An appointment is not an emergency
A boiler that dies is an emergency. A boiler service is an appointment. The distinction sounds pedantic and is worth real money, because emergencies need a buffer you hope never to spend, while predictable annual costs need a schedule.
Yet every year the same costs ambush the same households. The car insurance renews. Christmas arrives. The MOT comes round. None of it is a surprise (the dates have sat on a calendar for twelve months), and each one still gets treated as an unexpected expense, paid on a credit card, quietly undoing months of progress.
A sinking fund is the schedule those costs were missing. Divide a known future cost by the months until it is due, set that amount aside monthly, and when the bill lands the money is already there; the month it arrives in feels like every other month. The term comes from corporate finance, where companies set aside money over time to retire a bond at maturity. The household version is identical in structure and considerably more useful.
Twelve dates, one standing order
List each irregular cost, what it comes to, and the month it falls due, and enter anything you have already put aside in the "have" column. The planner works out how many months remain (wrapping round the year end, so a January renewal viewed from October is three months away, not minus nine) and divides what is left across them. The total at the side is the one number you need: a single monthly amount covering every fund at once. Set one standing order for that figure into a separate savings pot on payday and the entire system runs itself.
Costs worth including: insurance premiums of every kind, vehicle tax, MOT and servicing, Christmas and birthdays, holidays, professional subscriptions and memberships billed yearly, school uniform and trips, dentist and optician, haircuts if you batch them, appliance replacement, and the annual software renewals that surface in your subscription audit. Which of those genuinely deserve a fund of their own is a judgement call, and the setup guide works through where most people draw the line.
Two of the planner's assumptions deserve a sceptical eye as you fill it in. It takes your target amount and due date as accurate, and both drift: insurance premiums rise, so a fund built on last year's number will fall short. Add a margin of ten per cent or so to anything inflation-sensitive (the ONS's inflation and price indices show how quickly prices are currently moving), and re-check the figures each year rather than rolling them forward blindly. It also spreads the remaining gap evenly across the months left, the simplest approach but not always the best one; if your income is seasonal, or you receive a bonus, front-loading a fund is more realistic than an even split.
Upkeep is one sitting a month. Bump the "have" column by whatever you transferred, and mark a fund back to zero after you spend it; that is the whole maintenance burden. The list itself is private: it is stored in this browser and never sent to us, and no account or bank link is needed.
One account, many labels
Keep the money in a single savings account and track the split here rather than opening a dozen accounts. The arithmetic is what matters, not the containers. Some banks offer named pots or spaces, which are pleasant if you have them, but they are a convenience rather than a requirement, and chasing them across providers is a good way to lose interest on the balance. What matters is that the money sits somewhere separate from your current account, because money in a current account has a way of becoming spent money.
The maths ignores interest earned on the balance. Over a year on a few thousand pounds that is a small effect, and ignoring it errs on the safe side; you will have slightly more than the tool says. It equally ignores the erosion of purchasing power on money held as cash, minor over a twelve-month horizon but not nothing. The Bank of England's inflation calculator makes the long-run version of that erosion vivid, and the inflation calculator here shows the scale of the effect over longer periods. Which marks the real boundary: this is not a place for money you will not need for many years. Cash is right for a known cost twelve months out and wrong for a goal a decade away, where the FIRE calculator and invested returns are the relevant tools.
That monthly total belongs as a line in your payday allocator under "save". It is a bill, not a leftover. Treat it with the same seriousness as rent and the system holds; treat it as optional and it collapses the first busy month.
A Christmas fund with an optimistic name
Sinking funds and your emergency fund are different jobs and should not share a balance. An emergency fund covers unknown events (job loss, illness, a car that dies), is sized against your essential monthly spending and your income security, which is what the emergency fund calculator works out, and you hope to leave it untouched for years. A sinking fund covers known events with dates attached. It is designed to be spent, on schedule, and refilled. If you spend your emergency fund on Christmas, you do not have an emergency fund; you have a Christmas fund with an optimistic name.
Keeping them separate also stops a genuine emergency from quietly destroying your annual plan, and stops predictable costs from making you feel like you are constantly raiding your safety net.
The catch-up year
The most common objection is time. If the car insurance is Ā£620 and it renews in two months, Ā£310 a month is not going to happen. That is real, and the honest answer is that the first year of sinking funds is always the hardest, because you are catching up on costs you did not fund in advance.
Two things help. Part-funding still counts: Ā£150 a month for two months means Ā£320 on a card instead of Ā£620. And the moment a fund is spent, its clock resets to a full twelve months. Set its saved amount back to zero, set the next due month, and the tool recalculates the monthly figure over the new, longer runway, at which point it drops to something comfortable. Year two is easy in a way year one is not.