Amex · Categorised export
Categorise Amex Statements by Spending Type
American Express statement PDFs go in, a CSV with a Category column for each transaction comes out. Built for the way Amex formats Australian card statements specifically.
Supports PDF files up to 10MB
What comes out
| Description (from statement) | Category |
|---|---|
| BUYMEACOFFEE.COM | Uncategorised |
| DAN MURPHY'S ONLINE | Shopping |
| QANTAS AIRWAYS SYDNEY | Travel |
| AMEX MEMBERSHIP FEE | Fees & Interest |
Real examples from Amex statements used during testing.
What makes Amex statements harder to categorise
Amex statement coverage sits lower than most transaction accounts in testing, and the reason is structural rather than a parsing bug: a credit card is used almost exclusively for discretionary, one-off purchases, restaurants, retailers, and small local businesses that never repeat. A savings account statement is dominated by a handful of recurring, predictable line items like interest and transfers. A credit card statement is closer to the opposite, every second row can be a merchant name that's never appeared before and won't appear again.
That's the honest ceiling of a keyword-based system on this kind of statement. What does move the needle is generic keyword fragments rather than hardcoded merchant names, things like "petro", "pharm", "barber" or "supermarket" as fragments rather than exact chains, since those generalise across whichever specific business happens to be on the statement.
Categories used
Every category comes from a fixed list of keyword and pattern rules, not a model guessing at intent. Anything that doesn't match a rule is left as Uncategorised rather than assigned incorrectly, so you can see exactly what still needs a manual look.
Built without AI
Categorisation runs as static pattern matching in the same request that already parses your PDF. Your Amex statement is never sent to an AI model to work out what your transactions mean, and nothing about your spending is stored afterwards to train one. It's a fixed set of rules, applied the same way every time.
What Our Customers Say
“Banks like Westpac only let you download CSV files for the past 18–24 months. If you need older data, you're stuck downloading PDFs and manually extracting transactions from pages of formatting and bank jargon. That job is a real slog. Your product handled it instantly and gave me clean data.”
Mark
Manufacturing, former HR/Finance Systems Consultant
“As someone preparing tax returns, going through PDF bank statements manually is a real hassle. This tool dumps everything into Excel format instantly. Huge time-saver.”
David
Director, Tax & Accounting Firm
“Great tool for dealing with PDF statements from clients. The data comes out clean and ready to import, which saves a lot of repetitive work. I had one client with statements going back several years, and this handled them all without a hitch. Highly recommend for accountants and bookkeepers.”
Tamara
Bookkeeper, Self-Employed
“Converted several years of bank statements in minutes and saved me a lot of effort. I initially wasn’t sure if it would handle my older statement formats, but once I tried it, it worked well. Would recommend if you’re dealing with PDFs.”
James
Owner, Construction Company
Questions
Does this use AI to categorise my Amex statement?
No. Categories are assigned with static keyword and pattern matching, not a language model. Nothing in your statement is sent to an AI provider, and nothing is stored to improve a model over time.
How accurate is the categorisation?
On real Amex statements tested during development, coverage sits around ~79% of transactions, depending on account type and how many one-off, unpredictable merchant names appear. Anything not matched is labelled Uncategorised rather than guessed at.
Can I edit the categories afterwards?
Yes. The CSV is a normal spreadsheet column, open it in Excel or Google Sheets and adjust anything that isn't quite right before you file it or hand it to your accountant.