How to convert a bank statement PDF to Excel: every method, ranked

A bank statement PDF is a picture of a table. Getting it back into an actual table — rows you can sum, sort, filter and import — is one of the most common jobs in small-business bookkeeping, and there are more ways to do it than most people realise. Here they are, ranked by how often they produce a spreadsheet you can trust, with the failure modes of each. At the end is the one check that works no matter which route you took.

Before anything else: try the bank

Most online banking will hand you transactions as CSV or Excel directly, no PDF involved. If the account is open and the period is recent, that's the answer and you can stop reading.

The catch is the window. Many banks only export the last 12–24 months (some as little as 90 days), closed accounts usually can't export anything, and the download is a transaction list rather than a statement — no declared opening and closing balances, and sometimes different descriptions from the printed version. Everything below exists for when the export isn't there: older years, closed accounts, statements a client emailed you, and paper. (How to get old statements covers the first two.)

1. Copy and paste — free, and worse than it looks

Select the table in your PDF viewer, paste into Excel. On a clean single-page statement it's almost fine. On anything else, the same five things go wrong: columns collapse into a single cell; descriptions that wrap onto a second line become separate rows; negative amounts lose their minus sign or gain a stray "CR"; dates arrive as text; and, depending on your locale, thousands separators get read as decimal points, quietly turning £1,250.00 into 1.25.

Budget ten minutes a page including the fixing, and accept that you'll miss something. Fine for a dozen rows you'll check by eye; a liability at any scale.

2. Excel's built-in PDF import

Microsoft 365 Excel can open a PDF through Power Query: Data → Get Data → From File → From PDF. It detects tables per page, previews them, and loads the ones you pick. Numbers stay numeric, which already beats pasting.

Its limits are structural. Each page becomes its own table, so a twelve-page statement is twelve tables to append, each with the header row repeated. Multi-line descriptions still confuse the detector. And it's a text extractor, not OCR — a scanned or photographed statement yields nothing at all. Windows got the feature first; Excel for Mac followed later and tends to lag.

The best free route for a short, clean, digitally generated statement with a simple column layout. Unreliable the moment the layout isn't.

3. Adobe Acrobat's Export to Excel

Acrobat Pro (and Adobe's online export tool) will convert a PDF to XLSX, and crucially it OCRs scans along the way. The output, though, reproduces the page — merged cells, spacer columns, the bank's page header landing in the middle of your data — rather than a transaction table, so a clean-up pass follows every export. Signs and CR/DR conventions aren't interpreted; a column of unsigned figures with a "Paid out" heading is your problem to fix. Worth it if you already have Acrobat and need OCR. It was never a bookkeeping tool.

4. Google Drive's OCR

Upload a scanned PDF to Google Drive and choose Open with → Google Docs: Google OCRs the text and drops it into a document. The recognition is surprisingly good on clean scans. The result, however, is prose, not a grid — columns come out as run-together paragraphs that you rebuild by hand. A last resort for a page or two of paper, and only if you'll verify every number afterwards.

5. Open-source table extractors

For the technically inclined: Tabula has a point-and-click interface for pulling tables out of PDFs; Camelot and pdfplumber are Python libraries that do the same under your control. On digital PDFs with well-aligned or ruled tables they are precise and fast, and because they're scriptable they scale to hundreds of statements.

The costs: there's no OCR unless you wire in Tesseract yourself, every bank's layout needs its own tuning, and the tuning breaks when the bank redesigns its statement. Excellent if you convert a single bank's statements in volume and can write code. Wrong tool for a mixed client base.

6. A dedicated bank statement converter

Purpose-built converters read a statement as a statement. They know a date column from a balance column, that "CR" means a credit, that a description can wrap, that a scanned page needs OCR before anything else, and that the output should be the file your accounting software wants rather than a facsimile of the page. The good ones do the one thing none of the methods above do: they check their own result, which is the subject of the next section. We compared the main options in the best bank statement converters in 2026.

NoRekey is ours. Drop in a PDF — digital, scanned or photographed, password-protected or not — and download Excel, CSV, OFX, QFX or JSON. Every conversion is verified against the statement's own opening and closing balances before you see it, and the free plan covers ten pages a month with no card.

The check that makes any method trustworthy

Whichever route you took, the statement itself will tell you whether the spreadsheet is right. It declares an opening balance and a closing balance, so:

opening balance + sum of all amounts = closing balance

In Excel, with the opening balance in B1 and signed amounts in column C:

=B1+SUM(C2:C200)

Compare that to the closing balance the statement prints. If they match to the penny, every row is present and correctly signed — that is the definition of a complete extraction. If they don't, the usual suspects in order: a page break swallowed rows; a sign flipped on a refund or fee; a decimal shifted; a repeated header row was parsed as a transaction. If the statement prints a running balance, you can go further and check row by row: each balance should equal the previous balance plus the amount, which pins the error to its exact line.

Do this before the data goes anywhere near your books. Accounting software will import wrong numbers as happily as right ones.

Ranked

Method	Digital PDFs	Scanned	Verifies the result	Cost	Best for
Bank's own CSV export	n/a	n/a	No	Free	Open accounts, recent periods
Copy and paste	Poor	No	No	Free	A dozen rows you'll eyeball
Excel Get Data → From PDF	Fair	No	No	Microsoft 365	Short, clean, single-layout statements
Adobe Acrobat export	Fair	Yes	No	Acrobat Pro	Scans, if you already pay for Acrobat
Google Drive OCR	Poor	Fair	No	Free	One or two paper pages, last resort
Tabula / Camelot / pdfplumber	Good	No	No	Free + your time	One bank, high volume, you can code
Dedicated converter	Good	Yes	Yes (the good ones)	From free / ~1¢ a page	Mixed banks, scans, anything going into the books

Which one, then?

• A few rows from an open account: the bank's export, or paste and check.
• One clean digital statement, no budget: Excel's PDF import, then the balance check.
• Paper or scans: skip the free OCR routes unless you enjoy rebuilding tables — use a converter that does OCR and verification.
• More than a handful of statements, or anything destined for accounting software: a converter. The time you'd spend cleaning up the output of methods 1–5 costs more than a penny a page.

And whatever you choose, finish with the arithmetic. The statement already knows the right answer; the only question is whether your spreadsheet agrees with it.