For most people the best answer is neither extreme: import the last two to four months of statements so you can see one complete pattern, then carry on month by month from there. Starting from today gives you a clean but empty dashboard for weeks. Importing years of history costs real time and rarely changes what you do next. Go further back only if you want a year-on-year comparison.
What do you get from starting fresh?
- Zero cleanup. Every row you see is one you chose to record.
- A habit built on the present, not on an archive you must finish.
- No old merchant names, closed accounts or one-off events cluttering your categories.
The cost is that your first dashboards are thin. One month of data cannot tell you whether a high bill was normal or a one-off. You also cannot see how often a yearly payment, such as an insurance premium, really arrives.
What do you get from importing recent months?
A few months of history gives the tracker something to learn from. In SpendixAI, categorisation follows your own rules first, then your history and similar merchants, so the more you have corrected, the fewer rows are left for later steps. Importing also lets you see a normal month against an unusual one, such as the month a festival or a family trip landed.
How far back should I go?
| Your goal | Months to import | Why |
|---|---|---|
| Just start tracking | 0 to 1 | Keeps the first sitting short |
| See a normal month vs an odd one | 3 | Enough for one repeating pattern |
| Spot quarterly or half-yearly bills | 6 | Catches payments that come twice a year |
| Compare this financial year with the last | 12 or more | Needs full April to March periods on both sides |
| Collect records for your CA | Whatever the CA asks for | Check what your CA actually needs |
If you work in April to March years, remember that a comparison is only fair when both years are complete. A tracker that starts in July has nothing to say about the April to June quarter.
What can go wrong when importing old months?
- Overlapping files. If two statements both contain the same days, the same rows can arrive twice. SpendixAI flags possible duplicate imports, but check the import history yourself.
- Old accounts. Pick the right account before each import, because each transaction keeps its account.
- Different currencies. Choose your base currency first. It cannot be changed once transactions exist.
- Layout surprises. Older statements may be laid out differently. Review rows before saving, and expect a few that do not parse perfectly.
A worked example with made-up figures
Say you spend about ₹14,000 a month on groceries and household items, and pay a ₹12,000 insurance premium once every six months. If you import only one month, the premium either appears or it does not, and the month looks either ordinary or alarming. Over six months you would see it twice: 6 x 14,000 = ₹84,000 on groceries plus 2 x 12,000 = ₹24,000 on premiums, so ₹1,08,000 in total across those two groups, or ₹18,000 a month when averaged. Those numbers are invented, but the point stands: a longer window shows a bill that a short one hides.
What should I do first?
- List your accounts and pick the one or two you use most.
- Download the last three months of statements for each as PDF, CSV or Excel.
- Import the oldest first, review the rows, and save.
- Fix the biggest wrongly categorised lines, then set a rule for the repeating ones.
- Check the import history to make sure every file has a clear status.
When is it fine to skip history altogether?
If your statements are hard to download, if you share an account with someone, or if the old months are tangled with a job change or a move, start from the first day of a new month. A clean boundary is easier to reason about than a messy backfill. You can always import older months later when you have the time.
Useful ideas start with a clearer picture of your own spending.
Explore your reports


