Construction
This sector comprises establishments primarily engaged in constructing, repairing and renovating buildings and engineering works, and in subdividing and developing land. These establishments may operate on their own account or under contract to other establishments or property owners. They may produce complete projects or just parts of projects. Establishments often subcontract some or all of the work involved in a project, or work together in joint ventures. Establishments may produce new construction, or undertake repairs and renovations to existing structures. — Statistics Canada, NAICS 2022A
- Establishments · CanadaA
- 157,598
- Under 10 employeesA
- 81%
- Establishments · USA
- 800,651
- Employment · USA
- 7,361,847
- Payroll · USA
- $534.4B
Size and shape
How many businesses there are and how small they are. Fragmentation is the first thing an entrant — or anyone selling software into this industry — needs to know, and it is one of the few things that is actually measured.
Canadian establishments by number of employeesA
Of 157,598 Canadian establishments with employees, 81% have fewer than ten — an industry of very small operators.
Where they areA
Largest four provinces by establishment count. Establishments with employees only — sole operators with no payroll are not in this table, so in trades and personal services the true number of businesses is higher.
Measured, not forecast: the share of US establishments opening in one year that were still active later. It counts good operators and bad ones together, which is exactly why it is the honest answer to “what are the odds”. It is for the whole sector rather than this market, and the ten-year figure comes from an older cohort because no younger one has reached ten years.
Market screens and studies
Market-entry records filed along this branch of the hierarchy. A record at or beneath this code is about this industry; one above it is about something wider that contains it.
A builder finances land, permits and construction for eighteen months before a single dollar arrives, on a sale price set by a market that can move against them the whole time. Spec building is the purest version of that bet; custom building trades the price risk for client risk and a fixed-price contract signed before material costs are known. The one durable advantage is a land bank bought before the cycle, which is a capital position rather than a construction skill. The construction software serving this industry is screened separately at 2362 and 236.
The demand is unusually good — fibre builds, water main renewal and electrical distribution all carry funded multi-year programmes — and the work is awarded almost entirely through prequalified bid lists held by utilities and municipalities. Getting onto those lists requires bonding, safety record and prior work of the same type, which is the circular barrier facing every new entrant. Adjacent to the grid study at 221121, which found the same buyers and a different wedge.
Entitlement is the product: buy raw land, carry it through rezoning and servicing, sell lots. That process runs three to seven years with no revenue, funded by equity that cannot be recalled, against a municipal approval that can be refused at the end. It is the highest-variance market in this research and the one least suited to a first-time operator without a balance sheet.
Roads are rebuilt whether or not the economy is growing, the owner is a government that pays, and the record at Utility System Construction already covers the prequalification-and-bonding gate that guards all public civil work. This group has a second, harder gate behind that one. Public paving is awarded to the lowest compliant bid, and the lowest bid belongs to whoever has the cheapest hot-mix asphalt delivered to the job — which means whoever owns the asphalt plant, the quarry feeding it and the liquid-asphalt supply nearest the road. The consolidators say so in their filings. Construction Partners describes its strategy as vertical integration across hot-mix plants, paving, aggregates and liquid asphalt; in fiscal 2025 it grew revenue 54% to US$2.81B, only 8.4% of it organic, by buying five companies that brought 27 asphalt plants, four aggregate facilities and an asphalt terminal [A]. In Ontario, Colas paid C$913M for Miller McAsphalt — about C$1.3B of revenue at a 7% operating margin — to get road crews and the national bitumen terminal network together [A]. An entrant with pavers and a bond but no plant buys its mix from the firm it is bidding against. The size bands fit: 38% of the 2,092 establishments have twenty or more employees, and the US average is 33 per establishment. The reachable fringe — driveways, parking lots, sidewalk work — is real but is private paving, not this market.
The pre-screen called this the corner of heavy civil where a small contractor can still get in, and the business counts half agree: 62% of the 1,430 Canadian establishments have fewer than ten employees [A]. But the code holds two different businesses. The open end — parks, trails, drainage, small shoreline work — is open because nothing protects it: any contractor with an excavator can bid a municipal tender, and it meets the same prequalification and bonding wall that cut Utility System Construction (2371), without that record's funded programmes behind it. The end that is protected is on the water, and what protects it is the fleet. Great Lakes Dredge & Dock, the largest US dredger, reported record 2025 revenue of $888.3M and spent $143.9M on capital in the same year — sixteen cents of every revenue dollar — of which $69.1M went into a single vessel, the Acadia, and $32.3M into another [A]. In February 2026 it agreed to be bought by Saltchuk for $17.00 a share, a $1.5B transaction [A]: the price of a working fleet with its crews and its backlog, which fell from $1.2B to $763.2M of dredging work in the year [A]. Marine work is awarded to the contractor who already owns the barge, the crane and the dredge and can show them working. An entrant has to buy the plant before it can qualify to bid, then keep it busy on lumpy public tenders. The enterable end is undefended; the defended end is bought with steel.
