# Online OCR PDF to Word: From Upload to Saved File

- Author: Schema Law (https://www.schemalaw.com/)
- Published: 2023-04-11T11:09:59Z
- Updated: 2026-10-11T19:22:20Z
- Category: Converting scans
- Canonical: https://www.schemalaw.com/blog/online-ocr-pdf-to-word

> An online OCR PDF to Word conversion from start to finish: upload, wait, review, download, save to the matter and remove the scan. Plus stalled uploads.

An online OCR PDF to Word conversion has six steps, all in a browser tab: open the converter, upload the scanned PDF, wait while the pages are read, look at the result, download the Word file, and save it to the matter. Then remove the scan from the converter and the stray copies from the computer.

The recognition is the part you cannot see and do not control. Everything on either side of it is ordinary browser work: choosing a file, keeping a tab open, finding a download, signing out. Those are also the steps where a file goes missing or a copy is left behind, so they repay a little care.

## What happens to a scanned PDF between the upload and the Word download?

Between the upload and the download, a scanned PDF passes through four stages: the file travels from your computer to the service, it waits its turn, each page is read into text, and a Word file is built from that text. Only then is there anything to download.

The post on online OCR text recognition explains [what happens to a page between upload and recognized text](https://www.schemalaw.com/blog/ocr-text-recognition-online#what-happens-to-a-page-between-upload-and-recognized-text) and where the reading takes place. For the purpose of getting a Word file, what matters is what the browser shows as the job moves along and what you should be doing meanwhile.

| Moment | What the browser shows | What you do |
| --- | --- | --- |
| Upload | A progress bar or a spinner beside the file name | Leave the tab open and the computer awake |
| Waiting and reading | A status such as processing, or a page counter | Nothing; a long scan takes longer than a short one |
| Result | The recognized pages on screen, or a notice that the file is ready | Look at the pages before exporting, if the service shows them |
| Export | A choice of format and a download button | Choose Word and note the file name |
| Download | The browser's own download notice | Find the file and move it to the matter folder |

Schema Law's homepage lays out its version in three steps: upload a PDF, PNG or JPEG of up to 20 MB from the browser; each page is read into text and tables and shown page by page; then you choose the file type and download it. The [how it works section](https://www.schemalaw.com/#how-it-works) has the wording.

### Two things the stages tell you

First, the Word file does not exist until the reading is done. Closing the tab during the upload stops the job. Closing it during the reading may or may not, depending on the service, and the only way to know is to sign back in and look.

Second, the copy on the service and the copy on your computer are separate. Downloading the Word file does not remove the scan from the service. That is a job of its own, at the end.

## Running an online OCR PDF to Word conversion from a browser tab

Open the converter in a current browser, sign in if it asks, upload the scanned PDF, keep the tab open until the pages are ready, review them, export Word, and save the download to the matter folder under a clear name. The numbered procedure takes each step in turn.

1. **Check the file before you open the browser.** Note its size in megabytes and its page count. Confirm it opens and that the pages are the right way up.
2. **Decide whether the document may be uploaded at all.** For a client document, your firm's policy on online tools comes first.
3. **Open the converter and sign in.** Type the address or use a bookmark you made yourself.
4. **Check the limits against your file.** Compare the size and page count you noted with what the service accepts on your plan. The post on [fitting a document inside a free OCR allowance](https://www.schemalaw.com/blog/free-ocr-pdf-to-word) helps when the page count is over a free cap.
5. **Upload.** Use the upload button or drag the file onto the page. Watch for the file name to appear, so you know the right file went up.
6. **Wait in the same tab.** Leave the computer awake. Work in another window if you like.
7. **Review the result on screen** if the service shows it. Look at the first page, the last page and one from the middle. A blank or scrambled page here means the scan needs attention before you export anything.
8. **Export as Word.** Choose DOCX or Word from the list of formats and start the download.
9. **Open the download and count the pages.** Compare with the scan. Read the first and last lines of the document against the first and last pages of the PDF.
10. **Save it into the matter.** Use Save As to put the file in the matter folder under a name that says what it is, such as `Ferris v Okonjo - lease 2019 (OCR conversion).docx`.
11. **Clean up.** Remove the scan from the service and the leftover copies from the downloads folder.

### Timing your attention

Steps 1 to 5 need you at the keyboard, as do steps 7 to 11. Step 6 is waiting. If you have several files, start with the one you need first, and use the wait to set up the folder and file names for the rest.

