The emergency furnace in January is only an emergency if nobody budgeted for the furnace. Maintenance doesn't surprise; it arrives on a schedule. What feels like a surprise is usually a budgeting failure wearing a repair's costume.

The rules of thumb — and their limits

All three are fine as rough reserve targets. None of them know how old your water heater is.

The budget that actually works: age-adjusted

Major systems have published service lives. Roofs run 20–30 years depending on material; water heaters 8–12; HVAC 15–20; major appliances around 10. Inventory the big systems with their ages and you have a capital plan: if the water heater is eleven years old, it isn't an "if" — it's a line item with a date range.

This is a fifteen-minute spreadsheet, and it eliminates most of what people call surprise.

Two budgets, not one

Keep them separate, or the reserve evaporates into a hundred small repairs and the roof arrives unbudgeted anyway.

Planned versus emergency pricing

The same repair, two ways

January, no heat: emergency callout — after-hours labor, expedited parts, one bid

September, scheduled: quoted price, three bids, off-peak scheduling

Scheduled work gets quoted; emergency work gets paid. After-hours labor rates, expedited parts, no time to collect competing bids, and you're negotiating in January from a position of weakness.

The seasonal calendar

Good maintenance budgeting makes repairs boring. That's the whole ambition.

A budget is only as good as the condition data under it: you can't schedule what you can't see, which is what periodic inspections are for (Kept's condition records exist largely to feed this — turning "surprise" into "scheduled" is most of the product).

The goal of a maintenance budget isn't to spend less, exactly. It's to convert emergencies into appointments. Good maintenance budgeting makes repairs boring. That's the whole ambition.