Import the last twelve months of transactions, summarise older years as monthly totals in a separate file, and archive everything else as a read-only copy. Twelve months is enough to compare this year with last year, and it keeps messy old data out of your new records. Go further back only if you need transaction-level detail from earlier years.
What are the three layers?
| Age of data | What to do | Why |
|---|---|---|
| Last 12 months | Import as transactions | Lets you compare this month with the same month a year ago, and spot seasonal bills |
| 1 to 3 years back | Keep monthly totals by category in a summary file | Good enough for trends, without importing every row |
| Older than that | Archive the full sheet, read-only | Reference only, in case you need to find one old payment |
Why not import everything?
Because old data is usually messier than recent data. Categories changed over the years, cash entries were skipped, and formulas turned some amounts into text. Every old row you bring in carries those faults into the new records, and reports are only as clean as the worst year. It is quicker to import a clean year and have a trustworthy base than to spend a weekend fixing 2019.
How do you make the monthly summary?
From the old sheet, build a small table of month by category. If dates are in column A, amounts in C and categories in B, the cell for one category and one month is:
=SUMIFS($C$2:$C$5000,$B$2:$B$5000,$G2,$A$2:$A$5000,">="&H$1,$A$2:$A$5000,"<"&EDATE(H$1,1))
Here G2 holds the category name and H1 the first day of a month. Copy it across and down. Paste the result as values into its own file called something like 'Monthly totals FY 2023-24 to FY 2025-26'. A pivot table with month in columns and category in rows gives the same answer without formulas.
When is it worth importing more?
- You want a full year-on-year comparison for a category, such as school fees, that spans several years.
- The old sheet is already clean, with one row per transaction and real dates.
- You are preparing for a loan, a visa or a dispute where a long, itemised history helps.
- You want to search for a one-off item in old data without opening an archive.
If none of these applies, resist the urge.
How do you decide for your own household?
Ask what question you will actually put to the data. If it is 'how much did we spend on this last year', twelve months is enough. If it is 'are school fees rising faster than everything else', you need totals for several years, and the monthly summary will do. If it is 'did we pay this bill in 2022', you need the full archive, not the app. Match the layer to the question, and you will import less and use it more.
How long should you keep the archive?
For anything linked to tax or legal matters, ask your CA how long to retain records, because the right period depends on your situation and may change. For everyday spending, keep the archive as long as it costs you nothing to hold. Save it as a file with the date in its name, keep a copy in two places, and set it to read-only so an accidental edit cannot change history.
Importing the twelve months well
Use bank and card statements for the months you want rather than exporting the sheet, so dates and amounts match what the bank recorded. SpendixAI supports batch uploads, keeps an import history with status, flags possible duplicate imports and sends a notification if an import fails, so you can load a year in pieces and see what happened. If you make a mess, deleted transactions go to a recycle bin for 30 days.
Useful ideas start with a clearer picture of your own spending.
Explore your reports


