# Smallpdf OCR: What the Online Tool Does to a Scan

- Author: Schema Law (https://www.schemalaw.com/)
- Published: 2026-10-09T03:06:44Z
- Updated: 2026-10-11T19:22:20Z
- Category: OCR in everyday software
- Canonical: https://www.schemalaw.com/blog/smallpdf-ocr

> Smallpdf OCR adds a hidden text layer to a scanned PDF in the browser. What Smallpdf's own pages say about limits, file types, languages and deleting uploads.

Smallpdf OCR is a browser tool that takes a scanned PDF and returns a copy with a hidden text layer, so the words can be searched, selected and copied. Smallpdf says the tool is free with limits and that uploads are deleted after one hour. A lawyer should read what sits behind both statements before uploading a client's scan.

This post describes the tool only from Smallpdf's own pages: its PDF OCR tool page, one article on its blog, its pricing page, its Privacy Notice and its Terms and Conditions, all opened on October 9, 2026. It does not rate how well the tool reads a page. Where Smallpdf's pages say different things about the same point, both statements are given with the page each came from.

## What does Smallpdf's OCR do to a scanned PDF?

Smallpdf's OCR adds text to a file that had none. Smallpdf's article on [searchable and non-searchable PDFs](https://smallpdf.com/blog/searchable-vs-non-searchable-pdfs-what-you-need-to-know) says the tool "creates a new version of your PDF with a hidden text layer" and adds: "The original layout and images stay in place." You get back a PDF that looks the same and can be searched.

The [Smallpdf PDF OCR tool page](https://smallpdf.com/pdf-ocr) gives the procedure in four steps: import or drag the file in, wait while recognition runs, edit the PDF with Smallpdf's other tools if needed, then download or share the searchable file. The blog article adds that you can import from Google Drive, Dropbox or OneDrive as well as from your device.

Three points follow from those descriptions.

- **You receive a new file.** The scan on your computer is not altered. You now have two files, and your naming has to show which is which.
- **The output is a PDF, not a Word document.** The tool page treats an editable Word file as a second step through a different Smallpdf tool, PDF to Word.
- **Nothing on the pages opened for this post says the tool marks doubtful words.** Plan to check the result yourself.

What a text layer is and how a viewer uses it is explained in the guide to [how to make a scanned PDF searchable](https://www.schemalaw.com/blog/how-to-make-a-scanned-pdf-searchable). The tool page also makes a modest claim about accessibility: recognized text can help screen readers, but "full PDF accessibility may require further changes."

## Is Smallpdf OCR free, and what does Pro add?

Smallpdf says yes, with limits. The tool page's FAQ reads: "Our online PDF tools are free to use, though some limits apply." Smallpdf's pricing page marks PDF OCR on the Free plan with the words "Daily download limit" and lists batch conversion as a paid feature. Neither page puts a number on the daily limit.

Here is what the [Smallpdf pricing page](https://smallpdf.com/pricing) showed for the rows that bear on OCR when it was opened for this post. Prices are left out on purpose, because they change; read them on that page.

| Row in Smallpdf's comparison table | Free | Pro |
| --- | --- | --- |
| PDF OCR | "Daily download limit" | Included |
| OCR (PDF to Word, PDF to Excel, PDF to PPT) | Not included | Included |
| Batch Convert | Not included | Included |
| File size limit | "Limited" | "Unlimited" |

The plan summaries on the same page describe Free as having "Limited document downloads" and Pro as having "Unlimited document downloads".

### What batch means here

The blog article has a section headed for Pro users on converting multiple PDFs. It says "Smallpdf Pro lets you run OCR on multiple PDFs in one go": you sign in with a Pro account, upload several PDFs at once, start OCR and download the results. On the free tier, by the pricing table, you work one file at a time.

The same article gives advice that applies on either tier: "If you work with many scans, start with one file first to check the OCR result before converting a full batch."

### What the pages do not say

No page opened for this post states how many downloads the daily limit allows, how large a file the Free plan accepts, or how many pages one scan may have. If you are deciding whether the free tier fits a job, the only way to find out is to try a non-confidential file of the same size and see where the tool stops you.

One more point for firms. Smallpdf's terms attach a Data Processing Agreement to the Team and Business plans by name. Section 16 of the [Smallpdf Terms and Conditions](https://smallpdf.com/terms) says the agreement covers processing of personal data "in connection with the Smallpdf Team Plan and Smallpdf Business Plan". If your firm needs a signed processing agreement before a vendor touches client data, ask Smallpdf which plans carry one.

## Which file types and languages Smallpdf says its OCR accepts

Smallpdf's tool page gives two lists of file types that do not match, and no list of languages. The upload box names PDF, DOC, XLS, PPT, PNG and JPG. The FAQ on the same page says the tool "accepts PDF, Word, PowerPoint, Excel, JPG, BMP, GIF, TIFF, and more."

