How to record cash spending without inventing the numbers
Cash is where a household ledger quietly goes wrong: no statement, no notification, and the purchases you lose are the small frequent ones you wanted to see. Four approaches, honestly compared.
Look at almost any household ledger that has been running for a few months and you will find the same wound in the same place. The card spending is complete, or close to it. The transfers are there. The rent is there. And then there is a fuzzy region — market stalls, the barber, a round of coffees, the parking meter, the note handed to a child on the way out of the door — where the record thins out, becomes approximate, and eventually stops.
This is usually described as a discipline problem. It is more accurate to describe it as a structural one, because cash is missing three things that every other payment method gives you for free, and no amount of resolve replaces them.
Why cash goes missing, specifically
First, there is no statement. A card transaction is recorded twice — once by you, if you bother, and once by the bank whether you bother or not. That second copy is a safety net: forget to enter something and you can still find it a week later. Cash has no second copy. If you do not write it down, the only remaining record is your memory, and memory of a small transaction decays fast enough to be unusable within about forty-eight hours.
Second, there is no notification. The banner a card produces is a prompt arriving while the expense is still fresh, and nothing at all happens after you hand over a note.
Third, and most damaging, cash is not a random sample of your spending. The purchases people still make in cash skew small, frequent, local and immediate: the bakery, the top-up, the tip, the school fundraiser, the flowers, the taxi in a place where the driver would rather not take a card. Card spending skews large, planned and online. So the part of your ledger that goes missing is not a random third of the total — it is concentrated in exactly the daily, discretionary, "where does it all go" region that made you start tracking in the first place.
Reconstruction from memory is biased, not merely incomplete
The standard rescue plan is to sit down on Sunday and reconstruct the week. The result is not a slightly thin version of the truth; it is a systematically distorted one.
What survives recall is what is distinctive: the large amount, the unusual purchase, the thing you had feelings about. What disappears is the repeated and the ordinary, which is where the volume lives. Three coffees become "a coffee, I think"; six small purchases become four; the sandwich you bought because you were late becomes nothing at all. Every one of those errors runs in the same direction, so a reconstructed week is an underestimate with a confident face on it, and it understates exactly the categories you wanted to act on.
This matters because a wrong number is worse than a missing one. A missing week announces itself; a reconstructed week looks like data, gets averaged into your monthly figure, and quietly makes your baseline look better than it is.
Four approaches, and what each one actually costs
Record at the counter
The only method that produces categorised, accurate cash data. You enter the amount while the change is still in your hand. It takes a few seconds and it works because the interval between the event and the record is essentially zero, which removes the memory problem entirely.
Its cost is social and it is real: some people are uncomfortable standing at a till with a phone out while somebody waits behind them. The honest mitigation is to step aside — you do not have to do it at the counter, only before you have thought about something else. The other requirement is technical: the entry must save in seconds and it must save with no signal, because market halls, basements, car parks and foreign countries are exactly where cash gets spent. In Fambook, entries are written to the phone’s own database first and pushed to the server afterwards, with anything still waiting shown as a count rather than hidden, which is what makes recording at the till possible rather than aspirational.
Wallet reconciliation
A different idea entirely, and an underrated one. You do not track cash purchases at all. Instead you record the withdrawal as an event, and then at a fixed interval — every Sunday, or the first of the month — you count what is physically left in your wallet. The difference between what you took out and what remains is what you spent, and you book that difference as a single entry.
The trade is explicit and it is a good trade for a lot of households: you give up all category detail on cash and in exchange your totals become exactly correct. Nothing is missing, nothing is invented. If you withdrew two hundred and have thirty left, you spent a hundred and seventy, and no act of memory was involved in producing that number.
What it will not tell you is what the hundred and seventy bought, so it is wrong if cash is where your problem categories live and right if cash is a background hum you want the totals to include. It also requires remembering that money moves in as well as out — a refund, a repayment from a friend or a note from a relative all break the arithmetic if unrecorded.
A single daily cash line
One entry a day, at a fixed moment — usually the evening — covering everything you spent in cash that day, in a single category. It is a compromise between the two above: a day is short enough that recall is tolerable, and one entry a day is a habit small enough to survive.
Its specific failure is the day you forget: unlike the wallet method there is no self-correcting mechanism, so a missed Tuesday simply vanishes. Pair it with an occasional wallet count as a check.
