Maintenance reserve for short-term rentals: the per-night number most hosts get wrong

The dishwasher died on a Sunday in February, the day a five-night booking checked in. Replacement plus install ran $710. Two months earlier the air-conditioning compressor had gone — $1,840. Three months before that, a slow leak under the kitchen sink cost $260 in repair and $480 in two cancelled nights while the floor dried. Twelve months, three failures, $3,290. The listing grossed $38,400 that year. The maintenance line on my spreadsheet had read 1.5% of gross at the start of January.

A 1.5% reserve set aside $576 for the year. The actual repair bill was $3,290. That is the gap. This post is the math.

The 1% myth, and where it came from

Property-management advice imported the 1% rule from long-term rentals. On a long-term unit you set aside 1% of the property's value per year for maintenance — a $300,000 house reserves $3,000. The number traveled into short-term-rental forums and got rewritten as 1% of gross revenue, and then dropped further to "round it down, it's mostly fine."

It is not mostly fine. Long-term and short-term wear at different rates because a short-term unit turns over every two to four nights instead of every twelve to twenty-four months. Every turnover is a guest who does not know the appliances, the locks, the boiler timer, or how hard the cabinet doors close. Cumulative wear per dollar of revenue is higher on short-term units, often two to four times higher, and the failures cluster around the same six items every host eventually replaces.

The closer benchmark is what hotel operators carry: a 5% to 6% reserve for replacements as a share of revenue (hotels call this FF&E reserve — furniture, fixtures, equipment). Short-term rental sits between a long-term landlord and a small hotel; the reserve should too.

The three buckets — what the money actually covers

A useful reserve is not one line; it is three.

Wear-out — the things you will replace on schedule

Items that age out whether or not anything dramatic happens. They have a known service life and a known replacement cost.

On a $40,000 gross year, that table alone is 4.9% of gross, and it is not done — it does not yet include the small stuff (broken lamp, lost remote, replacement keys, dented kettle) or the occasional surprise. Wear-out alone overshoots the 1% rule by 5×.

Break-in — the things that fail without warning

Acute failures: a pipe gives up, a compressor seizes, a window cracks, a fridge dies on day six of its eighth year. These are not on a schedule; they are on a probability curve. The pattern across forums and host conversations is one acute failure every 18 to 30 months in a unit run hard, costing $400 to $1,800 each time.

Run the expected-value math. If a $1,200 acute failure happens on average every 24 months, the annualised reserve for it is $600/year. Add the smaller stuff — toilet running, faucet dripping, doorknob loose — and call it $800.

Soft cost — the cancelled nights and the rush surcharge

When the AC dies on a 95°F afternoon, two things cost you money: the repair and the night you refunded because the guest opened a temperature-grievance ticket. A burst pipe takes the unit offline for three to five nights of drying time. A cancelled three-night stay at $170 ADR is $510 of revenue you do not earn, and the platform still bills its cleaning fee back if your policy was strict.

This is the bucket every host forgets. Out of those three failures I opened the post with, the $480 of cancelled nights from the under-sink leak was a real cost — same dollars as if I had paid the plumber that amount. Annualise: one offline incident per year costing two to four nights of lost revenue equals $340 to $680 at $170 ADR.

What the three add up to

That is the bill for a unit grossing roughly $40,000 a year. Eight percent, not one.

Per-night reserve — turn the percentage into a number you actually use

Percentages of annual gross are accountant-speak. The number you need at the bottom of every booking is a per-night figure you can subtract.

The conversion is straightforward: total annual reserve divided by total nights booked. On the same $40,000 unit at, say, 220 nights booked and a $180 ADR:

$3,265 / 220 nights = $14.84 per night booked

Round to $15 per night and that is the line you remove from each booking before you call the net "yours". On a 4-night stay grossing $720, you set aside $60. On a 30-night month at 75% occupancy (about 22 nights), you set aside $330.

If you are pricing your nights without removing this line from each one, your real margin is lower than your spreadsheet thinks.

Adjustments — when 8% is too much or too little

Eight percent is a working average. Three things move it.

Age of the unit. A two-year-old furnish-out with appliances all under warranty runs closer to 4–5% — most wear-out items are years from replacement. Years three to seven hit the sweet spot at 7–9%. After year ten the rate climbs to 10–12% as multiple wear-out items reach end-of-life concurrently.

Climate and humidity. Coastal units corrode faster (HVAC compressors, metal fittings, electronics). Tropical and humid coastal climates add 1–2 percentage points. Dry mountain climates often run a point under average.

Turnover frequency. A unit running 30 turnovers a month (mostly one-night bookings) wears at roughly 1.5× the rate of one running ten turnovers (mostly week-long stays). Minimum-night rules are not just about cleaning cost — they push the reserve number down too. The interaction with cleaning logistics is in cleaning buffer days.

A 12-year-old beach condo running short stays needs the reserve closer to 11–13%. A new mountain cabin doing seven-night minimums could honestly carry 5%. Most units are somewhere between.

Where to keep the money

A reserve is only a reserve if it is hard to spend. Three ways hosts solve this:

• Separate savings sub-account at the same bank. Free, takes ten minutes to open. Set up an automatic transfer of the per-night amount × monthly nights booked. This is the minimum-viable setup.
• High-yield savings at a separate bank. A reserve that earns 4% interest while it waits is the same reserve, just less expensive. Marcus, Ally, Wise — any of them work.
• Treasury bills or a money-market fund for hosts running multiple units. Higher yield, slightly more friction to access. Match the maturity to your acute-failure frequency: don't lock a year of reserve into a 12-month T-bill if you have any history of mid-year failures.

The wrong place is the operating account. If the reserve sits next to last week's payout, "I'll just borrow $300 from maintenance to cover the cleaner this week" becomes a habit, and the reserve evaporates before any of it gets spent on a repair.

Annual reset — the look-back that fixes your number

The number you set in January is a guess. The number you set the following January is data. Pull the year's actual maintenance spend — repairs paid, replacement appliances bought, refunded nights, cancelled bookings tied to failures — and divide by gross revenue.

If that fraction landed above your reserve rate, you under-reserved and the difference came from somewhere else (probably your own margin). Raise the rate by two points and revisit next year.

If it landed below, do not lower the rate immediately. One light year is noise; two consecutive light years is signal. Tying reserve adjustments to two-year rolling averages is the same discipline that hotel asset managers use, and it stops you from cutting the reserve right before a multi-replacement year.

One opinionated take

Hosts treat maintenance the way they treat dental insurance: an expense they hope they don't have to use, kept just barely funded, and then they're shocked when the cleaning crew finds a broken thing at 8am on a turnover day. The reserve is not a tax. It is the deferred half of the income statement. Every night booked has a small wear cost attached that you choose to either set aside in advance or pay out of next year's margin in surprise lumps.

The hosts running the most profitable single-listing operations I know have one habit in common: they price every night with the reserve removed first, and call the rest income. The math feels worse on day one — your ADR-times-nights number drops by 5 to 8 percent — and immeasurably better on the morning the HVAC compressor dies, because the money is already there.

If you want the reserve line surfaced against your real nights booked and actual repair history, that is what RentTools is built to track. And if you want to see how the reserve interacts with cleaning costs and channel fees, the full operator math is in channel manager break-even math.