### What the procedure assumes

It assumes a scan that fits the service's limits and a network that allows the upload. When either assumption fails, the upload stalls or is refused.

## Why does an online PDF to Word upload stall or fail?

When an online PDF to Word upload stalls or fails, check four things first: whether the file is larger than the service accepts, whether it is a type the service takes, whether the connection dropped in transit, and whether your firm's network stopped the upload. The error message, or the lack of one, points to the cause.

| What you see | Likely cause | What to check |
| --- | --- | --- |
| An immediate refusal naming a size | File over the size limit | The file's megabytes against the stated limit |
| An immediate refusal naming a format | Unsupported file type | The file extension; whether it is really a PDF |
| An immediate refusal naming pages | Over the plan's page limit | The page count against your plan |
| The progress bar stops partway | Connection dropped or slowed | Whether other sites load; whether you are on Wi-Fi or a VPN |
| The bar never starts, or a blank page | Network filter, or a browser extension | Whether a firm block page appeared; a different approved browser |
| A request for a password, or an error on a locked file | The PDF is protected | Who set the protection and why |
| You are returned to the sign-in page | The session timed out | Sign in and start the upload again |

### File size

A scanned PDF is a stack of pictures and can be heavy for its page count. Find the service's stated limit and compare it with the file. The post on converting large scans explains [how large a scanned PDF an online converter can take](https://www.schemalaw.com/blog/convert-large-scanned-pdf-to-word-online#how-large-a-scanned-pdf-can-an-online-converter-take) and how to split one that is over.

### File type

Check what the service accepts. Schema Law, for example, takes a PDF, PNG or JPEG of up to 20 MB. A scan saved as TIFF, a phone photo saved as HEIC, or a PDF inside a ZIP file will be refused by a service that does not list those types. The post on [turning a scanned image into text](https://www.schemalaw.com/blog/scanned-image-to-text) covers other image formats.

### Protected PDFs

If the file asks for a password or reports that it is protected, someone restricted it on purpose. Go back to whoever sent it and ask for an unrestricted copy or for the password. The post on [converting a protected PDF to Word](https://www.schemalaw.com/blog/convert-protected-pdf-to-word) explains the kinds of protection you may meet.

### The connection

A large upload over a weak Wi-Fi signal or a busy VPN can time out. Test by loading another site. If the connection is the problem, move to a wired connection or a stronger signal and upload again from the start. A partly uploaded file is of no use to the service.

### Firm network blocks

A firm may filter the sites its computers can reach or stop files being uploaded to services it has not approved. If you see a block page, or uploads fail on the office network and nowhere else, that is a decision your firm has made. Do not route around it with a phone hotspot or a personal laptop. Ask IT whether the service is approved, and if it is not, ask what is.

## Where did the downloaded Word file go, and which copy is current?

The downloaded Word file is in your browser's downloads folder unless you chose somewhere else. The current copy is the one you saved into the matter folder and have been editing. Every other copy, in the downloads folder or exported again from the service, is a duplicate to delete or ignore.

### Finding the download

Google's help page on [downloading a file in Chrome](https://support.google.com/chrome/answer/95759) gives the default location on Windows 10 and later as the Downloads folder under your user name, with matching folders on Mac and Linux. The same page describes a Downloads entry in Chrome's More menu that lists what you have downloaded. Other browsers have their own downloads list in a similar place.

If the file is not where you expect:

1. Open the browser's downloads list and use its option to show the file in its folder.
2. Sort the Downloads folder by date modified, newest first.
3. Search the computer for the file name the service gave the export.

Chrome's help page also describes a setting, under Settings and then Downloads, to ask where to save each file before downloading. With it switched on you choose the matter folder at the moment of download, and nothing lands in Downloads.

When that download is an answer and counterclaim, save it into the matter folder before you edit a word. Check the caption, each affirmative defense and the relief the counterclaim asks for. A paragraph number that shifted in conversion will throw off every later reference to that paragraph.

### Duplicate names

Download the same export twice and the browser keeps both, adding a number to the second: `lease.docx` and `lease (1).docx`. Export again next week and there is a third. None of the names says which one you edited.

The fix is a habit, not a tool:

- **Download once.** If you need the file again, look in the matter folder first.
- **Rename on saving.** The name the service gave the export is not the name the file should keep.
- **Edit only the copy in the matter folder.** Never the one in Downloads.
- **Delete the Downloads copy** as soon as the matter copy is saved and opens.

