Commercial contractors are adopting AI tools faster than the broader construction industry, but most still cannot put a dollar figure on the return. A survey of 1,020 commercial mechanical, electrical and plumbing (MEP) owners and executives, commissioned by ServiceTitan and fielded by Thrive Analytics in July 2026, shows 62% of firms have piloted or deployed AI. Yet only 15% of AI users report a significant positive impact with clear ROI.
The survey, published 24 September on GlobeNewswire, captures a sector that has moved past the trial question and run straight into the evidence question. ServiceTitan's own disclaimer states it "provides no assurances (express or implied) with respect to the accuracy of the survey data." The findings are vendor-commissioned research about a market the vendor sells into - useful, directional, and not neutral.
The two numbers that define the moment
Thirty-three percent of firms are, in the report's phrasing, "actively using it or have embedded it across their businesses." As a technology priority, AI more than doubled year over year, rising from 15% in 2025 to 33% in 2026. The appetite is real and growing.
Then the hard part: among contractors using AI, 59% report a positive impact. But just 15% can point to a significant impact with clear ROI. Six in ten say it helped. One in seven has a number. That gap is not a verdict on the technology. It reflects what owners and general managers can demonstrate, not what is true. Many contractors never instrumented the before-state, so proving ROI on a dispatch tool requires a baseline that does not exist.
The honest summary: the sector has moved from "should we try this" to "can we show it worked," and has not yet answered the second question at scale.
Why the 62% cannot be compared to the 9%
In July, this newsroom analyzed two datasets pointing in opposite directions on AI adoption in construction. A McKinsey study concluded AI could automate 39% of nonphysical work. A global DEWALT survey found 9% of construction professionals used AI day to day. Placing ServiceTitan's 62% beside DEWALT's 9% and declaring one wrong would be a category error.
The DEWALT figure counted individual professionals across six countries using AI in their own daily work. ServiceTitan's counts firms - surveyed at owner, executive and GM level - that have piloted or deployed AI anywhere in the business. One measures a person's Tuesday. The other measures a company's procurement. A single pilot in one department registers as a yes in the second and as a no for almost everyone in the first.
The 33% "actively using or embedded" figure is the closest analogue to daily use, and it is still well above 9%. But it represents a different population - commercial MEP rather than construction broadly - and is reported by executives, not by the people doing the work. Directionally, the specialty-trade back office looks further along than industry-wide numbers suggest. That is a hypothesis this report supports, not a fact it establishes. For operations managers looking to build internal capability, AI for Operations Managers Courses address the measurement and deployment skills that the ROI gap exposes.
Cash flow, not AI, is the real headline
The finance findings explain why contractors are reaching for tools in the first place. Forty percent now rank improving cash flow among their top three business goals, up from 28% in 2025 - the largest year-over-year shift in the survey. Increasing net profit margins ranks first at 42%. Acquiring new customers trails both at 29%. A sector that puts margin and cash ahead of growth is a sector managing a squeeze.
That squeeze is visible in the payment cycle. Eighty-two percent of contractors send invoices within three days of completing work. Ninety-six percent wait at least 15 days to be paid, and 30% wait more than 30 days. The contractor's own process is fast. The money is slow anyway. That asymmetry is the structural fact of subcontracting, and invoicing software alone cannot close it - the delay sits on the payer's side of the transaction.
Two more pressures are quantified. Seventy-three percent say tariffs have materially impacted their business over the past year. And 46% now have more than half their commercial customers on service or maintenance agreements, up from 42% in 2025 - recurring revenue as a hedge against project volatility.
Alex Kablanian, ServiceTitan's senior vice president and general manager of commercial and construction, said in the release: "Contractors are looking closely at how they can improve cash flow, strengthen recurring revenue, and make their teams more productive. Technology, including AI, has an important role to play in helping contractors operate more efficiently and turn the information they already have into better outcomes for their businesses and customers."
Where contractors expect AI to land
Asked where AI will have the greatest impact, respondents named scheduling and dispatch (37%) and predictive maintenance (31%), with the report citing further opportunity in estimating, remote diagnostics and customer inquiries. That ordering matters. Much of this year's construction-AI coverage has clustered around preconstruction, drawing review, estimating and progress tracking. These contractors put dispatch and maintenance first - the operational middle of a service business, where the asset is a technician's day rather than a document.
One field finding underlines the point: 69% cite warranty coverage and agreement details as a top obstacle for technicians, alongside having the correct spare parts and access to equipment service histories. That is an information-routing problem, not a reasoning problem. It describes a technician standing in a mechanical room who cannot find out whether the part is covered. For service managers building dispatch workflows, AI Service Operations Courses target exactly this kind of retrieval and routing challenge.
Why this matters for management and operations
If the biggest self-reported constraint is getting known information to the person in the field, then the AI that pays for itself in this segment first is unglamorous retrieval rather than generation. Retrieval saves a phone call at a time. It does not produce a line on a P&L unless someone was already counting the phone calls. The firms that reach the 15% with clear ROI are likely the ones that measured the thing they automated before they automated it. Everyone else will keep reporting "positive impact" and struggling to say how much - not because nothing happened, but because nobody counted before. For operations leaders, the practical takeaway is that AI adoption without a measurement baseline produces sentiment, not evidence. Building that baseline before deployment is what separates the 15% from the 59%.
Your membership also unlocks:
