Most people can describe their large fixed costs accurately and are substantially wrong about everything else. Tracking for a defined period resolves it.
This describes the exercise rather than advising on anybody's circumstances.
Why estimates are unreliable
The categories where people are consistently wrong share characteristics.
Frequent small transactions, which are individually forgettable and collectively substantial. Coffee, snacks, transport, small conveniences.
Irregular purchases, which are remembered as occasional and occur more often than recalled.
Subscriptions, which are invisible because they require no action.
And anything on a card rather than in cash, since the friction of handing over notes produces a memory that a contactless tap does not.
Research on spending recall consistently finds underestimation in exactly these categories.
Why three months
One month is misleading because it may be unrepresentative — a month with an unusual expense, or an unusually quiet one.
Three months captures more variation and includes at least some irregular costs.
A full year would be better and almost nobody sustains it, which is why the shorter period is the practical recommendation.
The compromise is to track three months carefully and separately list the known annual costs, which covers most of the picture.
How to do it without it becoming a burden
The methods, in rough order of effort.
Reviewing statements retrospectively, which requires no ongoing effort and works if most spending is on cards. This is the lowest-effort approach and is frequently sufficient.
Automated aggregation through banking applications or account information services, which categorise transactions automatically, with categorisation that is approximate and correctable.
And manual recording, which is the most accurate and the least likely to be sustained.
For most people, statement review plus a note of cash spending is adequate and takes an hour per month.
The categorisation question
Where the exercise frequently gets bogged down.
Fine-grained categories produce arguments with yourself about whether something is groceries or household, which adds effort and no information.
Broad categories — housing, transport, food, subscriptions, discretionary — are sufficient to reveal the pattern.
The purpose is to see where money goes, not to produce accounts, and precision beyond that is wasted.
What it typically reveals
The findings that recur.
Food spending, particularly the split between groceries and eating out, which is nearly always higher than estimated.
Subscriptions in total, which almost everybody underestimates and which frequently includes forgotten items.
Small frequent transactions, which aggregate to a figure that surprises people.
And the gap between total income and total spending, which reveals whether the arithmetic works at all and which some people have never calculated.
The uncomfortable part
Worth acknowledging since it stops people.
The exercise frequently reveals spending that is difficult to look at, and the instinct is to stop.
What generally helps is treating it as data collection rather than as judgement, and deferring any decisions until the period is complete.
Nothing needs to change during the tracking. Changing behaviour while measuring it produces a measurement of the changed behaviour rather than the actual one.
What to do with the result
The decisions that follow.
Identify the two or three categories where the number was most surprising, since those are where the largest available adjustment sits.
Distinguish between spending that produced value and spending that did not, which is a more useful frame than necessary and unnecessary.
Total the irregular annual costs and convert them to a monthly figure.
And then build the automated system described elsewhere, so that the tracking does not need to continue.
Repeating it
Worth doing periodically rather than continuously.
Circumstances change, and a pattern established three years ago may no longer describe anything.
Repeating the exercise annually, for a month rather than three, is sufficient to catch drift.
Which is a manageable commitment and considerably more likely to happen than permanent tracking.
Shared finances
Where the exercise requires more care.
Households with combined or partly combined finances need both parties' spending to produce a complete picture, which requires a conversation rather than an individual exercise.
Doing it separately and comparing afterwards is frequently more productive than doing it jointly, since it avoids the review becoming a negotiation while the data is still being gathered.
And agreeing in advance that the purpose is information rather than judgement is what determines whether the conversation is useful.
Cash and the invisible portion
The gap in any statement-based method.
Cash withdrawn appears as a single transaction and disappears into unrecorded spending.
Which means statement review alone systematically underestimates the categories where cash is used, and attributes the whole withdrawal to nothing in particular.
For anybody using cash regularly, noting roughly where it goes for the tracking period closes the largest gap in the exercise.
For anybody using cards almost entirely, which is increasingly common, the statement method is close to complete on its own.
Open banking and the aggregation services
Worth knowing since it makes the exercise considerably easier.
Regulated account information services can read transactions across multiple accounts with your permission, producing a combined view without manual entry.
Access is granted for a period and must be renewed, which is a deliberate protection rather than an inconvenience.
The things worth checking before granting access are whether the provider is regulated in your jurisdiction, what data is retained, and how to revoke permission.
Revocation is generally available through the bank as well as the service, which is the reliable route.