How to Extract Tables from Scanned PDFs into Excel (Without Retyping a Single Cell)

You've got a stack of scanned invoices, a folder of archived purchase orders, or a PDF that came straight off a photocopier. The data you need is right there — visible on the page — but every tool you've tried either garbles the columns, misses rows, or simply refuses to process an image-based file.

This is the core frustration with scanned PDFs: unlike native digital documents, they're essentially just pictures of text. Standard copy-paste doesn't work. Manual retyping is slow, error-prone, and frankly a waste of time. The good news is that AI-powered extraction has made this problem largely solvable — and this guide shows you exactly how to do it.

Step 1: Understand What You're Working With

Before you start, it helps to know what type of document you have. Not all "scanned" files behave the same way.

• Scanned image PDFs: The entire page is a flat image. No text layer exists. These require OCR (Optical Character Recognition) to extract anything.
• Poorly exported PDFs: They may have a hidden text layer, but the layout is scrambled — tables break into single cells or columns merge unpredictably.
• Photo images (JPG, PNG): Taken with a phone or scanner, these are purely visual and need the same OCR approach as scanned PDFs.

The extraction method is the same for all three, but knowing the source helps you set realistic expectations for the first pass.

Step 2: Choose a Tool That Understands Tables, Not Just Text

Most generic OCR tools can read words off a page — but they collapse table structures completely. What you need is a tool that recognises rows, columns, and cell boundaries as a structure, not just a sequence of characters.

This is exactly where AI-driven extraction earns its keep. Instead of treating a table as lines of text, it interprets the grid layout and maps each value to the correct cell in a spreadsheet.

For scanned documents specifically, the Scanned PDF to Excel Converter preset is purpose-built for this scenario — handling skewed scans, low-resolution images, and multi-column layouts without manual cleanup. If your source is a photo rather than a PDF, the Image to Excel Converter preset covers the same ground for JPG and PNG files.

Step 3: Prepare Your Document for Best Results

AI extraction is resilient, but a little preparation reduces errors significantly — especially with older archive documents.

• Scan at 300 DPI or higher. Lower resolution makes character recognition unreliable, especially for small fonts or dense tables.
• Keep pages straight. A slight rotation of even 5–10 degrees can confuse column alignment. Most scanners have an auto-deskew option — use it.
• Remove physical obstructions. Fingers at the edge of a phone photo, sticky notes, or folded corners all interfere with extraction.
• Separate multi-table pages if possible. When one page contains two unrelated tables, cropping them into separate images before processing often yields cleaner output.

Step 4: Run the Extraction

With the right preset selected and your file ready, the extraction itself is straightforward:

1. Upload your scanned PDF or image file.
2. Select the appropriate preset (scanned PDF, image, invoice, bank statement — whatever matches your document type).
3. Review the column mapping preview before confirming the export.
4. Download the resulting Excel or CSV file.

For common document types like invoices or delivery notes, there are dedicated presets that go further than raw extraction — they also label columns correctly and handle repeated line items. The invoice to Excel preset and the delivery note preset are good examples: they're pre-configured to recognise the typical structure of those documents so you don't have to map fields manually.

Step 5: Clean and Verify the Output

Even with excellent OCR, a final review pass is essential — especially for archive documents that may be faded, stamped, or printed in non-standard fonts.

• Check numeric columns for transposed digits (e.g., 8 read as 6, or 1 read as l).
• Verify totals and subtotals against the source document.
• Look for merged cells where two columns should be separate.
• Confirm that headers have been correctly identified and are not treated as data rows.

Once you're satisfied with the output, you can use Tablola's built-in AI table editor to make further edits — renaming columns, filtering rows, reformatting dates — directly within the spreadsheet view, without needing to open a separate application.

Step 6: Scale It Up — Batch Processing for Archive Projects

If you're digitising a large archive rather than a single document, processing files one by one quickly becomes its own bottleneck. Batch processing lets you feed multiple scanned PDFs in a single run and receive a unified, consolidated spreadsheet as output.

The merge multiple documents into one table preset is designed for exactly this: it extracts data from each file individually and then stacks all the results into a single, consistent Excel table — ready for analysis, reporting, or import into another system.

Tip: When processing large batches of historical documents, start with a small pilot — five to ten files — and validate the output before committing the full archive. Different scan qualities, fonts, or table formats across older documents can require small adjustments to your preset configuration, and it's far easier to catch that early than after processing five hundred files.