What Cheap Commissioning Actually Costs
The numbers on commissioning cost and payback are good — and they are the reason the industry is being hollowed out. Here's the arithmetic, honestly, including the parts that cut against my argument.
The best available data on commissioning economics comes from Lawrence Berkeley National Laboratory. The 2020 study by Crowe, Mills and colleagues covered 1,482 North American buildings — the 643 from the earlier 2009 assessment plus 839 more — and it's the closest thing this industry has to a reliable evidence base.
The headline figures:
- Median commissioning cost, existing buildings: $0.26 per square foot. Down about a third from the 2009 study.
- Median commissioning cost, new construction: $0.82 per square foot. Down about half from the 2009 study.
- Median simple payback, existing buildings: 1.7 years, with an interquartile range of roughly 0.8 to 3.5 years.
- Median whole-building energy savings: 3–16% depending on market segment. Existing-building projects run through utility programs came in around 5%; monitoring-based commissioning programs around 9%; projects outside utility programs around 14%.
Take a moment on that cost trend, because it is the whole story of this website in two numbers.
The price of commissioning fell by half. Ask why.
There are three explanations for a 50 percent real decline in the cost of new-construction commissioning over roughly a decade, and they are not mutually exclusive.
Efficiency. The tools genuinely got better. Cloud-based issue tracking replaced spreadsheets and email. BAS trend extraction is vastly easier than it was in 2009. Automated fault detection can pre-screen problems that used to require manual hunting. Practitioners are more experienced and there are more of them. Some of this decline is real productivity, and I don't want to pretend otherwise.
Scale. More projects, more competition, more standardization of a process that used to be improvised each time. Some cost decline is the normal maturation of a service line.
Scope erosion. The service being priced in 2019 is not the service that was priced in 2008. Design review dropped. Seasonal testing dropped. The ten-month warranty review dropped. Sample rates fell from 100 percent of terminal units to 20 percent to "a representative sample" to whatever fits in the fee.
I cannot tell you from the data how the decline splits across those three. Nobody can, because the studies measure cost and savings, not scope. But here's the thing that makes me suspect scope erosion is a large share: savings did not go up. If the decline were mostly productivity, you'd expect the same work at lower cost, which means roughly constant savings per project. The 2020 medians are not dramatically better than 2009's. We are paying less and getting a similar or slightly lower result, which is what scope reduction looks like from the outside.
I'll hold that loosely. It's an inference, not a finding.
The number nobody publishes
Every study measures what commissioning saved. No study measures what bad commissioning cost, because the counterfactual is unobservable. The building that was signed off untested doesn't file a report. It just runs badly, forever, and its owner attributes the performance gap to the design or the weather or the tenants.
We can bound it, though, roughly. Suppose a 150,000 square foot office building. Drive-by Cx at $0.20/sf costs $30,000. Real Cx at $0.80/sf costs $120,000. The apparent saving is $90,000, and that number is extremely visible at bid time.
Now the other side. That building spends perhaps $1.75/sf/year on energy, so about $260,000 annually. The LBNL data suggests real commissioning yields on the order of 10 percent on new construction — call it $26,000 a year. Drive-by Cx captures little of that, because nothing was actually corrected. Over a ten-year hold, that is on the order of $260,000 of energy, undiscounted, against a $90,000 procurement saving.
And energy is the small part. The expensive failures are the ones that don't show up on a utility bill:
- Equipment cycling itself to death because staging was never tuned. A chiller compressor or a boiler reaching end of life at year twelve instead of year twenty-two is a six-figure event.
- Warranty claims that expire undiscovered. This is the one owners feel most sharply in hindsight. Manufacturer warranties run one to two years. A fault found in year four is the owner's problem; the identical fault found in month eight is the contractor's. The commissioning process is, functionally, the mechanism by which an owner converts latent defects into warranty claims while they're still someone else's money. Drive-by Cx forfeits that conversion entirely.
- Comfort complaints, which become tenant retention and lease renewal problems, and which nobody costs correctly because they arrive as a thousand small annoyances.
- Ventilation deficiencies. Minimum outdoor air that was never verified is an indoor air quality liability with a long tail and, increasingly, a legal one.
- Rework at year three, when someone finally hires a retro-commissioning firm to find what the new-construction CxA didn't. You are now paying twice, and the second time it's your money instead of the contractor's.
Add those and the $90,000 procurement saving becomes, plausibly, the most expensive line item on the project.
Where my argument is weakest
I should say the parts that cut against me.
Not every building needs the full process. A 12,000 square foot retail shell with two rooftop units and a packaged control system does not require the same rigor as a research building. Some of what I'd call scope erosion is appropriate right-sizing, and a good CxA should be willing to say so.
The savings estimates have selection bias. The LBNL data draws heavily on projects reported through utility incentive programs, which are projects that were pursued specifically because someone expected savings. The median building probably does worse.
Payback is not the only frame, and owners know it. A developer building to sell in thirty months does not capture a 1.7-year payback in any way that shows up in their exit price. Telling that developer they're being irrational is wrong; they're responding correctly to their actual incentives. The problem is that the person bearing the cost of bad commissioning is usually not the person buying it. That's a split-incentive problem, and split-incentive problems don't get solved by better arguments to the wrong party. They get solved by changing who decides — which usually means the lender, the insurer, or the eventual buyer's technical due diligence.
That last point deserves its own piece, and it'll get one. The short version: the most promising lever I know of isn't persuading developers. It's getting acquisition due diligence to treat the commissioning record as a document worth reading — because the moment a building's sale price turns on whether it was actually tested, the entire procurement calculus flips.
Sources: Crowe, E., Mills, E., Poeling, T., Curtin, C., Bjørnskov, D., Fischer, L., Granderson, J. (2020). Building commissioning costs and savings across three decades and 1,500 North American buildings. Energy and Buildings. Lawrence Berkeley National Laboratory.