How do you actually decide how much to set aside for maintenance each month?

Question

I've been thinking about this a lot lately. Most advice I see is generic ("save 1-2% of property value per year") but that never accounted for the actual age of my systems.

A roof that's 3 years old and a roof that's 22 years old shouldn't have the same reserve treatment, but most of the rules of thumb I've found treat them the same.

Curious how others here actually calculate this. Do you use a flat percentage? Track individual system age? Just wing it and hope for the best?

Rachid Abadli · Answer

@Giuseppe Cavucci — @Rami David and @Mike Fisher nailed the component-level approach. The "1–2% of property value" rule is basically useless because property values have nothing to do with what a water heater costs to replace.

What I do for my Sacramento units: I built a simple replacement schedule for every major system — roof, HVAC, water heater, appliances, flooring, exterior paint. Each one gets a replacement cost, estimated remaining life, and a monthly set-aside. For your situation — furnace at 9 years (roughly halfway through its life) and roof at 14 years into a 20–25 year span — you'd want heavier monthly reserves on the roof than the furnace.

The number I've landed on across my portfolio is closer to 8–9% of collected rent for maintenance + CapEx combined, not property value. That's on a mix of properties from the 1950s to 2000s builds. Newer construction is obviously lower.

One thing nobody mentioned yet: your maintenance history should adjust your reserves, not just system age. If a unit has had three plumbing calls in 18 months, that plumbing line item needs to go up regardless of what the "expected life" spreadsheet says. I track every maintenance request by unit and system — when the same system shows up twice in six months, I bump the reserve for that component. The data from your own properties is more accurate than any rule of thumb.

Mike's point about climate is huge too. Sacramento summers destroy HVAC systems faster than the "15–20 year" estimates assume. I budget 12–15 years on AC compressors here, not 20.

Jim Johnson · Answer

I lean more toward the condition of the property than a fixed percentage. A newer property with updated systems has very different reserve needs than one with an aging roof, HVAC, plumbing, or exterior components.

We like to keep track of the expected lifespan of the major systems and adjust reserves as those replacement windows get closer. It's not an exact science, but it's a much more practical approach than treating every property the same.

I'd rather be a little overprepared than have a major repair become an unexpected financial strain.

Eduardo Cavasotti · Answer

Giuseppe, I would separate the reserve formula from the evidence behind it.

The roof and furnace each need a replacement target, but the thing that changes the target is the paper trail: inspection note, last invoice, vendor bid, photos, and every work order tied to that system.

I'm the founder of DoorVault, so this system age evidence file is something I work on every day, and the pattern that matters is when the same component keeps showing up before its expected replacement year.

For your 14 year roof, I would keep the monthly reserve heavy until a current bid or inspection note proves the remaining life.

Giuseppe Cavucci · Answer

That's a useful distinction I hadn't made explicitly, the formula tells you what to expect, the paper trail tells you how much to trust that expectation. A 14 year old roof with a recent inspection saying "5 more years, no issues" is a very different situation than the same age with no documentation at all, even though the formula spits out the same number for both.

Makes me think the reserve shouldn't just be age-based or even repair-frequency-based like Rachid mentioned, it should also widen or narrow based on how recently and how thoroughly something's actually been checked. No recent evidence probably means treat the estimate conservatively, like you said, keep it heavy until proven otherwise.Appreciate everyone stacking different angles on this thread, feels like between age, climate, repair history, and now documentation, there's a much fuller picture forming than any single method would give on its own.

Giuseppe Cavucci · Answer

The maintenance history point is the piece I was missing. I've been thinking about this purely as an age and lifespan problem, roof at X years, furnace at Y years, done. But you're right that a system throwing repeat calls is telling you something the age alone can't, it's basically real-time data overriding the theoretical estimate.That actually changes how I'd want to track this going forward: not just install date and expected life per system, but a running count of maintenance requests tied to each one, so if something spikes I bump its reserve manually instead of waiting for the age-based number to catch up.Combined with what Mike said about climate skewing the estimates too, it's starting to feel like the real formula is: base lifespan from standard tables, then adjust down for both climate and actual repair frequency. Appreciate you laying out the Sacramento numbers specifically, useful to see how much the AC compressor estimate shifts from the textbook 15-20 down to 12-15 in practice.

