When a statement import counts totals and header rows as transactions, your spending is overstated by whatever those lines add up to. They look like rows in the preview but are not money moving, so spot them and take them out before saving.
Which lines are not transactions?
- A closing-balance or opening-balance line.
- A footer such as Total debits or Total credits.
- Repeated column headings, such as Date, Narration and Amount, at the top of each page.
- A carried-forward or brought-forward line.
- A line quoting the statement period or the account number.
What they share is that they sit in the file but do not represent a payment or receipt on a particular day.
How do I recognise one in the preview?
| Clue | Example | Verdict |
|---|---|---|
| No date | Row reading Total debits | Not a transaction |
| Words in the amount column | Amount shown as the word Debit | Heading row |
| Amount equals a sum of other rows | A figure equal to all debits above it | Total row |
| Narration says balance | Closing balance 43,261 | Balance line |
| Very large amount for the month | One entry bigger than the rest combined | Check it first |
Why leaving them in doubles a total
A total row repeats money that is already in the rows above it. Say a month's debits genuinely add up to ₹52,000, and the file ends with a footer row showing Total debits 52,000. If that footer is saved as a transaction, your spending reads 52,000 + 52,000 = ₹1,04,000. Categories take on the same distortion: the footer may be filed under something like Other, which then looks enormous.
A balance line is worse in another way. A closing balance of ₹43,261 saved as a debit would show up as a huge purchase, and the balance proof would break.
Removing them before you save
- Scroll to the top and the bottom of the list first, since footers and headings cluster there.
- Look for rows with no date or no real narration.
- Remove each non-transaction line in the review step.
- Check the page boundaries if the statement is a PDF.
- Count the rows against the statement and run the balance proof before saving.
What if they are already saved?
Find them by sorting for the largest amounts, or by searching for words such as total, balance, brought forward and narration. Delete them, and remember that in SpendixAI deleted transactions sit in a recycle bin for 30 days. Then recheck your category totals, since the junk may have inflated one.
Do CSV files have the same problem?
Yes. A CSV exported from net banking often has a few lines of text above the table, such as the account holder's name and period, and a summary line at the end. Those are read as rows if the importer cannot tell them apart. Excel files can carry them too.
Do this check for every account, not just the first one you import. A habit that begins with the first file and fades by the fourth is how a stray footer ends up in your year-end totals without anyone noticing.
A quick audit for any category
Totals and headings tend to be filed under a category like Other or Uncategorised, because they have no merchant. Open that category after an import and read what is inside. Anything with a name such as total, balance, opening or closing is a junk row. Anything with no date, or a figure that looks like a month's whole spending, deserves a closer look. If the category contains only small, ordinary payments you do recognise, you are fine. This ten-second look is a useful second net, because it catches what you missed in the preview.
None of this means the import is bad; it just means the review step matters. Two minutes of checking the top and bottom of every file keeps one footer from turning into a false extra month of spending.
Useful ideas start with a clearer picture of your own spending.
Explore your reports