### File types

| Where on the tool page | What it lists |
| --- | --- |
| The upload box, under "Supported formats" | PDF, DOC, XLS, PPT, PNG, JPG |
| The FAQ answer on supported formats | PDF, Word, PowerPoint, Excel, JPG, BMP, GIF, TIFF, "and more" |

For a law office the gap matters for TIFF, the format some document productions and fax servers deliver. One list includes it and the other does not. Test a single non-confidential TIFF before you plan a job around it. Image formats as OCR input are discussed in the post on [converting a scanned image to text](https://www.schemalaw.com/blog/scanned-image-to-text).

### Languages

On languages, Smallpdf's pages speak in general terms:

- The tool page FAQ says the tool "can recognize text in many different languages".
- The blog article's FAQ says it "supports many widely used languages such as English, Spanish, French, German, Italian, Portuguese, and more", and that the engine can detect the document language automatically "in most cases".
- The same article, in its troubleshooting section, tells you to check that your OCR settings match the document language, "where supported."

Read together, those statements say detection is automatic in most cases and a setting exists in some. The tool page itself says the tool has "no complicated settings". So do not count on a language picker. If your file holds a Spanish-language exhibit or a French contract, run one page and search it for a word with an accented letter before you trust the rest.

## What does Smallpdf say happens to an uploaded file?

Smallpdf's tool page says: "Uploaded files are protected with 256-bit TLS encryption and deleted after one hour unless you save them to your Smallpdf account." Its Privacy Notice is more specific and ties the one-hour figure to signed-in use. Read both before relying on the shorter sentence.

### The deletion statements, side by side

| Smallpdf page | Statement |
| --- | --- |
| PDF OCR tool page | Files are "deleted after one hour unless you save them to your Smallpdf account." |
| Privacy Notice, section 6, signed-in use | "If you access our services via a User Account, we delete User Files within one hour unless you save them to your file storage." |
| Privacy Notice, section 6, no account or not logged in | "we will generally aim to delete User Files within a reasonable period of time after the last time they were opened. Please note that this retention period is extended every time you reopen the respective User File." |
| Privacy Notice, section 6, files you saved | "When you choose to delete saved User Files, we generally delete them within 14 days." |
| Blog article FAQ | Smallpdf "deletes your files automatically after a short period." |

The [Smallpdf Privacy Notice](https://smallpdf.com/privacy) carried the date 29 July 2025 when it was opened. The tool page invites use with "no signups". Under the notice's wording, a visitor who is not logged in falls under the "reasonable period of time" sentence, not the one-hour sentence. The pages do not reconcile the two. If the one-hour figure matters to your decision, ask Smallpdf in writing which rule applies to the way you would use the tool.

### Three more statements worth reading

**Shareable links.** Section 1.10 of the Privacy Notice says: "By default, the documents you process using our services are accessible to anyone with a unique sharable URL. This can be disabled by the User". Find that setting before you process anything sensitive, and do not paste the link into an email.

**Smallpdf's own access.** Section 3.2 of the terms says: "We will only access, view, or listen to your content in limited ways." Its examples are responding to support requests, addressing fraud, security, unlawful or technical issues, improving the services, and enforcing the terms.

**Where files are processed and whether they train models.** The Privacy Notice names a hosting provider used to process uploaded files and says the files may be temporarily stored on servers in the European Union. The blog article says Smallpdf "does not use your documents as training data". What those two kinds of statement are worth, and how to check them, is covered in the posts on [OCR data residency](https://www.schemalaw.com/blog/ocr-data-residency) and [OCR that does not train on your data](https://www.schemalaw.com/blog/ocr-that-does-not-train-on-your-data).

Encryption during upload is one sentence on Smallpdf's page and a subject of its own; see the post on [encrypted OCR](https://www.schemalaw.com/blog/encrypted-ocr). How deletion windows differ from one converter to another, and what to ask about backups, is in the post on whether [OCR software stores your documents](https://www.schemalaw.com/blog/does-ocr-software-store-my-documents). For a second published window to compare, the post on [iLovePDF's OCR PDF tool](https://www.schemalaw.com/blog/ilovepdf-ocr) quotes that company's figure.

## Smallpdf's own advice on which scans recognize well

Smallpdf says results depend on the scan. The tool page puts it in one sentence: "Results depend on the clarity, contrast, and resolution of the original scan." The blog article's troubleshooting section turns that into specific advice about resolution, flat and well-lit pages, language, fonts, stamps and handwriting.

