without looking wrong. From 150 DPI the same page had no errors. The Tesseract engine that MAXScanner uses recommends at least 300 DPI, which leaves a margin for creased paper, faint print and photos taken at an angle.
So if you are scanning now, scan at 300 DPI. If all you have is an old 100 DPI scan, run it anyway, then check every number by eye.
Tables and bank statements: go to Excel, not Word
Word is a poor target for a scanned table. OCR returns each row as a line of text, with spaces or tabs between the columns, as our timetable showed, and you would rebuild the grid by hand. For a scanned bank statement, run OCR PDF first and then open the searchable PDF in Bank Statement to Excel, which checks every running balance, so a misread digit shows up as a broken balance.
Other ways to do it, and their limits
Desktop Microsoft Word can open a PDF and convert it. Microsoft says this works best with documents that are mostly text and that the result may not match the original page for page. Google Docs converts PDFs and images too, but only files of 2 MB or less with text at least 10 pixels high, and it keeps bold, italics and line breaks but usually not tables or columns. Free online converters often keep OCR for paying users: on the pages we read, Smallpdf and iLovePDF both mark their OCR option as a paid feature, and both upload the file to their servers.
Sources
- Microsoft Support: Scan and edit a document (checked Nov 10, 2026)
- Google Drive Help: Convert PDF and photo files to text (checked Nov 10, 2026)
When this is not the right approach
If the document has to look exactly like the original, such as a certificate or a signed form, keep it as a PDF. Make it searchable with OCR PDF, or change the few words that need changing with Edit PDF, which can replace text on scanned pages.
- Handwriting: the engine is trained on printed text and reads handwriting poorly.
- Complex layouts: two columns, text boxes and footnotes come out in reading order, not in their positions.
- Long documents: OCR took 3 to 6 seconds per A4 page in our runs on a computer; a phone is slower.
- Privacy: PDF to Word and PDF to Text run the OCR in your browser and do not upload the PDF. Check the privacy terms of any converter that does.
Frequently asked questions
Why can’t I edit my scanned PDF?
Because each page is a picture, and editors and converters can only change text that exists as characters. OCR turns the picture into characters: MAXScanner’s PDF to Word and PDF to Text both run it in your browser and give you a Word file you can edit.
Can I convert a scanned PDF to Word without losing formatting?
Not completely. OCR recovers the words and paragraphs, but fonts, exact positions, columns and table lines have to be rebuilt in Word. If the look matters more than editing, make a searchable PDF with OCR PDF instead.
Why is my Word file just a picture of the page?
The converter found no text in the PDF, so it copied the scanned page into Word as an image. Before we changed it, our own PDF to Word did this too: a 3.1 MB Word file with no words in it. Use a converter that runs OCR, such as MAXScanner’s PDF to Word or PDF to Text.
Is OCR accurate enough for numbers and dates?
On a clean scan of 150 DPI or more, our test page came back with no errors, times and dates included. At 100 DPI, 8 pt print had 16 wrong characters out of 788, mostly digits. Always check amounts, dates and names against the scan.
How do I convert a scanned PDF to Excel?
For a bank statement, run OCR PDF and then open the searchable PDF in Bank Statement to Excel, which checks the running balance. Avoid going through Word for tables: OCR into Word loses the columns.
Is it safe to convert confidential scans online?
With MAXScanner the OCR runs in your browser and the PDF is not uploaded. Many online converters upload the file to their servers, so read their privacy terms before using them for contracts, ID documents or bank statements.
Keep reading
What DPI Means: Pixels, Print Size and KB
DPI only tells software how large to print the pixels. We wrote 72, 300 and 600 DPI into one scan and every file stayed 437,719 bytes. Here is what really changes the size, and the pixels each print needs.
Read the related guide →Which Barcode Do I Need? Types and Sizes
Products sold in shops need an EAN-13 or UPC-A number licensed from GS1; your own stock and shipping labels can use Code 128 for free. We encoded the same data in each type and measured width and scanning.
Read the related guide →