Reading a statement you cannot sort
A bank statement arrives as a PDF because that is what a bank produces, not because anyone wants to read a year of transactions that way. The moment you need a total — what went out in bank charges, how much interest was paid, what the EMIs came to — the format is working against you.
This tool reads the statement, adds it up, and shows you the year in the categories people actually ask about. What it will not do is pretend to be more certain than the document allows, and the rest of this page is mostly about where that line falls.
The cross-check is the part that matters
Any tool can print totals. The question is whether the rows behind them were read correctly out of the PDF, and there is exactly one way to know: where a statement carries a running balance, closing minus opening must equal credits minus debits.
That check is run, and it is the single most valuable number here. When it balances, the figures came across intact. When it does not, something was missed or misread — and being told that is worth more than any category total, because a category total on rows that did not all arrive is confidently wrong.
The reading itself is shared with PDF to Excel, which is deliberate. Two separate parsers would drift apart, and the reconciliation is the only reason to believe either of them.
Categories are totals to check, not a classification
“Bank charges”, “GST”, “EMI”, “interest” and the rest are found by matching words in the narration, because the narration is the only place a bank says what a transaction was. That works most of the time and fails in ways worth knowing about:
- A payment to a shop called GST Traders looks like a tax payment.
- A transfer with EMI in the reference looks like a loan repayment.
- A UPI handle can put a bank’s name in the middle of a person’s name.
So these are offered as totals to check, not as an accounting classification. They are a fast way to find the transactions worth looking at, and a starting point for a conversation with whoever does your books. Anything built on top of them — a tax return, a loan application — needs a human to have looked at the rows.
What it reads
Statements from the major Indian banks are supported, including the awkward ones whose tables have no ruled lines and whose narration wraps across several rows. Wrapped narration is genuinely hard: a single payee name can be split over three lines, and stitching it back together wrongly is how “Ansi Nanani Hdfc” happens. Where a name is uncertain the tool prefers showing you what was printed over inventing a tidier version.
If your statement is already a spreadsheet rather than a PDF, use the Excel and CSV versioninstead — it does the same analysis without the reading step, which removes an entire class of error.
If you just want the rows
Sometimes the analysis is not what you need and the transactions are. PDF to Excel converts the statement to a workbook you can sort and filter yourself, using the same reader and the same reconciliation check. From there, a pivot table will summarise it however you like without opening Excel.
Your statement
A bank statement is about as sensitive as a document gets, so the handling deserves stating plainly rather than in the abstract. The file is sent over an encrypted connection, read to produce your result, and discarded. It is not written to a database, not kept after your download, not used to train anything, and not shared with anyone. Nothing about its contents is logged. No account is needed. Our privacy policy says the same thing at greater length.