### A worked example

A paralegal converts a scanned lease for the fictional matter Ferris v Okonjo on Monday and opens the export straight from the browser's download notice. She corrects twelve names and figures and saves. On Thursday a colleague needs the file, cannot find it in the matter folder, signs in to the service and exports it again. That copy has none of the twelve corrections. Monday's corrected file is still in the paralegal's Downloads folder under the export's original name.

One step would have prevented it: Save As into the matter folder before the first correction. When you do find yourself with two copies and no memory of which is which, Word can compare two documents and mark the differences, which shows at once which copy carries the corrections.

## Converting online from a locked-down firm computer or a borrowed one

A browser converter needs no installation, so it runs on a firm computer where you cannot install software. Whether you are permitted to use it there is a separate question for your firm. On a shared or borrowed machine, the conversion is the easy part, and clearing your traces afterwards is the real work.

### On a locked-down firm computer

Not having install rights is a restriction on software. It says nothing about which websites are approved. Before using an online converter on a firm machine, find out whether the firm has approved one, and use that. The post on [online OCR versus desktop OCR for law firms](https://www.schemalaw.com/blog/online-ocr-vs-desktop-ocr-for-law-firms) sets out the wider choice.

Three practical points on a managed machine:

- Downloads may be redirected to a network location. Find where yours go before the first conversion.
- The browser may be a fixed version. If the converter's page does not display properly, report it to IT instead of trying another browser you were not given.
- The screen may lock after a few minutes idle. Find out whether an upload carries on behind a locked screen on your machine before you rely on it for a long file.

### On a shared computer in the office

A shared workstation at a reception desk or in a copy room keeps what the last person left: a signed-in browser, a downloads folder, a list of recent files. Before you leave it, confirm each of these.

- Signed out of the converter, using its sign-out control and not only by closing the tab
- Signed out of the account you used to sign in, if the browser kept it
- Declined the browser's offer to save your password
- Moved the Word file to the matter folder on the network
- Deleted the scan and the Word file from the local Downloads folder
- Emptied the recycle bin
- Cleared the browser's downloads list

### On a borrowed computer

A borrowed computer, whether a home machine, a client's office PC or a hotel business centre, is outside your firm's control. Ask first whether the document should be on it at all. Your firm's policy may answer that with a flat no for client files.

If you may go ahead, add these to the shared-computer list: use a private browsing window so the browser does not keep history or sign-ins after you close it; do not save the scan or the Word file to the machine if you can avoid it; and if you must, delete them and empty the bin before you hand the machine back. Remember that a file you opened in Word may also appear in Word's list of recent documents.

## Removing the scanned PDF from an online converter once the Word file is saved

Once the Word file is saved in the matter folder and opens correctly, delete the scanned PDF from the converter, then delete the local leftovers. Do it in that order. Until the saved Word file has been opened and its page count checked, the copy on the service is your way to export again.

### The order of cleanup

1. **Confirm the saved file.** Open the Word file from the matter folder, not from Downloads. Check the page count against the scan.
2. **Decide whether you will need another format.** If you may want a searchable PDF or a spreadsheet of the same document, export it now. Schema Law's site says you can export the same document again in a different format without a second upload.
3. **Delete the document in the service.** Find the document in your list and use the delete control. Confirm that it has gone from the list.
4. **Delete the local leftovers.** Remove the downloaded export from the Downloads folder, and the copy of the scan if you put one on the desktop for uploading.
5. **Clear the browser's downloads list** if the machine is shared. Chrome's help page notes that removing an item from that list does not delete the file from the computer, so do both.
6. **Sign out.**

### What deleting from the service covers

What a delete button removes, and how soon, is for each service to state. Read the service's own page on it before you rely on it for a client document. Schema Law's commitment reads: "Delete any document whenever you choose. The original file and everything parsed from it are removed." The [security commitments](https://www.schemalaw.com/#security) are set out on the homepage.

The post on [whether OCR software stores your documents](https://www.schemalaw.com/blog/does-ocr-software-store-my-documents) explains what a converter can keep after you finish and the questions to put to a vendor.

### A note in the file

For a client matter, record the conversion in one line: the date, the service, the document, and the date you deleted the upload. If anyone later asks where a client document has been, the answer is in the file.

## 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)