What they are actually for
A category is not there to classify: it is there so you can decide something. If the monthly split leads to no decision at all, you have too many categories or they are cut in the wrong places.
The test is simple. Look at last month’s summary and ask: what would I do differently next month with this? If “Online shopping” took £240 you know nothing —groceries? clothes? a gift?— but if “Eating out” took £240 you know exactly which lever you have.
That gives the criterion: categories are cut by type of decision, not by type of shop or method of payment.
How many you need
Between fifteen and thirty. Fewer than fifteen and everything falls into buckets so wide they say nothing; more than thirty and you start hesitating over every purchase, which is precisely what breaks the habit.
Sumant ships 27 expense categories and 11 income ones by default, and that number is the reason: it covers the normal cases without making you think. The ones you never use, you archive. Creating your own categories is a PRO feature.
A split that works in almost any life, grouped by the kind of decision it enables:
What you do not decide monthly (housing, utilities, insurance, contributions, commuting): four or five categories. You do not save here day to day, you save once a year by renegotiating.
What you decide weekly (groceries, eating out, going out, clothes, treats): six to eight. These are the ones you look at when you want to change something.
What happens occasionally (health, gifts, travel, learning, pets, home): six or seven. Not controlled month to month; reviewed over the year.
The line between the first two blocks is fixed versus variable expense, and it is the one that pays off most if you are only going to make one cut.
The rule for not hesitating
When you do not know where something goes, do not ask “what did I buy”. Ask “what was it for”. The answer is almost always single, and always the same for the same case, which is what makes your data comparable across months.
Forty pounds at the supermarket for Saturday dinner with friends is not “Groceries”: it is “Going out”. A charger bought at the same shop as the groceries is not “Groceries”: it is “Home”. The shop is irrelevant; intent decides.
And the second half of the rule, the part that actually holds the system up: when in doubt, pick fast and move on. One miscategorised expense breaks nothing. Thirty seconds frozen in front of your phone does, because that is what stops you logging anything tomorrow.
The six hard cases
1. The mixed receipt
Eighty pounds at the supermarket, twenty of which is detergent and cleaning products. Do not split it. Put the whole thing in the dominant category and move on. Splitting receipts is the habit most people abandon in week one, and the error it corrects is about 2 per cent.
2. Amazon, eBay and the like
Never create a category named after a shop. On Amazon you buy food, clothes, gifts and things for the house: an “Amazon” category merges four different decisions and lets you take none of them. Categorise by what was in the box.
3. Subscriptions
You have two options and both are fine, but pick one and stick to it: either all together under “Subscriptions” —good if your aim is to review them in one go once a year— or each in its usage category, with music under “Going out” and cloud storage under “Home” —good if your aim is to know what your leisure really costs.
4. Cash withdrawals
Taking £100 out of a cash machine is not an expense: it is moving money from one account to another. Log it as an expense and you count it twice on the day you actually spend it. It goes as a transfer to a cash account, and from there it leaves as a normal expense when you use it.
5. Money paid back to you
You cover dinner for six and five people pay you back. That is not income: it is an expense that turns out to be smaller. Record it by reducing the original expense, not as money coming in. Log it as income and your “Going out” category will claim you spent £120 while your summary claims you earned £100, and both figures are wrong.
The same rule covers refunds on an online order and expenses your employer reimburses.
6. Costs split between personal and work
Fuel, the phone, the laptop. If you are self-employed you already have the answer: they go on the business account, not the personal one. If you are employed and your company does not pay it, it is yours and goes in its normal category. What does not work is splitting every tank of fuel 60/40: that gets abandoned within a fortnight.
Subcategories: when they earn it
The urge to split “Groceries” into “Supermarket”, “Greengrocer” and “Butcher” shows up early. It is almost always a bad idea, and there is a test: a subcategory earns its place only if you would take a different decision for it than for its parent.
“Going out → video subscriptions” passes the test: you can cancel one and keep the others. “Groceries → greengrocer” does not: you are not going to stop buying fruit because the number looks high.
And a size rule: do not create subcategories inside anything under 10 per cent of your monthly spending. Refining a small bucket buys precision on a figure that changes nothing, and costs you one more hesitation every time you log.
If yours are already a mess
Six months in with half of it under “Other”, the natural reaction is to want to recategorise the whole history. Do not: it is hours of work to repair data you are never going to use for a decision.
Do this instead. First, fix the categories themselves: remove the ones you never use, merge the ones that overlap and create the two or three you were missing. Second, recategorise the last month only, since that is the one you will compare against. And third, draw a line: from today, with the new categories.
You will lose comparability with earlier months for a while. That is a low price against the alternative, which is usually abandoning the app on a Saturday afternoon while recategorising March.
Keeping them alive
Once a quarter, look at two things. First: which categories you have not used once. Archive them; every option you never pick is one more hesitation at logging time. Second: which category ate more than 20 per cent of the month. If it is a single big one, it probably hides two different decisions and deserves splitting in two.
Once you are a few months in, the repetitive work can be taken off your hands: in Sumant categorisation rules assign a category on their own when the description matches, and your own categories let you go finer without multiplying the main menu. Both are PRO features. What the free plan gives you is the 27 defaults and archiving the ones you do not want.
If you are coming from a spreadsheet with twenty columns, the move from columns to categories is covered in the Excel template for monthly expenses, and if you are not logging anything yet, the order to start in is in the starter guide.