An import-ready sheet has one row per transaction, a single header row, and these columns: date, amount, description, category and account. Remove merged cells, subtotal rows and anything else that is not a transaction, make sure the amounts are real numbers, then save as CSV. Most import trouble comes from three things: merged cells, subtotal rows and numbers stored as text.
What should the clean layout look like?
| Column | Example | Rule |
|---|---|---|
| Date | 2026-10-14 | Real date values, one format all the way down |
| Amount | -1250.00 | A number only, no ₹ sign, no comma inside the cell |
| Description | Local grocery store | Text, one line, trimmed |
| Category | Groceries | One consistent name per category |
| Account | Savings account | The account the money moved from or to |
Use one sign convention: spending as negative, or a separate Type column. If your sheet has separate Debit and Credit columns, combine them with =D2-C2 (credit minus debit, so spending comes out negative) in a new column and use that for amount. A statement-style layout is the shape most importers expect. SpendixAI accepts CSV or Excel files with rows reviewed before saving, but do not assume any layout will be read perfectly; check the preview.
How do you remove merged cells and subtotal rows?
- Merged cells: select the area, choose Unmerge, then fill the blank cells. In Excel, Home, Merge and Center toggles it; in Google Sheets it is Format, Merge cells, Unmerge.
- Month headers in the middle of the data: delete those rows. They are labels, not transactions.
- Subtotal and total rows: filter the Description column for words like Total, select the visible rows and delete them. A subtotal left in will be read as a very large expense.
- Blank rows: sort the sheet by date to push them to the bottom, then delete them.
- Repeated header rows copied from old pages: filter and delete them in the same way.
How do you fix amounts stored as text?
Text numbers are left-aligned and often come with a green triangle in Excel. Count them with =COUNTA(B2:B500)-COUNT(B2:B500); a result above zero means some amounts are text. To convert:
- If there are ₹ signs or commas, use Find and Replace on the column: replace ₹ with nothing and the comma with nothing.
- If a hidden space remains, in a new column use =VALUE(SUBSTITUTE(B2,CHAR(160),"")).
- Alternatively type 1 in a blank cell, copy it, select the amounts and use Paste special, Multiply.
- Re-run the count formula until it returns 0.
What do you do with extra columns?
Old sheets collect extras: a Notes column, a Who paid tag, a Month label, a running balance. Keep what you will want later, move it to the far right and give it a plain heading. Delete columns that are formulas about other columns, such as Month or Running balance, because they can be rebuilt. Add one column called Old row, with =ROW(), pasted as values. If an imported row looks wrong, that number takes you straight back to its source row in the sheet.
Dates, and where to look for them
Check the date column the same way: =COUNTA(A2:A500)-COUNT(A2:A500) should be 0. If dates are in dd/mm/yyyy and the other end reads them as mm/dd, a 3 April entry can become 4 March, so test it with a few rows first. Where in doubt, format the date column as yyyy-mm-dd before exporting, which leaves no ambiguity.
Export, then look at the file
- Excel: File, Save As, and choose CSV UTF-8 (Comma delimited). Google Sheets: File, Download, Comma Separated Values.
- Open the CSV in Notepad. You should see commas between fields and no stray quotes or odd characters in names.
- Pick the account before you import, so every transaction keeps it. SpendixAI also flags possible duplicates, which helps if a month overlaps with a statement you already loaded.
- Spot-check the first and last ten rows after import, and compare the total with the sheet.
Useful ideas start with a clearer picture of your own spending.
Explore your reports


