Dario Compain used to lose half a day every time a customer needed a load calculation. His techs measured the house by hand, drove back to the office, punched numbers into software, and prayed nobody transposed a window dimension. Then his crews started carrying iPads, and now the tech walks the house waving the tablet while LiDAR maps every room, handing the homeowner a printed, ACCA-certified Manual J before the coffee goes cold. Fifteen minutes, door to quote.
Compain runs IAQ Medics, a shop in the DC metro area, and his story is the one ServiceTitan told when it announced it was buying Conduit Tech, the startup behind that iPad trick, in September 2025. Conduit was founded in 2022 by Marisa Reddy and Shelby Breger with a simple observation: the most skipped engineering step in residential HVAC now fits in a tablet, so there is no reason left to skip it. Good. Because the skipping has been expensive.
The most common math error in homebuilding
Walk onto almost any residential job site in America and ask how the AC got sized. The answer, more often than anyone admits, is a rule of thumb: about 500 square feet per ton, with no measurement of the windows, no accounting for which way the house faces, no blower door number, just square footage divided by a number somebody's boss used in 1994.
Building science journalist Martin Holladay put it bluntly years ago: sizing by rule of thumb almost always produces oversized equipment. And oversized is not a victimless rounding error. An air conditioner twice as big as the load runs in short bursts, shuts off before it wrings moisture out of the air, and leaves the house clammy. Homeowners then crank the thermostat below comfort to feel dry, which grows mold around the air handler. The compressor takes a beating from constant starting and stopping, and you paid extra for the privilege.
How bad is it out there? Energy rater Allison Bailes checked HERS files on Energy Star homes, houses that were supposed to be the good ones, and found air conditioners sized at one ton per 529 square feet, and 544, and 781, with his verdict on the first two running to a single word: terrible. His benchmark for a genuinely high-performance home is 1,000 square feet per ton or better, so read that again and let it land: production builders chasing efficiency certifications were installing roughly double the cooling the houses needed.
The math nobody ran
Methodology up front, because I am about to name a dollar figure and you should see the machinery. To see the machinery, take a 2,400 square foot new build and run both methods on it. Rule of thumb at 500 square feet per ton: 2,400 divided by 500 is 4.8, so the bid says 5 tons. Now run the actual load calculation on a well-built envelope at Bailes' 1,000-sq-ft-per-ton benchmark: 2,400 divided by 1,000 is 2.4, so a 2.5-ton system. Same house. Two answers. The wrong one is twice the right one.
ACCA Manual S, the equipment selection standard, says total cooling capacity should land between 95 and 115 percent of the Manual J load, and NREL's sizing guidance holds heat pumps in cooling-dominated climates to no more than 15 percent over. A 5-ton box on a 2.5-ton load is not 15 percent over. It is 100 percent over, sailing past the standard by a margin that would get an engineer laughed out of a plan check.
Now the money. A standalone Manual J from a third party runs $200 to $500 for a typical house, and Montgomery County's electrification program notes it is usually less than the cost of going one size larger on the equipment, and going one size larger is the polite version. Our 2,400-square-foot example is four half-ton steps oversized, and each phantom ton installed runs roughly $1,500 to $2,500 depending on efficiency tier and market. Call it $3,000 to $5,000 of equipment the house will never need, plus ductwork sized for airflow that will never happen. A $350 load calc against a $4,000 midpoint mistake pays for itself eleven times over before the first utility bill shows up.
One honest caveat on my own arithmetic: the 1,000-sq-ft-per-ton benchmark describes high-performance envelopes. A code-minimum build might land closer to 700 or 800 square feet per ton on a real Manual J, which still leaves the 5-ton rule-of-thumb unit 40 to 60 percent oversized. The conclusion survives even when you steelman the other side.
Why the calc kept getting skipped
If the math is this lopsided, why did contractors keep guessing? Because the load calc was unpaid consulting, and ACHR News surveyed contractors years ago with bleak answers: techs doing Manual Js before the close watched customers take the paperwork and shop it to cheaper bidders. One contractor said he had zero takers when he offered the calc for $300. Another called it "an option on all of our replacement work, but rarely selected." The industry had built an equilibrium where doing the engineering lost you the job, so nobody did the engineering.
