Why Metro Bank Statements Are Hard to Convert
banksWritten with AI
Michael
February 1, 2026
5 min

Why Metro Bank Statements Are Hard to Convert

Metro Bankbank statement conversionPDF to CSVUK banksbookkeepingPDF parsingbank statement problems

Metro Bank PDFs store transaction data differently to other UK banks. Here's why they break most converters and how we solved it.


If you've ever tried to convert a Metro Bank statement PDF into a spreadsheet, you've probably noticed something odd. The data comes out wrong. Transactions get jumbled. Amounts appear next to the wrong dates.

This isn't user error. Metro Bank structures their PDFs differently to almost every other UK bank, and most conversion tools can't handle it.

Here's what's actually happening and how we fixed it.

How normal bank PDFs work

When you extract text from a standard bank statement PDF, the data comes out in reading order. Top to bottom, left to right. A transaction looks like this:

  • Date: 1 Jan
  • Description: Coffee Shop
  • Amount: £5.00
  • Balance: £95.00

Simple. The converter reads line by line and everything stays together.

What Metro Bank does differently

Metro Bank PDFs don't store data in reading order. When you extract the text, you get something like this:

  • Line 1: £95 (a balance from transaction 9)
  • Line 2: 10 Jan (a date from transaction 10)
  • Line 3: £200 (a balance from transaction 11)
  • Line 4: Coffee Shop (a description from transaction 5)

The data is scattered. Dates from one transaction sit next to amounts from a completely different one. Standard text extraction produces chaos.

Finding the pattern

After analysing multiple Metro Bank statements, we discovered something useful. Not everything was scrambled.

Dates and descriptions extract in correct chronological order. If you read just those elements from top to bottom, they line up perfectly. 1 Jan, then 2 Jan, then 3 Jan.

Amounts and balances are the problem. They appear in seemingly random positions throughout the extracted text.

This gave us a way in.

The two-part solution

We built a hybrid approach that handles each data type differently.

For dates and descriptions, we read the text stream normally. They come out in order and we capture them sequentially.

For amounts and balances, we ignore the text stream entirely. Instead, we use coordinate mapping. Every element in a PDF has an X and Y position on the page. We find where each date appears vertically, then scan horizontally at that same height to find the corresponding amount.

Think of it like matching puzzle pieces by their position rather than trying to read scrambled text.

The multi-page problem

Our first version only extracted 7 transactions from a statement that should have had 47.

The issue was the summary box on page one. Metro Bank statements include an account summary near the top showing opening and closing balances. Our parser saw "Closing Balance" and assumed the document had ended.

But there were seven more pages of transactions after that summary.

The fix was to make the parser look ahead. Before stopping, it checks the next 50 lines for transaction indicators like column headers. If it finds them, it keeps going. It only stops when there's genuinely no more data.

The page layout problem

Another issue emerged with multi-page statements. Page one has a different layout to subsequent pages. It includes the address block, account information, and summary box, so transactions start lower on the page.

Pages two onwards are pure transaction data starting near the top.

This meant the Y-coordinates didn't match across pages. A date on page two might have the same vertical position as an amount on page one, causing incorrect matches.

The solution was simple but essential: only match elements from the same page. Page two dates only look for amounts within page two coordinates.

The result

The converter now handles Metro Bank statements reliably:

  • All dates and descriptions extract in order
  • Amounts match correctly using coordinate mapping
  • All pages process fully, not just page one
  • Page one's unique layout doesn't cause mismatches

Statements that previously produced unusable output now convert cleanly in seconds.

Why this matters for bookkeepers

If you have clients who bank with Metro, you've probably experienced the frustration of broken exports. Copy-pasting manually works but takes time. Other converters either fail completely or produce errors you have to fix by hand.

Understanding why Metro Bank statements behave differently helps explain why a specialised approach is necessary. It's not that conversion tools are poorly built. It's that Metro Bank's PDF structure requires specific handling that most tools don't provide.

Try it yourself

If you're dealing with Metro Bank statements regularly, you can test the converter free with no signup required. Upload a statement and check the output against the original before committing to anything.