The advice in that section, grouped by the problem Smallpdf attaches it to:

| Problem, in Smallpdf's words | What Smallpdf suggests |
| --- | --- |
| The scan is "blurry, dark, or skewed" | Rescan at 300 DPI or higher; make sure the page is flat and well-lit; use a scanner instead of a phone camera when possible |
| "words full of strange characters" | Check that OCR settings match the document language, where supported; avoid unusual decorative fonts and heavily stylized text |
| Text is missing or jumbled | Increase contrast before you scan, or use a scanner setting that reduces background noise; avoid scribbles, stickers or stamps over critical text |
| Handwriting | "OCR works best with printed characters." |

On handwriting the article's FAQ goes a little further: "For critical information in handwriting, you may still need to review and correct the recognized text manually."

### What a law office can and cannot act on

Some of that advice assumes you control the paper. You can rescan your own client's documents at a higher resolution. You cannot rescan the other side's production, and you cannot remove a received stamp from a filed pleading. For those files the useful part of Smallpdf's advice is the warning: a stamp over text or a faint fax is a place to look for errors, not a thing to fix.

Smallpdf's suggested remedy for leftover errors is to open the searchable PDF in its editor or a word processor "and manually fix the few lines that need correction." For a working copy that is fine. For a document whose text you will quote or rely on, check it against the page image; the method is in the post on [why OCR output is full of errors](https://www.schemalaw.com/blog/why-is-ocr-output-full-of-errors). How much resolution recognition needs is explained in the post on the [best DPI for OCR](https://www.schemalaw.com/blog/best-dpi-for-ocr).

### A two-minute acceptance test

The blog article describes a check you can run on any PDF, and it works on the file Smallpdf returns:

1. Open the downloaded PDF and drag across one line. Single words should highlight, not the whole page as a block.
2. Press Ctrl+F, or Command+F on a Mac, and search for a word you can see on the first page.
3. Repeat the search with a name or number from the last page.
4. Copy one sentence into a blank document and read it against the page.

If step 3 fails while step 2 passes, the text layer does not cover the whole file.

On the copy you download, look for the parties, the date of the notice or the recording stamp, and the exhibit label printed on the page. Where the scan is an affidavit or a signed agreement, check the date by the signature and one obligation you can read, then match those hits to the image. Put the file in the matter after a middle page passes the same look.

## Questions to settle before a client's scan goes into Smallpdf

Settle who is allowed to decide, which of Smallpdf's deletion rules applies to you, whether the shareable link is off, and what the document is. Smallpdf's Privacy Notice says the responsibility for personal data in an upload stays with you, so these are your questions to answer, not the tool's.

Section 1.10 of the Privacy Notice puts it this way: files "may contain personal data related to you or others and you remain fully responsible for any personal data contained in the User Files."

### The checklist

- **Does your firm allow it?** Look for a written policy on browser tools. If there is none, ask the person responsible for the firm's confidentiality practices before the upload, not after.
- **Does anything attached to the document forbid it?** A protective order, a client's outside counsel guidelines, or a confidentiality agreement may restrict who can receive the document. Check the document's own restrictions first.
- **Will you be signed in?** The Privacy Notice gives a one-hour rule for account use and a looser sentence for use without an account. Know which one you are under.
- **Is the shareable link disabled?** The notice says documents are reachable by anyone with the link by default.
- **Which plan is the account on?** The terms tie the Data Processing Agreement to the Team and Business plans.
- **Are you saving the file to Smallpdf storage?** Saved files are outside the one-hour rule, and the notice says deleted saved files are generally removed within 14 days.
- **Have you read section 3.2 of the terms?** It lists the limited ways Smallpdf may access content, and service improvement is on the list.
- **Does the client need to be told?** That depends on your jurisdiction's rules and your engagement terms. This post does not answer it.
- **Would a tool on your own computer do the job?** For a single sensitive document, that may remove most of the questions on this list.

### A worked example

Take Priya Nambiar, an associate with two scans on her desk. One is a 1998 recorded easement pulled from the county's public records. The other is a client's unsigned settlement draft with medical details in it.

The easement is a public record. The questions about confidentiality fall away, and the only test left is whether the output is good enough. The settlement draft is different on every line of the checklist: her firm's policy, the client's expectations, the account she would be signed in to, the default link. She runs the easement through the browser tool and takes the settlement draft to software her firm has already approved.

The point of the example is that the decision is made per document, not per tool. The broader question of [whether free online OCR is safe for legal documents](https://www.schemalaw.com/blog/free-online-ocr-safe-for-legal-documents) and the trade between [online OCR and desktop OCR for law firms](https://www.schemalaw.com/blog/online-ocr-vs-desktop-ocr-for-law-firms) each have their own post.

The same questions apply to every service that receives an upload, Schema Law included. Its four standing commitments about documents are printed in the [security section of the homepage](https://www.schemalaw.com/#security); read them with the same care you would give Smallpdf's.

## 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)