A budget accuracy score is one minus the total size of your category errors, over and under, divided by the total budget. For the made-up five-category month below, the errors add up to ₹5,200 on a ₹24,500 budget, so the error is about 21.2% and the accuracy about 78.8%. The score matters less than what it shows about which lines your budget gets wrong.
Why is the net total not enough?
Compare only total budget with total spending and over-runs in one category can hide behind savings in another. In the example, you budgeted ₹24,500 and spent ₹27,700, a net miss of ₹3,200. But three categories ran over by ₹4,200 together while two came in ₹1,000 under, and ₹4,200 minus ₹1,000 is the ₹3,200. The net figure blurs that. Absolute error treats an under-run and an over-run as equally wrong, because each means the budget did not describe your life.
How do you compute it?
| Category | Budget | Actual | Absolute error | Error as share of budget |
|---|---|---|---|---|
| Groceries | ₹9,000 | ₹10,200 | ₹1,200 | 13.3% |
| Eating out | ₹4,000 | ₹6,100 | ₹2,100 | 52.5% |
| Transport | ₹3,500 | ₹3,300 | ₹200 | 5.7% |
| Utilities | ₹3,000 | ₹3,900 | ₹900 | 30.0% |
| Shopping | ₹5,000 | ₹4,200 | ₹800 | 16.0% |
| Total | ₹24,500 | ₹27,700 | ₹5,200 | 21.2% |
The steps: find each category's error as ABS(actual minus budget), add the errors, and divide by the total budget. ₹5,200 ÷ ₹24,500 is about 21.2%, so accuracy is 100% minus 21.2%, about 78.8%. In a sheet with budget in B2:B6 and actual in C2:C6: =1-SUMPRODUCT(ABS(C2:C6-B2:B6))/SUM(B2:B6).
There is a simpler variant: average the five per-category percentages, (13.3 + 52.5 + 5.7 + 30.0 + 16.0) ÷ 5, which gives about 23.5%. It treats a ₹3,000 category and a ₹9,000 category equally, so the weighted version above reflects rupees better. Pick one and keep it.
What should you do with the score?
The per-line errors are more useful than the headline. Eating out missed by 52.5% and utilities by 30%, so those two budgets are fiction, not discipline problems. Transport missed by 5.7%, so it is fine as it is. Re-baseline lines with big errors by using the average of the last three months rather than a number you wish were true, then watch whether the error shrinks.
- Error above your own comfort line in three months in a row: the baseline is wrong, so reset it.
- Error only in one month: probably a one-off, so leave the budget and note the reason.
- Persistent under-runs: the budget is too generous, so the extra money can go to a goal.
A worked reset: suppose eating out came to ₹5,400, ₹5,900 and ₹6,100 in the last three months. The total is ₹17,400, so the new baseline is ₹17,400 ÷ 3 = ₹5,800. If next month it is ₹6,100, the error is ₹300, about 5.2% of the new baseline, down from 52.5% when the budget was ₹4,000. Same spending, far better description of it.
Is 100% accuracy the target?
No. A perfect score would mean your budget predicted every line to the rupee, which real months rarely allow. Use the score as a trend: if it moves from 70% to 85% over a few months, your budget is describing your life better. The point of budgeting is not to hit numbers but to make decisions before the money is spent.
Monthly routine
- At month end, list budget and actual for each category.
- Compute absolute error per line and the weighted score.
- Pick the two worst lines and decide: new baseline or one-off.
- Carry forward only what you changed on purpose.
SpendixAI has budgets on the Pro plan, so you can set a budget per category and compare it with what your categorised transactions add up to.
Useful ideas start with a clearer picture of your own spending.
Explore your reports