Janice Carter · Answer

@Giuseppe Cavucci I don't use a flat percentage. I look at the age and condition of the major systems and adjust my reserves accordingly. I also factor in the property's age and maintenance history. Newer homes typically require less while older properties usually need a larger reserve for unexpected repairs.

Giuseppe Cavucci · Answer

That lines up with what I'm seeing in this thread too, age and condition of the systems seems to matter a lot more than a flat percentage ever could. Curious if you track maintenance history formally somewhere, or if it's more of a general sense you build up over time owning the property.

Mike Fisher · Answer

Agreed that property value is the wrong variable, so I will add the operator's version of this rather than repeat the component math above.

The way I run it across a portfolio is two separate reserves, and I fund the CapEx one at acquisition, not just monthly. When I buy, the inspection gives me the age of the roof, furnace, water heater, and service panel. Anything already past two-thirds of its life gets pre-funded on day one, because on an older system the "remaining useful life" math is optimistic. A 22-year-old roof does not politely wait three more years to hit its number.

For the monthly maintenance side (the small, random stuff), I use rent, not value. Roughly 8 to 10 percent of collected rent on an older building, less on a newer one. That covers the faucets, disposals, and turn punch lists that actually do average out.

Two things specific to this climate skew the numbers, since I have invested and managed in the Chicago south and southwest suburbs since 1991:

1. Freeze-thaw punishes roofs, flashing, and concrete faster than the national tables assume, so I shorten roof and exterior life estimates by a few years versus a Sunbelt rule of thumb.

2. Furnaces get run hard from October to April. I reserve for heat as a when-not-if item on anything over about 15 years, because a January no-heat call is an emergency-rate repair plus a tenant-relations problem, not a scheduled replacement.

The part most owners skip: move the money to a separate account the day rent lands, and re-price the reserve every year. A reserve you can see in your operating balance gets spent. One you have to transfer out of savings stays a reserve.

What is the age spread on the heat across your units? In this climate that is usually the line item that decides the whole number.

Giuseppe Cavucci · Answer

The acquisition pre-funding point is something I hadn't fully separated out, that's a sharp distinction. Makes sense, a system already past two-thirds of its life shouldn't be treated the same as one that's brand new just because the monthly math says there's technically time left.To answer your question: furnace is about 9 years in, which by your logic still has some runway before it becomes a when-not-if item. Roof is the one closer to that line for me though, around 14 years into a 20-25 year expected life.The separate account point is one I've seen bite people specifically, myself included early on. A number on a spreadsheet is very easy to spend when it's sitting in the same balance as everything else. Moving it out the day rent lands is such a simple fix for something that causes a lot of damage when skipped.

Rami David · Answer

The percentage rules (1% of value, 50% rule, $X per door) are fine as a sanity

check, but they all fail the same way: they scale with the wrong variable. A

$400K house and a $200K house need the same $9K roof. Property value does not

drive your costs. Components do.

It works better to split the reserve in two, because the two halves behave

completely differently.

Maintenance is small, frequent, and roughly random. Leaky faucets, a failed

disposal, service calls, the small stuff between tenants. This genuinely does

average out to a percentage of rent, so a percentage rule is fine here.

CapEx is large, rare, and not random at all. Roof, HVAC, water heater, flooring,

exteriors. You are not guessing at these. You know roughly what each costs in

your market and roughly how much life each has left. Using a percentage rule for

something this predictable throws away information you already have.

So build the CapEx number bottom-up:

Replacement cost divided by remaining useful life, per component, per month.

A water heater at $1,400 installed, 10 year life, 4 years in, leaves 72 months.

That is about $19/month. Do that for the five or six components that actually

matter and add them up. The result is specific to that property rather than an

industry average, and it tells you something no percentage rule can: which

property is about to get expensive.

The uncomfortable part is that a house with a 22-year-old roof and an original

furnace produces an ugly monthly number. That is not the math being wrong. That

is the property telling you something you would otherwise discover all at once.

Two things that matter more than the exact figure:

Actually move the money. A reserve sitting in the same account as everything

else is not a reserve, it is a number you tell yourself. Separate account,

automatic transfer, same day rent lands.

Re-run it annually. Components age and replacement prices have moved a lot in a

few years. A number set in 2021 is not the number today.

What does the age spread on your roofs and HVAC look like? That usually drives

more of the answer than which percentage you land on.

How do you actually decide how much to set aside for maintenance each month?

Most Popular Reply