To import a bank statement CSV into Excel properly, do not double-click it. Open a blank workbook, choose Data > Get Data > From Text/CSV, pick the file and use the preview window to set file origin, delimiter and data types before anything loads. This is where you stop commas in narration from splitting columns and keep reference and cheque numbers as text.
Why not just double-click the file?
Double-clicking makes Excel guess for you. It guesses the separator from your Windows regional setting, turns long reference numbers into numbers, strips leading zeros from cheque numbers and may read 05/04/2026 as 4 May. Once the guesses are saved into a workbook, the original detail is gone. The import preview lets you decide each choice and reload.
What does the preview window ask for?
- Go to Data > Get Data > From File > From Text/CSV (older versions show a From Text/CSV button directly on the Data tab) and select the statement file.
- File Origin: choose 65001: Unicode (UTF-8) if the rupee symbol or a dash looks garbled.
- Delimiter: choose Comma for a normal CSV. Excel can also detect it, but confirm in the preview that the columns look right.
- Data Type Detection: choose Do not detect data types. Everything loads as text and nothing gets rounded.
- Click Transform Data rather than Load, so you can fix types in Power Query first.
What happens to narration that contains commas?
A well-formed CSV wraps such text in quotation marks. Say the file has this line:
- 03-04-2026,"UPI/CAFE COFFEE, MG ROAD/9876",450.00,41550.00
Excel reads the quoted part as one field, so the comma after COFFEE stays inside the narration and you get four columns: date, narration, amount, balance. If the narration does split across columns, open the file in Notepad. A line without quotation marks around the text means the bank wrote a loose file, and you need to merge those columns back with a formula or download a fresh copy.
What should a clean import look like?
Using made-up rows, the line above and one more should land in the sheet like this:
| Date | Narration | Amount | Balance |
|---|---|---|---|
| 03-Apr-2026 | UPI/CAFE COFFEE, MG ROAD/9876 | 450.00 | 41,550.00 |
| 04-Apr-2026 | UPI/GROCER/1122 | 1,200.00 | 40,350.00 |
When both rows are debits, each balance is the previous balance minus the amount, here 41,550.00 - 1,200.00 = 40,350.00. That small subtraction is a fast test that no column has shifted.
How do you set the right column types?
| Column | Set type to | Reason |
|---|---|---|
| Reference or UTR number | Text | Keeps every digit and leading zero |
| Cheque number | Text | Leading zeros stay, so 000123 does not become 123 |
| Date | Date, using locale | Day-first dates are read correctly |
| Amount and balance | Decimal Number | Lets you sum and sort |
For the date, select the column, choose Transform > Data Type > Using Locale, pick Date as the type and English (India) as the locale. Do the same for amounts if they contain thousands commas. Then choose Home > Close & Load.
Two small habits help here. Rename the query in the side panel to the account name, such as Savings April, so you can tell several statements apart. Add a note in the workbook with the date you downloaded the file and the statement period, which makes any later mismatch easy to trace.
Make the import repeatable
The query remembers its steps. Next month, save the new download over the same file name, open the workbook and choose Data > Refresh All. If you move the file, use Data > Get Data > Data Source Settings and change the path. Keep an untouched copy of every original download, because a CSV is the closest thing to the bank's own record.
If cleaning each month still takes too long, SpendixAI accepts statements as PDF, CSV or Excel, lets you pick the account first, and shows the rows for review before saving. Layouts differ between banks, so check the preview.
Useful ideas start with a clearer picture of your own spending.
Explore your reports


