PDF Tools·4 min read

Extracting Tables from PDF: Bank Statements to Excel

A bank statement PDF holds exactly the forty rows you need for a reconciliation, and every copy-paste attempt lands them in one useless column. Extracting tables from pdf is a solved problem once you know which kind of PDF you're holding and what a converter can honestly promise. This guide covers how the PDF to Excel tool on this site reads native and scanned statements, a worked row-by-row example, the merged cell cleanup that follows almost every export, and the mistakes that turn amounts into text.

Open the tool All insights

Why statement PDFs fight the copy-paste into Excel

Rows that look like a table but aren't one.

A PDF has no concept of a table. What looks like rows and columns is hundreds of positioned text fragments, each placed at an x and y coordinate. Copy-paste flattens those coordinates into a stream, which is why your statement lands in Excel as one long column of dates, descriptions, and amounts interleaved.

This bites hardest at month end: reconciliations, expense reports, tax filings, invoice disputes. The data exists, it's just trapped in a layout format. Rebuilding the grid means reading the geometry, which is what a real converter does and your clipboard can't.

Scanned vs native pdf tables: know which one you're holding

One has text, the other has pictures of text.

A native PDF, downloaded from your bank's portal or an accounting system, contains an actual text layer. Try selecting a line: if the cursor grabs words, extraction can read exact characters and positions, and accuracy is essentially perfect. The PDF to Excel tool reads that geometry directly and reconstructs columns from the coordinates.

A scanned statement is a photograph. There's nothing to select, so the tool detects the missing text layer and switches to OCR automatically, rasterizing each page and reading it with the language model you picked. OCR output is very good on clean scans and approximate on skewed or blurry ones, which is why the preview step exists: check the numbers before you trust them.

A worked example: one statement row from PDF into Excel

What the export actually writes.

Take a typical row as it appears in the PDF: 05/06/2026, NEFT SALARY CREDIT, 1,50,000.00. After Analyze & preview, the row shows in the on-page preview split into its columns. Download Excel writes the date and description as text and the amount as a real number: the cell holds 150000 with a thousands-separator format applied, so it displays like the statement but sums correctly.

Headers and letterheads keep their look too. The bank's centered name stays centered and bold, the column header row keeps its ruled borders, amounts stay right aligned, and the header row is frozen so it follows you as you scroll. Each PDF page becomes its own worksheet, named Page 1, Page 2, and onward.

Fixing merged cells after export

Ten minutes of cleanup, done in the right order.

Long descriptions are the usual culprit. When a transaction narrative wraps across two lines in the PDF, an exporter can read it as two rows, and a letterhead line arrives as one merged span across every column. The fixes are quick if you do them in order.

Unmerge first: select all, then unmerge every cell, because sorting and filtering refuse to work while merged spans exist.
Rejoin wrapped descriptions with a helper column that concatenates each row with the dateless continuation line under it, then paste values over the originals.
Convert any leftover text amounts with a multiply-by-one paste special, though exports from this tool already write pure amounts as numbers.
Strip repeated page headers with a filter on the header text, a two-minute job since each page arrived on its own sheet.

Mistakes that corrupt a statement export to Excel

Where the numbers quietly go wrong.

Most bad exports trace back to one of these four moves.

Running OCR in the wrong language. Reading a Hindi statement with the English model produces confident nonsense; the language select exists for exactly this case.
Skipping the preview. The on-page preview is exactly what the workbook will contain, and thirty seconds of scanning it catches a misread total before it reaches your books.
Trusting OCR totals blindly. On scans, verify the closing balance against the PDF; one misread digit in an opening balance poisons every running total after it.
Feeding in a password-locked statement. Remove the password with Unlock PDF first, or the extraction may fail outright.

Tips for extracting tables from pdf statements cleanly

Small choices, much less cleanup.

A few habits upstream of the conversion cut the downstream work in half.

Download the native PDF from your bank portal instead of scanning a printout; native text extraction beats even excellent OCR.
For a 60 page statement, isolate just the months you need with Extract PDF Pages before converting; less input means faster OCR and a smaller cleanup.
Scan at 300 DPI grayscale when you must scan at all; it's plenty for OCR, and 600 DPI color only slows the read.

PDF to Excel or another extractor: picking the right tool

Structured pages here, everything else elsewhere.

This converter is built for structured pages: statements, invoices, price lists. If what you have is a photo of a receipt rather than a document, Image to Text (OCR) pulls plain text faster and skips the table logic. For tabular data that already lives in developer formats, JSON to CSV and CSV to JSON cover the structured side.

And when the PDF just needs reading rather than analysis, PDF Viewer opens it in the browser, while Split PDF trims a giant archive down to the pages actually worth converting.

Try it now

Open PDF to Excel

The tool is one click away. No sign up, no upload, no payment.

Open PDF to Excel

Write a review