Reconcile bank statement and ledger CSVs
Paste a statement CSV and a ledger CSV, choose the date, amount, and memo columns, and get matched, suggested, and unmatched transactions with fuzzy memo scoring.
About this tool
Bank Statement Reconcile compares a bank-statement CSV with a bookkeeping ledger CSV. It matches transactions by signed amount and date window, then uses a fuzzy token score on the memo/description fields to separate confident matches from suggested matches that need review.
Use it when exported bank transactions and accounting rows use slightly different
memos such as COFFEE SHOP versus Coffee Shop downtown, or when ledger entries
land a few days away from the bank posting date.
Worked example
Statement CSV:
date,amount,memo
2024-01-02,-42.50,COFFEE SHOP
2024-01-05,-1200.00,RENT PAYMENT
Ledger CSV:
date,amount,memo
2024-01-02,-42.50,Coffee Shop downtown
2024-01-05,-1200.00,January rent
With a three-day date tolerance and a moderate memo threshold, the tool lists the coffee row as a match and the rent row as a lower-scoring suggested match for human review.
Limits & edge cases
- Each CSV is capped at 1000 data rows to keep the O(statement × ledger) matching pass bounded.
- Dates can be
YYYY-MM-DD,YYYY/MM/DD, or common slash/dot/dash dates with an optional time part ignored. - Amounts may include currency symbols, thousands separators, signs, or
accounting parentheses like
(1,234.56). - Matching is one-to-one and greedy in statement row order: each ledger row can be claimed once.
- Memo similarity is token-based. It is useful for descriptions, not a guarantee that two transactions are semantically identical.
FAQ
Do amounts need to be signed the same way on both files?
Yes. The matcher compares signed amounts. If the statement exports debits as negative values but your ledger stores expenses as positive values, normalize one file first or use a CSV calculation tool before reconciling.
What is the difference between matched and suggested matches?
Both pass the amount and date tolerance gates. A matched row also reaches the memo similarity threshold. A suggested match has the right amount and date but a weaker memo score, so it is listed separately for manual review.
Can I use column numbers instead of header names?
Yes. Every date, amount, and memo column field accepts a header name or a 1-based
index such as 1, 2, or 3. Header names are usually clearer for saved URLs
and CLI commands.
Why did one obvious match remain unmatched?
A ledger row can only be claimed once. If duplicate amounts and dates exist, the greedy pass picks the best memo score in statement order. Tighten the date window, review the suggested matches, or split duplicate-heavy periods into smaller batches.
Developer & Automation Access
Run it from the terminal
Same engine as this page, headless — via the gizza CLI:
gizza tool bank-statement-reconcile "date,amount,memo
2024-01-02,-42.50,COFFEE SHOP
2024-01-05,-1200.00,RENT PAYMENT" 'ledger_csv=date,amount,memo
2024-01-02,-42.50,Coffee Shop downtown
2024-01-05,-1200.00,January rent'New to the CLI? Get gizza →
Open it by URL
Pre-fill and auto-run this tool with query parameters — the names match the API/CLI:
https://gizza.ai/tools/bank-statement-reconcile/?statement_csv=date%2Camount%2Cmemo%0A2024-01-02%2C-42.50%2CCOFFEE%20SHOP%0A2024-01-05%2C-1200.00%2CRENT%20PAYMENT&ledger_csv=date%2Camount%2Cmemo%0A2024-01-02%2C-42.50%2CCoffee%20Shop%20downtown%0A2024-01-05%2C-1200.00%2CJanuary%20rent&stmt_date=date&stmt_amount=amount&stmt_memo=memo&ledger_date=date&ledger_amount=amount&ledger_memo=memo&date_tolerance_days=3&amount_tolerance=0.01&memo_threshold=70&delimiter=comma&output=markdownMachine-readable descriptor: tool.json — title + parameters JSON Schema for agents.