Automatic expense tracker in India: a question of coverage

Take a household with ₹1,30,000 of take-home pay, a home loan, a salary account at one bank and a credit card from another. Here is one month by route. The figures are illustrative; redo it on one month of your own statement.

The credit card row is spending in the month, 18 swipes. The payment settling last month's card bill also leaves the account but is excluded, because it pays for spending already counted. Counting both is the commonest double count in tracking, and it would push the total from ₹1,22,121 to ₹1,43,721.

A payment app's history sees payments made in that app. Give the main app 55 percent of UPI and three of the five mandates: it records ₹10,071, 8.2 percent of the month, and nothing from cards, EMIs or net banking. The bank's own app sees every debit on that account, but not a card from another bank beyond the lump bill payment: ₹1,00,521, 82.3 percent. Account Aggregator data, the bank ledger shared by consent through an RBI-regulated intermediary with no password, has the same reach across every account you link, and cards only where the issuer shares them.

SMS alerts are the widest single net: banks alert on every route, NACH and AutoPay included, and issuers alert on every swipe. That is all ₹1,22,121. Email depends on the bank; this household's bank emails card and transfer debits but not UPI or NACH, which gives ₹57,000. Read together, with each debit counted once, SMS and email cover all seven routes. Statement upload does too, a month late and only as often as you remember.

The payment app records 56 transactions, 42.4 percent of the count, and 8.2 percent of the rupees. Email records 28 transactions, 21.2 percent of the count, and 46.7 percent of the rupees. By count the payment app is twice as good; by rupees email is more than five times as good. Only the second is what a budget needs.

The costly case is a tracker that sees UPI across every app, the mandates and both cards, but not NACH or transfers. It records 124 of 132 transactions, which feels complete, and ₹47,921 of the money.

Full coverage is coverage of what left the account. Three things sit beyond any automatic source.

Cash after the ATM. The ₹8,000 is recorded to the rupee; what it bought is recorded nowhere. At a steady 6.6 percent, treat it as one line. If it is large or rising, the fix is a note of the two or three big cash payments each month.

One payment, several categories. A ₹2,600 supermarket debit is one row. If it was ₹1,800 of groceries and an ₹800 kettle, the chart says ₹2,600 of groceries until someone splits it.

What a transfer means. The ₹12,000 IMPS to a family member is, in every source, a transfer to a name. It could be support for parents, rent to a relative who owns the flat, a loan that will come back, or a move to your own account elsewhere, which is not an outflow at all. A tracker can remember what you told it last month; it cannot know the first time.

Logging 132 transactions by hand at 20 seconds each is 44 minutes a month, 528 a year, and that is typing, not reading. With entry automatic, the month reads in ten minutes, on a fixed date, in four steps.

Two minutes on the total: ₹1,22,121 out against ₹1,30,000 in. If the tracker's total is far from what the balance says, it is missing a route; find that first.

Three minutes on what next month is already committed to: ₹46,200 of NACH, ₹1,641 of AutoPay and ₹3,500 of maintenance, ₹51,341, or 39.5 percent of take-home before the month starts. ₹1,641 a month is ₹19,692 a year, and a mandate for a service you stopped using charges until someone cancels it.

Three minutes on the unlabelled: new payees, unnamed handles, any debit you do not recognise. This is also the fraud check.

Two minutes on cash and splits. Then stop: 120 minutes a year, against 528 for entry alone.

If most of your money leaves as cash, a tracker gives you a precise record of ATM visits. A notebook for the large cash payments does more.

If your outflows are few and fixed, an EMI, a SIP, rent by transfer and one card cleared monthly, the bank app and card statement already show the month. Another tool adds a permission and little information.

If the problem is that ₹51,341 of ₹1,30,000 is committed before the month starts, tracking will show it clearly and change none of it. That is a question about the loan, the premium and the SIP, answered by reading those contracts.

And on an iPhone with a bank that does not email UPI and NACH debits, an email-reading tracker sees ₹57,000 and misses ₹65,121. A monthly statement upload does better than a partial live feed that looks complete.

Coverage of all seven routes. Walk the rows of the matrix. After a month, its total should land within a few hundred rupees of what left your account; if not, the missing route is usually NACH or transfers.

No bank password, no UPI PIN. Alerts, emails and a consented Account Aggregator feed need neither. The PIN authorises payments; a tool asking for it wants a capability tracking does not need.

Merchant naming. UPI debits carry a handle, often a phone number; NACH debits carry a mandate reference. Ask what the tool shows when it cannot name one: the raw handle is honest, a guessed category is not.

Deduplication. The same ₹999 debit can arrive by SMS and email; a card swipe and the card bill are the same money twice; a transfer between your own accounts is not an outflow.

Android or iPhone. On Android a tracker can read transactional SMS, the widest source. On an iPhone no app can, which leaves email, Account Aggregator data and statements. On Android, read what the SMS permission covers and whether messages leave the phone.