Bank Statement Extraction QA Checklist

Bank statement extraction QA is the difference between a useful export and a spreadsheet that only looks right. Before extracted rows are trusted, the team should prove that pages, dates, amounts, descriptions, and balances survived extraction.

Quick answer

A bank statement extraction QA checklist should verify page coverage, row count, date accuracy, amount sign accuracy, balance logic, missing rows, duplicate rows, OCR confidence, and export format readiness.

Who this is for

Use this checklist when you convert bank statements with OCR, AI extraction, manual data entry, or any PDF-to-table workflow.

The operating problem

Extraction errors often hide in plain sight. A table can look clean while one page is missing, deposits are reversed, or balance rows are mixed with transaction rows. QA needs to catch extraction defects before bookkeeping review.

The workflow

QA the extraction before categorization. Otherwise the reviewer may spend time diagnosing data quality instead of bookkeeping judgment.

1. Compare statement page count to extracted page coverage.
2. Check the first and last transaction dates against the statement.
3. Sample rows from the beginning, middle, and end of the statement.
4. Tie total deposits and withdrawals to statement summary where available.
5. Check running balances if the statement includes them.
6. Flag ambiguous OCR fields, merged rows, split rows, and unreadable descriptions.
7. Export only after the extraction passes row-level and total-level checks.

Checklist

• All statement pages were processed.
• No header, footer, or balance summary row was imported as a transaction.
• Dates are parsed correctly.
• Amounts keep the correct sign.
• Descriptions map to the right row.
• Deposits and withdrawals are separated or signed consistently.
• Duplicate rows are removed.
• Known uncertainty is visible to the reviewer.

What to document before handoff

• Extraction tool or method.
• Page count checked.
• Row count.
• Sample rows reviewed.
• Known OCR issues.
• Totals checked.
• Export format.

Review signals that matter

• Reviewers trust the extracted table as source-adjacent evidence.
• Reconciliation failures are not caused by extraction defects.
• OCR issues are flagged before import.
• The same bank format becomes easier to process over time.
• QA notes are visible to whoever imports or reviews the file.

Where Wesley fits

Wesley keeps statement extraction close to review. Start with PDF to CSV or scanned bank statement to CSV when the first risk is extraction quality, then move the clean rows into bookkeeping review.

FAQ

Is visual spot checking enough?

No. Spot checking helps, but extraction QA should also include page coverage and amount logic checks.

What is the most dangerous extraction error?

A missing page or reversed sign is usually more dangerous than a messy description because it can break reconciliation.

Should extraction QA be done by the preparer or reviewer?

The preparer should complete first-pass QA. The reviewer should inspect exceptions and material risk.