A step-by-step guide for reviewing bank statements and flagging potential errors.
① Upload your file
Drag one or more CSV or PDF bank statements onto the upload area, or click Browse. You can drop an entire year's statements at once — they'll be processed in alphabetical order, so name them consistently (e.g. 2026-01-Jan.pdf) to ensure the right sequence.
② Review the extracted rows (PDF only)
For PDF uploads, the tool extracts transactions and shows them in an editable table before running any checks. Amber-highlighted rows are the ones it's least certain about — missing dates, amounts, or an unresolved debit/credit sign. Check these first, edit any cell that looks wrong, remove stray rows with ✕, or add missed rows with Add row.
Tip 1 — data doesn't look right? Click "⟳ Re-extract using OCR" in the extraction panel to re-read the PDF as an image rather than as a text file. This is slower and may need more manual correction, but often works better on scanned or photographed statements.
③ Set a materiality threshold (optional)
Enter an amount in the Materiality threshold box to flag every transaction at or above that value. Leave it blank if you don't need this check.
④ Run checks on these rows
Click "Run checks on these rows" (PDF) or upload a CSV to run automatically. The tool checks for:
- Balance chain breaks — any row where the running balance doesn't add up
- Possible duplicates — same date, amount, and payee appearing more than once
- Date gaps — missing statement pages or transactions
- Round-number transactions — exact round figures worth a sense-check
- Weekend-dated transactions — unusual for most business accounts
- Materiality threshold breaches — if a threshold was set
- Statistical outliers — amounts far outside the normal range for a payee
- Benford's Law deviation — informational flag on the leading-digit distribution
⑤ Review the results
Flagged rows appear at the top in red, with every other transaction below them. Use the High / Medium / Low filter buttons to focus on specific severity levels. Each flagged row shows the date, description, amount, balance, and the reason it was flagged.
Tip 2 — long or garbled descriptions? Paste your Google AI Studio API key into the
AI description cleaning panel and click
"Summarise descriptions" to automatically shorten them.
Note: this uses your daily free-tier quota, so only use it when descriptions are genuinely difficult to read — not as a routine step. Get a free key at
aistudio.google.com/apikey.
⑥ Export your data
Three export buttons are available once checks have run:
- Export flagged items (CSV) — flagged rows only, with severity and reason
- Export transactions for Xero — all transactions in Xero import format (*Date, *Amount, Payee, Description, Reference, Check Number)
- Export for WP — all transactions in working papers format (Date, Description, Debit, Credit, Balance)