PDF statements trip up budgeting apps for two reasons that stack on top of each other: a PDF has no real rows or columns, and European number and date conventions give a reading tool many ways to be wrong without looking wrong. Together they explain why an import can seem perfect and still be off by a factor of ten or by a month.
What does the app lose when it reads a printed page?
It loses structure. The statement looks like a table, but the file only records that certain words sit at certain positions. The reader rebuilds columns by alignment. If the alignment is regular it succeeds; if a description wraps or a column is narrower than usual, it shifts a number into the wrong place.
Why are decimal commas such a problem?
Because the same digits mean different amounts under different rules. Look at the table below, where each value is how a statement might print it and how a tool using the wrong convention could read it.
| Printed | Intended meaning | Wrong reading |
|---|---|---|
| 1.250,00 | one thousand two hundred fifty euros | 1.25, a little over one euro |
| 12,50 | twelve euros fifty | 1250 if the comma is treated as a thousands mark |
| 3.000 | three thousand euros, or three euros with three decimals? | Depends on the convention |
| -45,99 | a 45.99 euro payment | Text, so ignored in totals |
The dangerous cases are the ones that look plausible. A €12,50 coffee read as 1,250 would be spotted instantly. A €1.250,00 rent read as €1.25 might sit unnoticed among many small lines.
Do day-first dates and accents add more risk?
- A date like 05.03.2026 is 5 March to you and 3 May to a tool that expects month-first, and nothing flags it.
- Lines with accented letters in merchant names can be recognised as odd characters, so the same shop appears under several names.
- A statement in a language the tool did not expect may confuse the labels it uses to find the amount column.
- Thousands marks can be a full stop, a space or an apostrophe, depending on the bank.
How can I check an import is trustworthy?
- Choose a line with a day above 12 and confirm it landed on the right date.
- Find the largest amount in the file and compare it with the statement.
- Add up the imported amounts and compare with the movement on the statement.
- Check the count of rows against the count of transactions you can see.
- Spot check one line near each page break.
A worked check: a statement opens at €3,120.50, credits €2,800.00 and debits €2,945.35. Then 3120.50 + 2800.00 - 2945.35 = 2,975.15 is the closing balance. If the app's total of credits and debits gives 2,975.15 you are fine; if it gives something off by a round factor such as ten or a hundred, a decimal mark was read wrongly.
What is the simpler route?
Where a CSV is offered, use it, and read the delimiter and decimal convention in a text editor before importing; this is easier to verify than a printed page. Where only a PDF exists, prefer a tool that shows you each row before saving. SpendixAI is built around PDF, CSV and Excel statement import with review before saving, without promising every layout will parse perfectly. It is built around India and INR, so it is relevant mainly to readers who also keep Indian accounts, not euro ones.
Short list for any PDF import
- Know your decimal and thousands marks.
- Know your date order.
- Reconcile to the closing balance.
- Keep the original PDF.
- Fix, do not ignore, odd merchant names.
Useful ideas start with a clearer picture of your own spending.
Explore your reports

