Budgeting

Categorisation coverage: how much of your spend is labelled?

Coverage is categorised rupees divided by total rupees. See why low coverage makes reports mislead and which uncategorised rows to label first for the biggest gain.

Editorial illustration for Categorisation coverage: how much of your spend is labelled?

Categorisation coverage is the rupees you have labelled divided by all rupees spent. If coverage is low, every category total in your reports is understated and any chart built on them misleads. As a rule of thumb, not an official figure, treat reports below roughly 90% coverage with caution, and label the largest uncategorised items first, since a few big rows do most of the repair.

How do you measure coverage?

Measure it in rupees, not in counts. Coverage = (total spending minus uncategorised spending) ÷ total spending. In a sheet with a label for every row, using the word Uncategorised where nothing fits, and with categories in column B and amounts in column C, the figure is =1-SUMIFS($C$2:$C$2000, $B$2:$B$2000, "Uncategorised")/SUM($C$2:$C$2000). Blank cells will not match the word, so fill them first.

Why do rupees matter more than counts?

Thirty unlabelled transactions out of 300 look like 90% coverage by count. But suppose those thirty include a large transfer, an EMI debit and a few shopping payments, and add up to ₹26,400 of a ₹1,20,000 month. Then rupee coverage is ₹93,600 ÷ ₹1,20,000, only 78%. A report that says you spent little on shopping may simply have missed it.

Which transactions should you label first?

Sort the uncategorised rows by amount and start at the top. Suppose the ten largest of those thirty add up to ₹19,800.

StageUncategorised rupeesCategorised rupeesCoverage
Before₹26,400₹93,60078.0%
After labelling the ten largest₹6,600₹1,13,40094.5%

Ten rows, a third of the unlabelled items, lifted coverage from 78.0% to 94.5% (₹1,13,400 ÷ ₹1,20,000). The remaining twenty rows hold only ₹6,600, about ₹330 each on average, so finishing them gains little compared with the first ten.

How often should you check coverage?

Once a month, right after you import or enter the month's transactions. Coverage drops whenever a new merchant appears, a trip adds unfamiliar names, or a batch is added before you have reviewed it. A monthly check takes a minute and tells you whether the reports you are about to read are trustworthy. If you read your reports only once a quarter, run the check for each month in that quarter first.

What makes a label reliable?

  • Coverage is not accuracy. A transaction labelled Shopping that was really a gift to family counts as covered but still misleads.
  • After labelling, spot-check the five largest rows in each category.
  • Treat transfers and card bill payments as their own category, not as spending, so they stop distorting both coverage and totals.
  • Repeated merchants are the cheapest to fix. Label one and the rest of the same kind follow.
  • Large unfamiliar debits deserve a question to yourself before a label: who was this, and was it for me, the household or a business?

Here is a small example of how a gap misleads. Suppose the ₹26,400 of unlabelled rupees contains ₹9,000 of festival shopping. A report that reads Shopping ₹5,000 would be understating it by almost two-thirds, and you would walk away believing the month was frugal. The error is not in the arithmetic of the report but in what was missing from it.

How does this work in practice?

In SpendixAI, categorisation runs in stages: your own rules first, then your history and similar merchants, and an AI model only if you have switched AI on, for whatever is left. If AI is off, unmatched items are held for review, which makes the uncategorised pile easy to see. Your corrections become history, and you can add merchant aliases so the same shop is recognised next time.

  1. Compute coverage for the last three months.
  2. Sort the unlabelled rows by amount.
  3. Label the largest ten and add a rule for any merchant that repeats.
  4. Recompute coverage and note the figure each month.
SPENDIXAI JOURNAL

Useful ideas start with a clearer picture of your own spending.

Explore your reports