How to Convert a Scanned PDF to Excel on Mac (OCR Methods That Work)
You open a PDF, try to select a number in the table and nothing highlights. The file is a scan: a picture of a page, stored in a PDF wrapper. Excel can do nothing with a picture until optical character recognition, OCR, reads the pixels and turns them into characters. On a Mac you have several ways to do that, some of them free and local. This guide goes through them and, more importantly, how to check the result.
Quick answer: First test for a text layer: press ⌘F in Preview and search for a word in the table. If it is not found, the PDF is a scan. Export the page as an image (File > Export, choose PNG), then use Insert > Data From Picture in Excel (Microsoft 365). For a long scan, run OCR on the whole file in a PDF tool first, then copy or export the tables. Whatever you use, compare the totals with the page.
Is it really a scan?
Two quick tests settle it. Search for a word with ⌘F, and try dragging the text selection tool across a line. If either works, the PDF has a text layer, and you do not need OCR. Our guide to copying a table from a PDF into Excel is the better route, and more accurate. If neither works, you are looking at an image.
Some PDFs are hybrids: the page is an image, but an earlier OCR pass added an invisible text layer, which is how scanners and apps produce searchable PDFs. Those tend to select fine, but the invisible text can contain recognition errors. Check them like any other OCR result.
| Your scan | Try first | Then | Caution |
|---|---|---|---|
| Clean, one table per page, recent macOS | Preview: test whether text can be selected | Paste and split into columns | Table structure may not survive the copy |
| Clean, one table, Microsoft 365 | Export the page as an image, then Data From Picture | Review preview, Insert Data | One page at a time |
| Long document, many pages | OCR the PDF once (for example with Acrobat Pro), then extract | Copy or export tables | Paid software, check the licence |
| Poor scan: skewed, faint, handwritten | Re-scan at 300 dpi if you can | Then any route above | Expect heavy manual correction |
| Sensitive document | A local method only | Verify against totals | Avoid upload sites |
Method 1: Try Live Text in Preview
Recent versions of macOS can recognise text in images, a feature called Live Text, and it can work on pages within Preview. Open the scanned PDF, try selecting text in the table, and if it highlights, copy it. Paste into Excel as text with ⌃⌘V, then split it with Data > Text to Columns. The steps are in our guide to cleaning PDF data pasted into Excel.
The catch is that Live Text gives you text, not a table. We found no source confirming that it keeps rows and columns apart, so test on your own page. Quality also depends on the scan, and results vary between macOS versions.
Method 2: Export the page as an image and use Data From Picture
This is the better route when you have Microsoft 365 and a single clean table.
In Preview, go to the page with the table.
Choose File > Export, set the format to PNG or JPEG, and save. Alternatively, take a screenshot with ⌃⌘⇧4 of just the table.
In Excel, choose Insert > Data From Picture > Picture From File (or From Clipboard for a screenshot).
Review the preview, correct the highlighted cells and insert.
Our full guide to converting an image to Excel with Data From Picture covers the review step and a table of typical OCR mistakes. For a multi-page scan, you repeat this page by page, which is fine for five pages and tedious for fifty.
Method 3: OCR the whole PDF, then extract
For a long document, the sensible order is to OCR the entire file once and then treat it like any other text PDF. Adobe Acrobat Pro has a Recognize Text feature, and other PDF and OCR apps offer similar functions. These are generally paid tools, so check the licence before you plan a project around them. Once the PDF has a text layer you can copy the tables, or use a local extraction tool as described in the guide to turning a PDF into a CSV on a Mac.
Note that exporting a PDF to Excel inside Acrobat is also a paid Acrobat feature, and the free Reader can copy text but not export. That is a common source of confusion, and the reason people hit a prompt to upgrade.
Method 4: Online OCR converters
Websites that offer OCR to Excel are convenient, and some are good. They also receive your file. For scans of invoices, ID documents or bank statements that is a significant decision. Read the retention policy, and use a local method for anything sensitive.
Make the scan easier to read
OCR accuracy is mostly decided before you press any button.
Scan at 300 dpi or higher, in black and white or greyscale for text.
Keep the page straight. Skew makes column detection fail.
Avoid shadows, creases and the edge of a book.
Scan one table per page when possible.
Use a flatbed scanner or a scanning app with edge detection rather than a casual photo.
Verify every number
This is the step that matters most with OCR. A recognised table can look entirely plausible and still be wrong in a few places, a 5 read as an S, a missing decimal point, a dropped minus sign. Use at least three checks.
Totals. Sum each numeric column and compare with the printed totals on the page.
Row count. The number of rows should match the original.
Spot checks. Read the first, middle and last rows aloud against the scan.
Format checks. Number cells that are left-aligned are text, which usually means a letter crept in.
If the numbers feed anything that matters, such as accounts or a client report, do a full read-through. OCR is a way to avoid typing, not a way to avoid checking.
Common mistakes
Using OCR on a PDF that already has text. Copying is more accurate.
Trusting a clean-looking result. Errors are usually invisible without a total check.
Scanning at low resolution to keep the file small.
Uploading confidential scans to the first free site you find.
Expecting formulas. OCR returns values only.
Frequently asked questions
Can Excel for Mac convert a scanned PDF directly?
Not by itself. Excel can read an image of a table through Data From Picture, but it does not open scanned PDFs as data.
Is there a free way to OCR a PDF on a Mac?
Partly. Live Text in Preview and Data From Picture (with a Microsoft 365 subscription) cost nothing extra, and they work on one page at a time. Whole-document OCR usually means a paid tool.
Why are some digits wrong after OCR?
Similar-looking characters get confused, such as 0 and O or 1 and l. A low-quality scan makes it worse.
How do I OCR many pages at once?
Run OCR on the whole PDF in a PDF tool, which adds a text layer, then extract the tables from the result.
Does the scan language matter?
Yes. OCR tools work best when the language and character set of the document is supported, so check the options if your document is not in English.