The sixty-second answer
Categorised receipts answer three questions a bank balance cannot: which costs move with revenue, which are fixed no matter what, and which have crept upward without anyone deciding. The third is where correctable money usually sits. You need twelve months for it to mean anything, because anything shorter hides both seasonality and annual charges.
The balance is a result, not an explanation
Most owners know roughly what is in the account and roughly what came in. What is harder to answer is why two months with similar revenue produced very different amounts left over.
That question is unanswerable from totals. It needs the spending broken into categories that mean something operationally, and it needs enough history that one unusual month does not dominate the picture.
Fixed and variable is the first useful cut
Before any clever analysis, split costs by whether they move with the amount of work you do.
Variable costs scale: materials, fuel, subcontractors, payment processing. If you do half as much work, they roughly halve. Fixed costs do not: rent, insurance, most software, most phone lines. They arrive whether or not the month was good.
The ratio between them is the single most useful number this exercise produces, because it tells you what a bad month actually costs. A business that is mostly variable can absorb a slow quarter. One that is mostly fixed cannot, and needs a larger cash buffer to survive the same dip.
The costs that grew without a decision
The interesting category is neither fixed nor variable. It is the cost that has risen steadily and gradually, where no single increase was large enough to question.
Software seat counts do this as people join and nobody removes the leavers. So do shipping costs after a carrier price change, and insurance at renewal. Each individual step is small and defensible. The cumulative effect over eighteen months is not.
These are invisible month to month and obvious year over year, which is the entire argument for comparing the same month across two years rather than comparing this month to last.
Why categories have to be boring
The temptation is to build a detailed category structure. Detailed structures decay, because the person filing a receipt has to make a judgement call, and judgement calls made in a hurry are not consistent.
A category scheme works when each receipt has one obvious home. Most small businesses need somewhere between eight and fifteen. If you find yourself hesitating, the problem is usually two categories that overlap rather than a receipt that is genuinely ambiguous.
This is also where bank-feed categorisation quietly fails. A feed classifies by merchant, which is stable but wrong in a specific way: the same hardware store is inventory one week and a repair the next. The merchant did not change. The reason did, and only the receipt knows the reason.
Asking questions instead of building reports
The traditional output here is a report - a chart of spending by category, produced monthly, looked at for about forty seconds. Reports answer the question that was anticipated when they were designed.
Most real questions are not anticipated. They arrive because something felt off: why was March expensive, what did we spend at that supplier last year, is the fuel bill actually up or does it just feel that way. Those are one-off questions, and building a report for each is not realistic.
That is the job Penny does in MapleExpense - answering specific questions about your own categorised spending in plain language, with the real figures and dates from your records, so the answer takes a sentence rather than an afternoon.
What the exercise is worth
Be realistic about the payoff. This will not usually reveal a single dramatic waste. What it reliably produces is a short list of things that grew quietly, a clearer sense of how much a slow month actually costs, and the ability to answer a supplier question without going through a shoebox.
The records have to be kept anyway - generally until six years after the end of the last taxation year they relate to [1]. Given that the retention is compulsory, getting some operational use out of the same records is close to free.