Question 1

How do I parse PDF bank statements programmatically?

Accepted Answer

POST the file to https://statementbear.com/api/v1/documents with your API key as a Bearer token. The response is JSON: each account on the statement with its period and balances, and every transaction with its date, description, payee, amount, direction and category. Any language that can send an HTTP request works; the documentation has samples in cURL, Node.js, Python, Java, PHP, Ruby and Go.

Question 2

What exactly am I charged for?

Accepted Answer

Pages read successfully, at $0.08 each, so a four-page statement is $0.32. A file that is not a statement, an error on our side, or a call with a test key costs nothing, and a failed read gives its pages back. A photo is one page and every 40 rows of a CSV or OFX export are one page. Every response says what it cost in the X-Billable-Pages header.

Question 3

What do you keep, and for how long?

Accepted Answer

Each result is stored for your retention period, 30 days unless you set anything from 1 to 90 in the console, so you can fetch it again; then it is deleted. DELETE /documents/{id} removes one at once. The original file is not kept after reading, except files uploaded through a collection, which stay for the same period so you can download them.

Question 4

How is this different from Open Banking?

Accepted Answer

Open Banking needs the account holder present, consenting, and banking with a covered institution. Mortgage files, SME lending, tenant screening, immigration and legal disclosure still run on statements handed over by the applicant, often for accounts no aggregator covers. This is for those, and plenty of teams use both.

Question 5

Which banks do you support?

Accepted Answer

Any bank in any country and currency. Nothing is configured per bank and there is no template to maintain, so an issuer we have never seen reads the same way as one we have. A PDF holding several accounts, or ledgers in two currencies, comes back with each account separately.

Question 6

What happens if you misread a statement?

Accepted Answer

Every response carries a verification object: where a statement prints opening and closing balances, the rows have to account for the movement between them to the cent. When they do not, reconciled is false and the issues say where, so you can send that document to a person.

Question 7

Are scanned or photographed statements supported?

Accepted Answer

Yes, at the same $0.08 a page. Send a scanned PDF, or a JPEG or PNG photo, and it is read and checked like a digital PDF. A scan carries the image_only authenticity signal, because it has no text layer to inspect for edits.

Question 8

Can it spot an edited statement?

Accepted Answer

Every document has an authenticity block: whether the PDF was last written by an editor, modified after it was created, has balances that do not reconcile, or was already sent by your account, combined into a low, medium or high risk. They are signals to review, not a verdict.

Question 9

Can I get QuickBooks or OFX files instead of JSON?

Accepted Answer

Yes. Add format=qbo, ofx, qif or csv to a request and the response is a file QuickBooks, Xero or Quicken imports directly, at the same price. Errors still come back as JSON.

Question 10

Is there a sandbox?

Accepted Answer

Yes. A test key calls the same endpoint and returns a fixture document instead of reading your file, so it is free and safe to leave in CI. A header makes it return a multi-account statement, an unreconciled one, or any error you need to handle. For how well we read your own statements, use the 200 free pages. Neither needs a card.

Bank statements in, verified data out

Everything on the statement, as data

Everything in the API

Read

Verify

Integrate

Live in <2 minutes

Pay for the pages you read

Built for the job you are doing

Frequently asked questions

Read your first statement today