Indian Bank Statement PDF → Excel, CSV & Tally XML

Turn any Indian bank statement PDF into clean, balance-verified transactions in seconds — ready for Excel, Google Sheets, your accounting software or direct import into TallyPrime / Tally ERP 9.

Built for chartered accountants, bookkeepers, loan/DSA agents, lenders, tax preparers and developers who are tired of re-typing statements or fixing broken copy-paste.

Why this one

• ✅ Top Indian banks recognised automatically: SBI, HDFC, ICICI, Axis, Kotak, PNB, Bank of Baroda, Canara, Union Bank, IDFC FIRST — plus a generic parser for co-operative, regional rural and other banks.
• ✅ Running-balance check on every row: previous balance − debit + credit must equal the printed balance. Rows that don't add up are flagged, so you know exactly what to review.
• ✅ Confidence score per file (0–1) — feed it into your automation and only send low-confidence files for human review.
• ✅ Password-protected PDFs (the usual bank e-statement passwords) — give the password once per run.
• ✅ Scanned / photocopied statements via built-in OCR.
• ✅ Tally XML: one Payment/Receipt voucher per transaction against your bank ledger and a counter ledger (default Suspense) for re-classification.
• ✅ Multi-line narrations, Dr/Cr suffixes, overdraft balances, newest-first statements handled.
• 🔒 No retention: PDFs are processed in memory and never stored by the Actor. See Privacy below.

How to use

1. Upload one or more statement PDFs (or paste HTTPS links).
2. If they're password-protected, enter the password.
3. Click Start. Download results from the Output tab:
   • the dataset (one row per transaction) as Excel, CSV, JSON or via API;
   • file-N.xlsx, file-N.csv, file-N-tally.xml and a SUMMARY record in the key-value store.

Sample output (one dataset item)

{
  "file": "file-1",
  "bank": "HDFC",
  "date": "2026-04-03",
  "valueDate": "2026-04-03",
  "narration": "UPI/578381399028/DUMMY TELECOM/PAYMENT/okaxis/ICIC",
  "ref": "578381399028",
  "debit": 1202.0,
  "credit": null,
  "balance": 234149.67,
  "balanceCheck": "ok",
  "confidence": 1.0,
  "page": 1,
  "flags": []
}

SUMMARY (per file): bank, parser used, pages, OCR pages, transaction count, opening/closing balance, total debits/credits, balance-continuity pass rate, whether opening + credits − debits = closing, confidence and warnings.

Input

Field	Description
Statement PDFs	Upload PDFs (text or scanned)
PDF URLs	Or HTTPS links (API / automation)
PDF password	Optional, stored encrypted, never logged
Bank	Auto-detect (default) or force one
Extra output files	Excel, CSV, Tally XML
OCR scanned pages	On by default
Tally bank ledger / counter ledger / company	Ledger names used in the XML

Tally import

Gateway of Tally → Import → Vouchers → choose file-N-tally.xml. The bank ledger and counter ledger named in the input must already exist in the company. Every voucher is balanced (debit = credit).

Pricing

Pay only for transactions extracted — $1.00 per 1,000 transactions (that's about 2–3 cents for a typical 1-page statement). Files that can't be parsed produce no rows and cost nothing beyond the small run start fee.

Privacy & data retention

• The PDF is downloaded into the Actor's memory, parsed, and discarded. It is never written to the dataset, key-value store or logs.
• Logs contain only counts, bank name and confidence — never names, narrations, account numbers or amounts.
• The password field is marked secret (encrypted by Apify) and is not logged.
• Files uploaded through the input form are deleted from your storage after the run where the platform permits (Delete uploaded PDFs option).
• The results (dataset and output files) live in your Apify account under Apify's normal storage retention; delete them whenever you like. Download and delete runs if you process client data.
• Use only statements you are authorised to process.

Limitations

• Statement layouts change; if a bank changes its format the generic parser takes over and the confidence score will tell you. Report a layout and it will be added.
• OCR accuracy depends on scan quality. Always check rows flagged mismatch.
• Credit-card statements and passbook photos are not supported yet.

FAQ

Is my data used for training? No. Nothing is retained. Which languages? English-language statements (all major Indian banks issue English e-statements). Can I call it from Zapier/Make/n8n or my code? Yes — use the Apify API or integrations with pdfUrls.