Guide · 6 min read

How to Check a Converted Statement Before You Use It

· PDFeyes Guide

Any tool that reads a PDF is inferring structure that isn't in the file. Most of the time it infers correctly. The checks below take about two minutes and catch the cases where it didn't, ordered by how much each one catches.

1. Does it reconcile?

If the statement has a running balance, every transaction amount should equal the difference between consecutive balances. This single check catches misread digits, wrong signs, missing rows, and phantom rows all at once, because any of those breaks the arithmetic.

A good tool runs this automatically and tells you how many rows reconciled. If yours doesn't, spot-check five rows by hand: take a row's balance, subtract the previous row's balance, and confirm the difference matches the amount.

Where a break is real, expect it to flag two rows. The broken one, and the one after it, whose delta is now measured from a wrong starting point.

2. Is the row count right?

Count the transactions on the last page of the PDF and compare. Two failure modes show up here: rows dropped at a page break, and non-transactions counted as transactions. Daily balance snapshots and category headings are the usual culprits for the latter. See the rows that aren't transactions.

3. Are the dates right?

The most dangerous failure, because it's invisible. 03/04/2026 is March 4th in the US and 4th March in the UK, and a tool that picks wrong shifts your data by months with no error message.

Find a transaction you remember and check its date. Pay particular attention to statements where every date has both numbers under 13. Those are ambiguous, and a careful tool will tell you it's guessing.

Also check statements that print only MM/DD with no year: the year has to come from the period header, and a tool that ignores it may drop those rows entirely.

4. Are the signs right?

Sort by amount and look at both ends. Money out should be negative, money in positive. A payroll deposit showing negative, or a card purchase showing positive, means a column was misread, and every row in that column is likely wrong the same way.

5. Do descriptions look complete?

Long merchant names wrap onto a second line in the PDF. That continuation line belongs to the row above; if it's been dropped, descriptions are truncated, and if it's been treated as its own row, you have phantom entries with no amount. Scan for either.

Then clean up

If you are checking one version of a document against another rather than against its source, compare two PDFs shows which words changed.

Reconciliation runs automatically, and rows that don't add up are highlighted for review.

Try Statement → CSV