The 'buy a boring business' thesis found this industry several years ago and repriced it: private-equity platforms and search funds now bid for the same HVAC and plumbing books, and multiples that were three to four times earnings are commonly five to eight. The buyer must also be the operator — technicians follow the owner, and licensing sits with individuals — so an absentee acquisition is the failure mode. The contractor software market at this code is a full study; this record is the operating business beneath it.
Drywall, painting, flooring and tiling are the trades with the lowest entry cost in construction, which is exactly the problem: a crew and a truck is the whole barrier, so pricing is competitive to the point of being cyclical labour arbitrage. The general contractor holds the relationship and re-tenders it. Scale helps only through the ability to staff several sites at once, which is a recruiting business rather than a trade one.
Iron is the business: an excavator, a truck and a float before the first invoice, financed at rates that assume utilisation nobody can promise a new entrant. Work arrives through general contractors who already have a preferred sub, and volume tracks housing starts — the cycle that is currently softening in the same markets where rents are falling.
Software serving this industry
The vertical software markets filed along the same branch — who sells to these businesses and who they would have to displace — and then the generic categories every business buys whatever it does.
The picture is now a commodity, and the money sits with whoever owns the job, the listing or the product catalogue. Turning a phone photo of a kitchen, facade or yard into restyled renderings is something a weekend project can do: RoomGPT's whole room-redesign app is public on GitHub under an MIT licence, built on open diffusion models steered by ControlNet (Apache-2.0) [A]. So the visualisation layer is being bundled into products that already hold the customer. Hover, a San Francisco company founded in 2011 that turns smartphone photos into measured 3D models of a house, added generative-AI redesign ('Instant Design') in February 2025 so an exterior contractor can show siding, roof and window options during the first visit [B]. Its backers include The Home Depot (Series B, 2018) and three insurers (Travelers, State Farm Ventures and Nationwide led a $60M Series D in 2020) [B]. Houzz, the remodeling marketplace that raised a $400M Series E in 2017, added AI 2D-to-3D floor plans to Houzz Pro in April 2025 [B]. On the listing side, Virtual Staging AI's own terms now name Zillow, Inc. as the provider of the service [C]. The listing portal owns the staging tool. What the tools really do, and what they do not. The claim that these apps 'compete with engineering firms' is wrong where it matters. The tools produce images, room layouts, material and colour options, and increasingly price ranges (Renovate AI shows materials-and-labour estimates; Hover's core product is measurement and estimating) [C]. They do not produce drawings an authority will accept for structural, mechanical or electrical work. California's engineering board states that civil, electrical and mechanical plans 'permitted or released for construction' must be prepared by a licensed engineer and 'signed and sealed (or stamped)' [A]. The same guide shows how much residential work sits outside that rule: unlicensed people may design conventional wood-frame single-family homes of up to two storeys plus basement, and non-structural interior alterations. Even then, the building official 'shall require' a licensed architect or engineer for any part that departs from conventional framing [A]. So the honest picture is that these tools displace the interior designer's concept board, the remodeler's sales sketch and the real-estate photo editor. They feed the architect and engineer rather than replace them, and in Canada the provinces (PEO, EGBC) set equivalent seal rules that this pass could not open (both returned 403). How this differs from the neighbours. 5413 is professional authoring software — Autodesk, Bentley, Trimble — whose file formats are the lock. 2362 is Procore-style construction management. 5312 and 5313 are transaction and appraisal workflow. This record is the consumer- and sales-facing front end that sits before any of them. The architecture-side entrants here (Snaptrude, Maket) are trying to cross from picture into plan, and the plan is where 5413's incumbents and the licensing rules are waiting. Defensibility decides it: an entrant's renderer is reproducible from open weights, and its distribution must be rented from Houzz, Hover, Zillow or a retailer that can ship the same feature.
