Why a normal budget fails
A classic budget allocates a known income: £2,100 comes in, £2,100 gets assigned. The 50/30/20 rule works the same way: percentages of a figure you treat as certain.
When income swings between £900 and £3,400, those percentages mean nothing. Twenty per cent of a good month and twenty per cent of a bad one are different amounts, and your rent does not know that. The usual outcome: you save hard in March, dip into savings in April, and end the year where you started with the added feeling of not being in control.
The shift is to stop budgeting against what arrives this month and start budgeting against what arrives in the worst month.
First: your spending floor
Your floor is what your life costs in a month with nothing extra: rent or mortgage, utilities, groceries, transport, insurance, contributions and subscriptions. No going out, no treats, no travel.
Working it out needs data, not estimates. Take the last three months of transactions, strip out anything that does not repeat and add up what remains. If you do not have that recorded, this is the reason to start today: the rest of the method depends on this number.
It almost always comes out higher than people say from memory. Small recurring charges and annual bills spread across the year are what break the sum; the glossary covers it as the latte factor.
With the floor in front of you, you know two things: how much you need every month no matter what, and how much of a payment is genuinely available.
Second: the buffer month
The intermediate goal is not “save 20 per cent”. It is having one whole floor put aside: a full month of living costs, available tomorrow.
With a buffer month you stop earning and living in the same month. You get paid in June and that money pays for July. It turns irregular income into something that behaves like a salary: in June you already know what July can cost, because the money is already there.
After that first month, the next goal is a separate emergency fund, which is never used to cover a quiet month. They are two different pots and mixing them is exactly why the emergency fund never grows.
Third: what to do in good months
This is where the year is won or lost. A good month is not a month to spend more: it is a month to pay the bad ones in advance.
Split whatever sits above your floor in this order, and only move down when the step above is covered:
1. Finish the buffer month. Until it is complete, everything spare goes here.
2. The big payments you already know are coming. Tax, car insurance, the annual service, a trip you have committed to. If you know the date and the amount, it is not a surprise: it is a deferred expense.
3. The emergency fund, up to three or six months of floor depending on how steady your work is.
4. Everything else: goals, investing, something for yourself. By this point you have genuinely earned it, and you can even work out how much.
The rule holding all of this together is not raising your floor during a good run. Three strong months that lead to a pricier flat and two more subscriptions have turned variable income into fixed costs, and that only moves one way.
And the bad months
A month below your floor is not the method failing: it is exactly the scenario the buffer exists for. What matters is the reaction, and there are three rules.
Draw on the buffer, not the emergency fund. A quiet month was expected; a broken car was not. Mix them and the day the car breaks you will have nothing and will cover it with a card.
Do not touch your category caps that month. The temptation is to cut groceries and going out the moment less money arrives. If the buffer covers the month, cutting is unnecessary and actively harmful: it turns the system into something that punishes you when things go badly, and those are the systems people abandon.
Count consecutive bad months. One is noise. Three in a row is a trend, and at that point the answer is no longer financial: you need more work, a different price or a different kind of client. The budget only warns you; it does not fix that.
A full year with numbers
Someone with a floor of £1,450 a month —£17,400 a year— invoicing per project. Their year comes out like this:
Quarter 1: £2,900, £1,100 and £3,600. That is £7,600. They pay themselves £1,450 a month, so £4,350 goes out and £3,250 stays in the incoming account. That completes the buffer month —£1,450— and leaves £1,800, which goes to the second step: setting aside for car insurance and the annual renewal of two tools.
Quarter 2: £800, £0 and £2,400. That is £3,200 against £4,350 of salary. It eats £1,150 of the buffer. All three months they are paid the same as always and their life never notices. This is the system working, not failing.
Quarter 3: £4,100, £3,800 and £2,200. That is £10,100. They put the £1,150 back into the buffer and are left with £4,600 above salary. This is where the year is decided: that money goes to the emergency fund, not to raising their standard of living.
Quarter 4: £1,900, £2,700 and £1,300. That is £5,900, covering salary and leaving £1,550.
Total invoiced: £26,800. Total paid to themselves: £17,400. Difference: £9,400, part of which went on the big planned payments and the rest of which is saved. They lived twelve months on the same money every month while their income swung between £0 and £4,100.
Notice what did not happen: they did not save more in good months through willpower. They saved because the salary was fixed in advance and the surplus never reached their personal account.
What to check each month
Ten minutes, on the day you get paid. Have I covered next month’s floor? How much sits above it? Which step on the list does it go to?
Answering that needs two things: transactions recorded, and a cap per category that does not move every month. In Sumant monthly budgets per category are set once against your floor, not against what you earned, and savings goals let you keep the buffer and the emergency fund as two separate pots, each with its own progress.
If your irregular income comes from invoicing on your own account, the step before this one is separating business money from your own; that is in freelance finances without mixing accounts.
One warning to finish on, because this is where most people fall off. The method takes six to twelve months to show: you need a history to work out the floor properly, and you need to fill the buffer before it starts doing its job. For the first three months you will be recording data and underpaying yourself with no visible reward.
That part cannot be shortened, and it is also why almost everyone goes back to improvising. For reference: the month you stop earning and living in the same month is the one that changes how it feels, and that month only arrives if you keep recording through the ones before it.