Money habits
How to categorise expenses so the data is actually useful
Most category lists are built to describe the past. A useful one is built to answer a question you will actually ask. Seven rules from five years of keeping a household ledger.
Five years ago we started recording every transaction a household makes. Not the big ones. Every one. The coffee, the salary, the transfer from the current account to the savings account, the €4 top-up for a metro card in a city nobody in the house lives in any more.
For the first two years the data was almost worthless.
Not because it was incomplete or wrong. It was complete and it was accurate. It was worthless because the categories were wrong, and a ledger with bad categories is a very expensive way of producing a pie chart you already knew the answer to. Groceries: 18%. Yes. And?
The fix was not a better app, and it was not more discipline. It was realising the category list had been built backwards.
The mistake: categorising by what you bought
Almost everyone builds their first category list by looking at a bank statement and naming what they see. Supermarket, restaurants, fuel, pharmacy, Amazon, phone bill. It feels obviously correct. You are describing reality, and each transaction has exactly one home.
The problem is that this list describes what the money went to, and no decision you will ever make is shaped like that. You are never going to sit down and decide “I will do 12% less pharmacy this year.”
A category list is not a filing system. It is the set of buckets you will later compare, argue about, and act on. So it has to be built from the decisions, not from the receipts.
Rule 1. Write the question before you write the category
Before a category earns a place on the list, finish this sentence: “I want this category so I can answer …”
If you cannot finish the sentence, the category is decoration.
- “Subscriptions”, so I can see the total I have quietly committed to every month and cancel the dead ones once a year. Good. That number changes behaviour.
- “Amazon”, so I can see how much I spent at Amazon. Bad. Amazon is a shop, not a decision. That €600 was a mixture of nappies, a drill and a book, and lumping them together hides all three.
- “Eating out”, so I can see whether the restaurant habit is drifting. Good. And worth splitting from “coffee and lunch at work”, because those are two different habits with two different fixes.
This one rule kills about a third of most people’s lists.
Rule 2. Split by controllability, not by shop
The most useful cut in a household ledger is not what you bought. It is how hard it would be to change. Money divides naturally into four tiers, and the tiers behave completely differently.
Committed. Rent or mortgage, insurance, school fees, loan repayments, the phone contract. You cannot change these this month. You can change them once a year, with paperwork, and the saving is large.
Living. Groceries, utilities, fuel, transport, basic household. You control these continuously, in small amounts, and the total moves slowly. This is the tier where people burn enormous willpower for very little money.
Chosen. Restaurants, travel, hobbies, clothes beyond replacement, gifts, the good wine. Fully discretionary, high variance, and this is the important bit, it is where the money actually is. This is where a real decision gets made.
One-offs. The washing machine. The visa renewal. The flight home for a funeral. Individually unpredictable, collectively very predictable.
Once your top level is cut this way, a month tells you something. “Living was flat, Chosen was up 40%, and there was one big One-off” is a sentence you can do something with. “Groceries 18%, restaurants 9%, transport 6%” is not.
Rule 3. Keep the top level small, put the detail underneath
Eight to twelve top-level categories. If you have thirty, you will never look at the list, and worse, you will hesitate every time you enter a transaction. Hesitation is what kills tracking habits.
Detail is not lost. It moves down a level. Chosen → Restaurants, Chosen → Travel, Chosen → Hobbies. Or, for anything you only occasionally care about, it moves into the description field, which is searchable and costs nothing to maintain.
The test: could you assign a transaction to a top-level category without thinking, from a phone, standing in a queue? If not, the list is too fine.
Rule 4. Transfers are not expenses. This is the big one.
If you take one thing from this article, take this.
Moving €2,000 from your current account to your savings account is not an expense. Paying off your credit card is not an expense. Buying €500 of an index fund is not an expense. Sending money to your own account in another country is not an expense.
Every one of these is a transfer. Your net worth does not change, only its location. And yet almost every simple tracking system treats them as spending, which produces a household that appears to spend more than it earns while its savings grow. We have watched people conclude they had a spending problem when what they actually had was a transfer problem in their spreadsheet.
