Almost everyone who decides to track their expenses does it properly for nine or ten days. Every rupee logged, every category filled in, a small sense of virtue. Then a busy Thursday happens, two days go unrecorded, and the gap itself becomes the reason to stop — because now the numbers are wrong and fixing them is a chore.
The habit does not fail because people are lazy. It fails because the method most people invent for themselves has four specific weaknesses. This guide is about designing around each one, and about what the first month of data is actually for.
Record at the moment of payment, not at night
The single biggest predictor of whether expense tracking survives is when it happens. Logging at the point of payment takes about eight seconds and requires no memory at all. Logging at the end of the day means reconstructing seven or eight transactions from recollection, which takes ten minutes and gets two of them wrong.
Those ten minutes are the thing that eventually gets skipped. Nobody skips eight seconds. So the rule is: while you are still standing at the counter, or while the payment confirmation is still on screen, the entry goes in.
This is also why an app on your phone beats a spreadsheet for this particular job, even though a spreadsheet is better at almost everything else. The spreadsheet is on your laptop and your laptop is at home. A daily expense tracker that opens in two taps is available at the exact moment the expense exists.
Track everything for one month, then stop tracking everything
There is a real argument for logging every single transaction, and a real argument against it. Both are correct, at different times.
For the first month, log everything. Every chai, every rickshaw, every top-up. You are not budgeting yet; you are taking a measurement, and a measurement with holes in it is not worth taking. The small purchases are exactly the ones that surprise people, because a hundred rupees five times a day is fifteen thousand a month and nobody has ever felt that number happen.
After that first month, you are allowed to simplify. Most people find three or four categories carry seventy per cent of their variable spending. Keep logging those in detail and let the rest fall into one broad category. You have already learned what you needed to learn from the long tail.
Categories: fewer than you think
A new tracker with twenty-two categories is a tracker where every entry now involves a decision. Was the supermarket run groceries or household? Was the bus fare transport or work? Those decisions are small individually and exhausting in aggregate, and they are the second reason people quit.
Start with six to eight. Something like this covers most households:
- Groceries — food bought to cook or keep at home
- Eating out — anything you did not prepare yourself, including delivery and chai
- Transport — fuel, fares, ride-hailing, maintenance
- Bills and utilities — electricity, gas, internet, mobile, rent if you pay it
- Health — medicines, consultations, tests
- Household — cleaning, repairs, things for the house
- Personal — clothes, haircuts, subscriptions, gifts
- Other — the honest home for everything that does not fit
Add a category only when "Other" gets big enough to be annoying, and you will end up with a category list shaped by your actual life rather than by a template.
One rule worth keeping: a category should answer a question you would act on. "Eating out" is useful because seeing the number might change something. "Wednesday purchases" is not.
Cash is the part everyone gets wrong
If most of your spending goes through a card or a wallet app, you have a statement to fall back on when you forget something. If most of it is cash, you have nothing. The memory is the only record, and it is gone by the next morning.
This makes cash tracking both more important and more fragile. Two things help. The first is an end-of-day pocket check: count what is in your wallet, compare it against what you logged, and if there is a gap of four hundred rupees you will usually remember what it was within about ten seconds of looking. The second is photographing receipts you cannot log right away, so the record exists even if the entry does not yet.
There is more on this in our guide to tracking cash expenses without a bank account.
Expect to miss days, and plan for it
The fourth failure is the one that finishes most people off: a gap appears, the data feels ruined, and the whole thing gets abandoned rather than repaired. This is worth naming in advance because it is going to happen and the response is a decision you can make now.
The correct response is to log an approximate entry and move on. If Tuesday is missing and you know you spent roughly twelve hundred on groceries, put twelve hundred in. A budget built on nine hundred honest entries and one estimate is accurate enough to make decisions from. A budget abandoned on day ten is worth nothing at all.
What the first month is for
A month of data is not a report card. It is the baseline you need before a budget can mean anything, and it usually contains two or three genuine surprises.
Open the category breakdown at the end of the month and look for three things. Which category is largest — often not the one you would have guessed. Which category is largest relative to what you expected — that gap is where the money is leaking. And which single transaction was the biggest, because one-off large purchases distort a monthly average badly and deserve to be thought about separately from the daily pattern.
Only then set limits. Setting a grocery budget before you know what groceries cost you is just guessing, and a guessed limit gets missed in week two and ignored by week three. Our monthly budget planner guide covers what to do with the baseline once you have it, and the 50/30/20 rule is a reasonable starting shape if you want one.
Making it faster
Once the habit exists, everything that shaves seconds off an entry makes it more durable. Casshi's free plan covers the whole method above — unlimited transactions, unlimited categories, reports over any date range — and the paid tiers exist mostly to make the logging itself take less effort:
- Receipt scanning turns a photo of a till slip into a filled-in entry, which is faster than typing and much faster than remembering.
- Voice entry lets you say "eight hundred rupees, groceries" while you are still walking to the car.
- Auto-categorisation removes the one decision that every manual entry still requires.
None of them are necessary. People tracked expenses in notebooks for a hundred years. They just move the cost of an entry from eight seconds to two, and over three hundred entries a month that difference is what decides whether you are still doing this in March.