Convert any bank statementPDF to Excel, CSV, or JSON

Bank Statement Parser API is a REST API that converts bank statement PDFs — from any bank, any country, any language, text-based or scanned — into structured JSON or CSV. For every transaction row it returns the date, value date, description, process number, amount and running balance. There is no template to configure: a layout the engine has not seen before is analysed and learned on the first request.

Send a POST request to https://api.bankstatementparser.net/extract with your PDF as multipart/form-data and your API key in the X-Api-Key header. The response is a JSON object with a transactions array plus bankKey, bankName, pageCount, transactionCount and requestId. Each transaction contains date, valueDate, description, processNo, amount, balance and a verified flag.

All of them. There is no supported-banks list and no per-bank setup — the parser reads the structure of the statement itself, so it is not limited to a fixed catalogue. When it meets a layout it has not seen, it analyses the document, builds a parsing profile automatically, and every later statement from that bank is instant. On the rare document it cannot parse, that case is logged and turned into a fix, so coverage only grows.

No. Uploaded PDFs are held in memory only for the few seconds of processing and are deleted immediately afterwards — nothing is written to disk and nothing is kept. Your statement data is never shared with third parties. Transfers are TLS-encrypted end to end.

Both text-based (selectable text) and scanned (image-based) PDFs. Text PDFs go through a fast coordinate-based engine; scanned PDFs are read with our OCR engines and then parsed the same way. You always call the same endpoint — the engine picks the right path.

When a PDF comes from a layout the engine has not handled before, it analyses the line structure, detects transaction rows, date patterns and amount columns, and builds a parsing ruleset on its own. That profile is stored and reused for every future statement in the same layout, with zero manual configuration.

A JSON object with bankKey, bankName, pageCount, transactionCount, requestId and a transactions array. Each transaction has date, valueDate, description, processNo, amount and balance (all as strings, preserving the original formatting), plus a verified flag that is true when the row passes the running-balance check.

{
  "bankKey": "bank_v1",
  "bankName": "Example Bank",
  "pageCount": 3,
  "transactionCount": 47,
  "requestId": "3fa85f64-5717-4562-b3fc-2c963f66afa6",
  "transactions": [
    {
      "date": "2024-01-15",
      "valueDate": "2024-01-15",
      "description": "MARKET PAYMENT",
      "processNo": "2024011500042",
      "amount": "-45.90",
      "balance": "1254.10",
      "verified": true
    }
  ]
}

The free plan gives you 50 pages every month, no credit card and no time limit. Paid plans are Starter $25/mo (500 pages), Professional $65/mo (3,000 pages) and Enterprise $199/mo (15,000 pages); paying yearly saves 20%. One-time page packs — available to Starter plan holders and up — start at $15 for 500 pages and never expire. REST API and API-key access are included from the Starter plan up.

Yes. Scanned and image-based PDFs are fully supported by our OCR engines. The engine tries the fast coordinate-based path first, falls back to OCR for scanned documents, and learns any new layout automatically — all behind a single API call.