Envelopes
Physically dividing cash into envelopes or pockets for food, transport, going out, and so on, and spending only what is in each. It is the oldest method here, it long predates every app on your phone, and it stubbornly refuses to die, which is worth taking seriously rather than sneering at.
It survives because it inverts the whole problem. Every other method records what happened; envelopes constrain what can happen. There is no recall step, because the envelope is the ledger — you can see how much is left in the food envelope without opening anything or remembering anything, and when it is empty you have received the information immediately and physically rather than in a review three weeks later. That immediacy is exactly what apps struggle to reproduce.
Its costs are equally real. It handles only cash, it produces no history, so you cannot compare March to October or look back over a year to find your irregular costs, and it is awkward across two people. Many households end up with a hybrid — envelopes for the categories that need a hard stop, an app for everything else — and that hybrid is a legitimate design rather than a failure to commit.
The lump line is more honest than a plausible breakdown
When you reach the end of a period and know that some cash is unaccounted for, there are two things you can do. You can distribute it across categories according to what you think you probably bought, or you can enter one line — "unaccounted cash" — and leave it undivided.
Take the second option, always. A guessed breakdown launders an estimate into the same visual form as a measurement: six months later nothing distinguishes the forty you recorded at a till from the forty you assigned to groceries because it seemed likely, and that category is now contaminated in a way you cannot detect or undo. A lump line is self-labelling. It says the totals are right and the detail is not, and it says how much detail is missing — if it is two percent of the month, stop worrying about cash; if it is a quarter, you know your category analysis is not worth much and you know why.
A workable setup
- Decide which of the four methods you are using, per household, out loud. Drifting between them is what produces a ledger that is neither accurate nor honestly approximate.
- If you are recording at the till, test the app in airplane mode before you rely on it. An entry that needs a signal to save is not available in the places cash is spent.
- Record the withdrawal as a transfer, not an expense, or you count the same money twice.
- Set a single fixed moment for the wallet count or the daily line, and attach it to something you already do. Habits attached to a time of day survive; habits attached to an intention do not.
The goal is not a perfect record of cash. It is a record whose errors are visible, bounded and labelled.
Frequently asked questions
Is it worth tracking cash at all if most of my spending is on cards?
Track the totals, not the detail. Record withdrawals and do a wallet count every week or two so the household total is correct, and accept a single uncategorised line for what the cash bought. If cash turns out to be a small share of your month, that is a useful finding in itself and you can stop there.
Can I just reconstruct my cash spending at the end of the week?
You can, but understand what you are producing. Recall preserves the large and distinctive and loses the small and repeated, so a reconstructed week is biased low in exactly the everyday categories you were trying to see. It is not a thinner version of the truth, it is a skewed one, and it looks like data afterwards.
How do I record a cash withdrawal so I do not double count?
The withdrawal is a movement between your own pockets, not an expense, so record it as a transfer or simply as a marker of how much cash entered your wallet. The expense happens when the cash is spent. Booking both the withdrawal and the purchases as spending counts the same money twice.
What should I do with cash given to me by a relative?
Record it as income at the moment it arrives, otherwise it appears in your ledger as spending that came from nowhere and breaks a wallet reconciliation. This is the most common reason a careful cash count fails to balance — money moves in as well as out, and only the outward direction feels like something worth writing down.
Do envelopes still make sense when I have a budgeting app?
For the one or two categories where you want a hard stop rather than a report, yes. An envelope constrains what can happen; an app records what did happen, and those are different jobs. A hybrid — envelopes for the categories that need a limit you can feel, an app for the history and the totals — is a reasonable design rather than an admission of defeat.
Does the app need a signal to save a cash entry?
It should not, and that is worth testing before you commit to a method. Fambook writes each entry to the phone first and uploads it later, retrying on launch and when the network comes back, and shows a count of anything still waiting. Cash is spent in basements, markets and abroad, which is exactly where a network-dependent app fails.
Try it for one month
Fambook gives a household one shared ledger: anyone can add an entry in seconds, every entry says who spent it, and the month adds up in one place instead of two. Records with no signal and syncs afterwards. Recording, categories, budgets, statistics, CSV import and export, sync and sharing for two people are free — the subscription only buys you less typing.