How to track shared household expenses without a spreadsheet

Almost every household that decides to “start tracking expenses” starts with a spreadsheet. It is free, everyone can open it, and for about three weeks it works. Then someone forgets a Saturday, someone else enters a week in the wrong column, and by the second month the file is a record of good intentions rather than of money.

The spreadsheet is not the problem. The problem is that it needs a person to type, and typing is the one part nobody wants to do.

Why shared spreadsheets go stale

Three things happen, in this order.

The entry lag. You spend on Tuesday and record it on Sunday, from memory or from a bank feed that says TARJETA 4587 COMPRA and nothing about what you bought. Whatever you type is a guess.

The ownership gap. A shared file with two owners has none. One person becomes the accountant by accident, resents it quietly, and stops.

The reconciliation cliff. Once a month is missing, catching up costs an evening. An evening is easy to postpone, so the gap grows until the file is abandoned rather than wrong.

Notice that none of these are solved by a better template. They are solved by removing typing from the loop.

Start from what you already do

You already photograph receipts, or you can. A phone camera at the till costs three seconds and captures more than any bank line: the merchant, the total, the date, the items, the payment method. That photo is the data. Everything after it is transcription — which is exactly the kind of work to hand to software.

The four habits below are what makes a household ledger stay accurate. They work with any tool, and they are the habits Cluenta is built around.

1. Capture at the till, not at the desk

Decide that a receipt is captured the moment it is handed over: photograph it and send it wherever it needs to go. Put the destination inside a chat app if you can — a household group, a bot — so capture uses an app that is already open, not one somebody has to find.

The test: capturing an expense should take less time than putting the receipt in your pocket.

2. Let a machine propose, and a person confirm

Any tool that reads a receipt is sometimes wrong: a total that includes a deposit, a merchant name that is a legal entity nobody recognises, a date in the wrong century. So do not let a parser write straight into the ledger. Let it propose, and let a person confirm.

That one confirmation is what makes the ledger trustworthy. It is also small enough to do while the kettle boils, which is why it survives as a habit when weekly data entry does not. In Cluenta it is a review queue: one proposal at a time, the receipt beside it, confirm or correct the category or reject.

3. Budget by category limits, not by a single total

“We want to spend less” is not a budget. A number per category is, because it tells you something on the twelfth of the month rather than on the first of the next one.

Keep the categories few enough to remember — groceries, eating out, transport, home, health, everything else. Six is plenty for a household. Twenty categories mean every expense becomes a decision, and decisions are what you are trying to remove.

Give each category a monthly limit, and decide once what happens to what is left over: does it roll into next month, or reset? Both are defensible. Choosing one and keeping it is what matters, because a rule that changes when it is inconvenient is not a rule.

4. Make the numbers the same for everyone

If one person’s app shows €412 of groceries and the other’s shows €418, both stop believing either. This is why the arithmetic should happen in one place — on a server, once — and both screens should display what it computed rather than each working it out from a partial list.

It is also why the ledger should be shared rather than mirrored. Two people tracking the same money in two tools is two ledgers, and two ledgers eventually disagree.

What “done” looks like

A household that has got this right can answer three questions in under a minute, without opening a spreadsheet: how much have we spent this month, on what, and is anything about to go over. Nobody in that household does data entry. Someone spends about a minute a day confirming what a machine proposed.

That is the whole goal. Not a beautiful spreadsheet — the absence of one.