That equilibrium just broke: IAQ Medics went from hours or days per calc to fifteen minutes on site, and Compain says the speed closed sales instead of losing them: the homeowner watches the scan happen, sees their actual house turned into numbers, and the bid stops being a guess. Fairfax County accepted the output for permitting without friction. When the calc costs fifteen minutes instead of half a day, "we don't have time for that" stops being an explanation and starts being a confession, and any bid that still arrives without a load calc attached is a contractor telling you, in writing, exactly how much your money is worth to them.
The case against, at full strength
Now the part the press release skipped, starting with the uncomfortable one: speed is not accuracy. A fifteen-minute LiDAR scan is only as good as its inputs, and the inputs that matter most are the ones the iPad cannot see: actual insulation levels inside closed walls, real air leakage without a blower door test, the ductwork baking in a 130-degree vented attic. As one contractor put it in a detailed 2026 teardown, most software-driven calcs quietly fall apart on the latent load, the moisture-removal half of the job, because they match equipment at laboratory rating conditions instead of the 95-degree reality outside. A fast wrong answer is still a wrong answer. It just arrives sooner.
Second, and this is the one the industry should sit with: NREL's own simulation research found that oversizing had very little impact on overall energy use in hot-humid climates. The honest case against oversizing is comfort, humidity control, equipment longevity, and upfront cost, not the power bill, so anyone selling right-sizing primarily as energy savings is either misinformed or selling something, which is worth remembering the next time a glossy utility brochure promises double-digit savings from a swap the lab coats say barely moves the meter.
Third, equipment comes in half-ton increments and houses do not, so some oversizing is unavoidable, and modern variable-speed inverter systems tolerate it far better than the single-stage boxes the horror stories are about. A two-stage or modulating system sized a half-ton proud is a different animal than a single-stage 5-ton hammer on a 2.5-ton load. Precision theater helps nobody.
Fourth, follow the acquirer's incentives, because ServiceTitan framed the Conduit acquisition as sales enablement: scan the house, render the install in AR, close the deal on site. A tool that makes load calcs instant also makes oversized quotes instant, prettier, and harder for a homeowner to argue with because now there is a 3D model attached. The iPad does not care whether the number it prints is honest. Somebody has to.
What this analysis could not prove
Limits, stated plainly. The $1,500 to $2,500 per-ton installed delta is an industry heuristic, not a quoted figure, and it moves with SEER2 tier, brand, and local labor markets. I could not independently verify Conduit Tech's "ACCA-certified in under 15 minutes" claim beyond ServiceTitan's announcement and a single customer case study, and the acquisition was still pending close at announcement with integration benefits firmly in the forward-looking category. The NREL oversizing findings are EnergyPlus simulations, not field measurements. And my 11x payback figure assumes the homeowner actually acts on the calc instead of buying the bigger box anyway because the neighbor's house has one.
What to do with this
If you are building or replacing a system, demand a room-by-room Manual J in writing before you sign anything. It costs $200 to $500 standalone, and many contractors now bundle it into the bid. A contractor who says they do not need one is telling you they size by guessing. Believe them. That is a red flag, not a reassurance, and the online HVAC press has been saying so for years.
Run the thirty-second sanity check on every bid. Divide the conditioned square footage by the quoted tonnage. On a new build, anything under roughly 700 square feet per ton deserves a written explanation, and anything near 500 means you are buying the rule of thumb, not the engineering. Then ask for the Manual S equipment selection at design conditions, not just the Manual J total, because the latent capacity is where humid-climate systems go to die quietly.
If you run a shop, the old excuse is gone. Your techs already carry the sensor. Price the fifteen-minute calc into the bid instead of eating half a day, and watch what happens when the homeowner sees their house turned into math in front of them. The contractors who figure this out first will not just size systems correctly. They will take work from the ones still dividing by 500. Count on it.