Building Age Doubles Replacement Work and Does Nothing to Service Revenue

The Test That Says Age Does Not Matter

Ask any mechanical contractor whether a 1970s office building costs more to look after than a 2015 one and he will say yes without pausing. It is one of the safest assumptions in the trade.

We went looking for it in real billing data, across 25,912 commercial buildings, and the first pass found nothing. Total annual revenue per building was close to flat across vintages. On the strength of that we nearly wrote age off as a driver entirely.

That would have been wrong, and the reason it would have been wrong is the interesting part.

Split The Revenue And The Effect Appears

A contractor's revenue from a building is not one thing. It is routine service, which is inspections and small repairs, and it is replacement, which is a rooftop unit or a compressor going in. Those two lines behave completely differently with age, and a total-revenue test averages one against the other until both disappear.

Cut the same office buildings by vintage and split the revenue at the job level:

Routine service does not move. Not slightly: the figure lands within twenty dollars of itself, and across all contractors the within-contractor ratio is 1.01, which is a coin flip.

Replacement work nearly doubles.

Healthcare is stronger still, moving from $10,191 to $23,500, a 2.31x step on the replacement line.

Why Routine Service Stays Flat

This is the part that surprised us, and the mechanism is obvious once stated.

Routine service is set by the equipment count and the contract, not by the equipment's age. A building with eight rooftop units gets eight units inspected whether they went in during 1994 or 2019. The visits are scheduled, the scope is fixed, and the invoice looks the same. Preventive maintenance visits run roughly $150 to $600 and seasonal start-ups $350 to $1,200 in most markets, and none of that pricing has an age term in it.

Replacement is the opposite. It is driven by what is due, and what is due is a function of how long things have been running.

The Mechanism Is Staggered Renewal, Not A Replacement Cycle

It is tempting to explain this as equipment reaching end of life at some age. Do not say that to a facilities professional: he knows his 1968 building has newer rooftops than your model predicts, because someone replaced them in 2011.

ASHRAE's service-life medians span 11 to 37 years across eight different system families. Rooftop units sit around 16 years, chillers longer, boilers longer again, and controls shorter. Nothing lines up. So past about 30 years the annual probability that something in the building is due converges upward and then flattens, which is exactly the shape the billing data shows including the flattening. A single replacement cycle would predict a peak we do not observe.

One more caution on that ASHRAE number: the median service life is measured on units still in service, so it understates realised life. Chillers show 18 years in service against 25 at replacement. The column is a survivor's average, not a lifespan.

An Independent Measurement Agrees On The Magnitude

We would not lean hard on a finding that only exists in our own data. It does not only exist there.

The Public Buildings Reform Board's March 2026 report on the federal portfolio covers 1,958 buildings and 280 million gross square feet. It names building age as the single greatest correlation factor with building condition, and its deferred-maintenance liability steps from $55 per gross square foot in buildings 1 to 20 years old to $130 in buildings 31 to 75 years old.

That is the same roughly 2.4x, on a different metric, in a different population, and it plateaus past 76 years the way ours does.

Be precise about what that figure is, though. Deferred maintenance is work that has NOT been done on a chronically underfunded portfolio, not revenue that was billed. It corroborates the age effect. It does not corroborate the dollar amount.

For contrast, the industry's familiar "2 to 4% of replacement value" maintenance rule traces back to a 1950s-derived convention and contains no age term at all.

Retail Runs The Other Way, And We Are Not Papering Over It

In retail the effect inverts. Newer retail bills MORE replacement work than older retail: $5,568 against $3,276, a ratio of 0.59.

Two explanations are plausible and we cannot separate them with this data. New retail fit-outs and tenant turnover drive equipment installs, so a recently repositioned space generates capital work that has nothing to do with age. Meanwhile an older strip center on a thin budget defers.

Because we cannot explain it, retail ships with no age adjustment at all. The same is true of four other segments. Age earns a multiplier in 3 of the 12 segments we publish, it inverts in 5, and in the rest it is not distinguishable from flat. A driver that works in a quarter of cases is a real finding; a driver applied everywhere because it worked once is a wrong answer waiting to happen.

What A Contractor Should Do With This

Three things follow, and the first two are targeting decisions you can make from public records.

Building age is a why-now signal, not a cost signal. A portfolio weighted toward 30+ year buildings is not a more expensive book to service, it is a book with more replacement work in it. That is a fundamentally different revenue mix: lumpier, higher-ticket, and far more sensitive to whether you can fund the material before the customer pays.

Stop pricing service contracts off vintage. If you have been adding a premium to older buildings on the assumption that routine work costs more there, the data says you are pricing a difference that does not exist. Price service off equipment count and scope.

Do not read a quiet year as a lost account. Replacement revenue is episodic by nature. A 40-year building that billed you $8,000 last year has not gone cold; it has a compressor that has not failed yet.

How This Was Measured

One mechanical contractor's billed revenue from one building, over a year. A building counts once it shows at least six months between its first and last completed job, annualised on that actual window rather than a calendar year. Each contractor holding five or more buildings of a type contributes one average, then those averages are averaged, so no single firm can carry a figure.

Jobs of $15,000 or more count as replacement and jobs under $5,000 as routine service. Work between the two sits in neither bucket on purpose: a low-end changeout and a compressor swap land in that gap, and calling the split exhaustive would be false.

The figure includes relationships that ended, not only the ones that lasted. Buildings whose contractor relationship ended bill about 36% less, so this is what an average account is worth including churn, rather than what a good one is worth.

These coefficients are HVAC and mechanical only. The ordering of building types by value is measurably trade-specific: against this ordering, electrical rank-correlates at rho 0.58 and refrigeration at 0.25, and for roofing, grocery moves from the most valuable type to near the bottom. Carrying these numbers into another trade would be visibly wrong rather than approximately right.

The full per-segment table, the size gradients, the held-out validation and the cells we measured and chose not to publish are all in the revenue per building by segment dataset, free to cite under CC BY 4.0. If you want the number for a specific building, the building revenue calculator will take an address.