If you think a bank statement imported cleanly just because the numbers appear in the right cells, you have already lost. The three checks shared by Juve_M are not a nice-to-have, they are the minimum viable diligence for anyone who moves money data by hand. Opening plus sum of transactions should equal closing. If it does not, a row is missing or a digit is wrong. That single check catches more errors than most people want to admit exist.
The second check is where the real detective work lives. Recompute the running balance yourself, row by row. The first row where your number differs from the statement is where the damage started. This is not guesswork; it is a traceable audit trail that pinpoints the exact transaction that broke the chain. Most people stop at the opening-closing tie-out and assume the rest is fine. That assumption is the quietest source of financial drift. The third check is a trap that catches even experienced users: debits and credits in separate columns. Many PDFs append a sign or "CR/DR" after the number, and text-to-columns turns "1,234.50-" into text. SUM ignores it. Your books balance on paper and your spreadsheet says zero. The bonus on dates is equally sharp, 03/04/2025 is March or April depending on your bank's locale, and unless you check against a date you already know, you will never spot the mismatch.
These checks connect directly to the broader challenge of pulling data across sheets reliably. In Pull sales data across sheets by date range with a smarter spreadsheet formula, we looked at how a single formula can aggregate transactions across time periods without manual copying. That approach depends on clean source data. If your statement import is silently wrong, your sales report is wrong too, no matter how elegant the formula. The same principle applies to budgeting. In Protecting your reserve first before discretionary funds are released, we covered how a reserve threshold protects a group from overspending. That protection evaporates if the underlying transaction data contains a hidden error from a PDF import. A reserve calculation built on bad data is worse than no calculation at all, it gives false confidence.
What Juve_M missed is worth noting. They asked what others check, and the silence in the comments is telling. Few people have a systematic answer. That is the real takeaway here: the most dangerous errors are the ones you do not think to look for. The running-balance recompute is the closest thing to a universal debugger for statement imports, yet almost nobody uses it. The date-locale check is trivial and takes ten seconds, but it never makes anyone's checklist until after a reconciliation disaster. Our advice is to build a two-minute review into every import process, not as a correction step but as a verification step. Treat every imported statement as guilty until proven clean. The first time you catch a missing row or a flipped sign, that two minutes will feel like the best investment you made all quarter.