Your system needs three transaction types, not two.
| Type | Net worth | Example |
|---|---|---|
| Income | up | Salary, dividend, refund from a shop |
| Expense | down | Groceries, rent, a flight |
| Transfer | unchanged | Current to savings, credit card payment, buying an investment |
The moment transfers are separated out, two numbers become trustworthy for the first time: your real monthly spend, and your real savings rate. Before that, both are fiction.
The awkward cases, resolved:
- Credit card spending is an expense on the day you buy, and the repayment is a transfer. If you record the repayment as the expense, your categories are wrong by a month and your card debt is invisible.
- A refund is negative income if it relates to something you did not categorise, but it is cleaner as a negative expense in the original category. That way the category total tells the truth.
- Buying an investment is a transfer into an asset. What eventually shows up as income is the gain or the loss, not the purchase.
- A mortgage payment is two things in one. The capital portion is a transfer, because you are buying your house back from the bank. The interest portion is an expense. Splitting these is the difference between understanding your housing cost and badly overstating it.
Rule 5. Give one-offs somewhere to live
A washing machine dies in April. If it lands in “Household”, April looks like a catastrophe, your twelve-month average is polluted, and the useful signal is destroyed. That signal being: your ordinary household spending was completely normal.
Keep a top-level One-offs category, or a flag you can filter out. Then look at it two ways. Excluded, to see whether your ordinary life is under control. Included, once a year, to see the truth.
Because the annual total of “unpredictable” one-offs is remarkably stable. In most households it is a genuinely large number and it recurs every single year. The only unpredictable parts are which month and which appliance.
Rule 6. Never create a category you will be tempted to lie to
If you have a category you feel slightly bad putting things into, you will stop putting things into it. Then you have accurate-looking data that is quietly false. This is the most dangerous failure in personal tracking, because it does not look like a failure.
The practical version of this rule: keep an Unclassified bucket and use it freely at the moment of entry. It is far better to record a transaction in five seconds with the wrong category than to not record it at all. Then, once a month, sit down and clear the bucket properly. Speed at entry, accuracy at review.
Rule 7. Change the list rarely, and fix history when you do
Categories are only useful compared over time. If you rename and reshuffle every few months you have a series of unrelated snapshots rather than a trend, and the whole point of tracking evaporates.
So change the structure at most once a year, do it at a clean month boundary, and when you do, recategorise the history. A mapping from old to new, applied in bulk, is an hour of work that preserves years of value. Not doing it costs you the ability to answer the only question that matters: is this getting better or worse?
A starting structure
This is a shape, not a prescription. Adapt the second level to your life. The first level should survive almost unchanged.
| Top level | Underneath |
|---|---|
| Income | Salary, bonus, side income, interest and dividends, refunds |
| Housing (committed) | Rent or mortgage interest, service charge, property tax, home insurance |
| Committed | Insurance, loan repayments, school and childcare, subscriptions, phone and internet |
| Living | Groceries, utilities, transport and fuel, health and pharmacy, household basics |
| Chosen | Restaurants, coffee and lunch, travel, hobbies, clothes, gifts, personal care |
| One-offs | Appliances and repairs, medical events, admin and visas, moving costs |
| Unclassified | Emptied every month |
| (Transfers) | Not a category. A transaction type. |
Note what is missing. No “Amazon”, no “Miscellaneous”, no thirty-item long tail. And note that Housing has been pulled out of Committed as its own top-level line. For most households it is between a quarter and a half of everything, and hiding it inside a bigger bucket wastes the most important number on the page.
How to move an existing mess onto this
Do not re-do five years by hand. You will start and not finish, and a half-migrated ledger is worse than the mess you had.
- Write the new list first, on paper, using Rule 1 on every line.
- Write a mapping from each old category to a new one. Most map cleanly. The ones that do not are exactly the categories that were wrong.
- Apply the mapping in bulk to history. A find-and-replace in a spreadsheet, or a rules-based recategorisation if your app supports it.
- Fix transfers next, because that is where the real distortion is. Search history for round-number payments to your own accounts and to credit cards, and retype them.
- Then stop. Do not hand-clean individual old rows. The trend is what you need, and the trend survives a small amount of noise.
Categories are not admin. They are the entire interface between the effort of tracking and the value of having tracked. Get them wrong and you can record every transaction for five years and still not answer a single question about your own money. Get them right and a two-minute look at last month tells you what to do next.
That is the whole return on the habit.