Home builder and remodeler software is two markets that share a NAICS code, and both already have an owner. The custom builder and remodeler side — schedules, selections, change orders, client portal, estimates and job costing for firms building a few to a few dozen homes or renovations a year — is held by Buildertrend, which took its first institutional money from Bain Capital Tech Opportunities and HGGC in December 2020 [A for the round, terms undisclosed] and used it to buy its nearest rival, CoConstruct, in February 2021; the combined business said it served 23,000 clients and over 1.1 million users in 2020 [C via the release, carried by Pro Builder]. Behind it sits a crowded, self-funded or lightly funded field — JobTread (Dallas; says customers have sold and managed over $5B of jobs on it [C]), Builder Prime (Staten Island; Series A led by Blueprint Equity in 2023 [A for the round]), Contractor Foreman, BuildBook, Projul, UDA Technologies and Houzz Pro, which rides the Houzz homeowner marketplace. The production builder side — purchasing, scheduling and supplier orders for builders closing hundreds or thousands of homes — is consolidated into three suites: Hyphen Solutions (Dallas; BuildPro and SupplyPro) says 1 in 3 new homes in America is built on its software, 330,133 homes in 2025, with 565 builders and 18,000 suppliers on the network [C]; Constellation HomeBuilder Systems (NEWSTAR, BuildTopia; part of Toronto-listed Constellation Software) says forty percent of the Top 200 North American homebuilders use its software [C]; and ECI Software Solutions sells MarkSystems ERP for production and semi-custom builders alongside Lasso CRM, LotVue, AvidCX and AvidWarranty [C]. Incumbent vulnerability decides it. The production side is a network business — Hyphen's value is the 18,000 suppliers wired to its builders, and a builder's ERP holds every lot, option and purchase order — so neither switching nor a new entrant is cheap. The custom/remodel side is not locked in the same way, but it is where the money is already competing: Buildertrend is private-equity-backed and acquisitive, JobTread and a dozen cheaper tools undercut it, and Procore's 10-K says its core customers span the residential and non-residential segments [A]. The buyer base is real but cyclical: US private housing starts ran at a 1,221,000 seasonally adjusted annual rate in September 2026, 7.4% below a year earlier, with single-family starts at 859,000 [A]; Canadian starts were 248,896 (SAAR) in September 2026 against 280,669 a year before [A]. Builder software is bought and cut with that cycle. How this differs from its neighbours: 2361-residential-building-construction is the builder as a business (capital intensity decides it) and names no software; 2362-nonresidential-building-construction is commercial GC software (Procore, Autodesk, Oracle, Trimble, Sage) and is not re-researched here; 238-fsm-field-service-management is dispatch for service trades (ServiceTitan, Jobber, Housecall Pro), which overlaps only at the remodeler/home-improvement edge where Builder Prime and Joist sit. Bridgit (Waterloo) was checked and sells workforce planning to general and specialty contractors, not to home builders, so it is listed only as a startup row.
The most-cited vertical SaaS success story, and therefore the most thoroughly contested. Procore plus Autodesk have covered general contractor workflow end to end; the remaining gaps sit in specialty-trade workflows, which is where 238220 and 238160 lead instead.
The switching cost is the estimating database, not the software — a contractor's historical production rates live inside HCSS and are the asset that wins bids. Trimble and Kiewit's InEight have the rest. The building-construction record at 2362 faced Procore; this one faces a data moat that is older and harder to copy. Nothing is disclosed here. Every vendor named on this record is private, or sits inside a parent that does not break the line out, so no revenue floor can be built and the market size is genuinely unknown rather than estimated.
FSM is the horizontal dispatch layer under every business that sends workers to a site — booking the job, scheduling and routing the crew, the work order on the technician's phone, the quote, the invoice and the payment. It spans the specialty trades (the 238 anchor), appliance and equipment repair, cleaning, pest and pool routes, security patrol, land-survey and GIS field crews, and utility and telecom field forces. The neighbouring records — roofing (238160), electrical (238210), mechanical (238220), cleaning (561722), pest (561710), pool (561799), security guard (561612) and geospatial (541370) — are the vertical specialisations of this same layer; this record is the layer itself, and it does not repeat their vendor work. Incumbent vulnerability decides it: there is no weak incumbent anywhere in the stack. At the SMB and mid-market trades end, ServiceTitan (Nasdaq: TTAN) reported FY2026 revenue (year ended 2026-01-31) of $961.0M, up 24%, platform revenue of $925.4M, $82.1B of gross transaction volume invoiced through it, about 10,800 active customers, net dollar retention above 110% and gross dollar retention above 95%, and passed a $1B annualised run rate [A]. That is a leader still compounding, not one coasting. At the enterprise end, Salesforce Field Service and ServiceNow FSM ride inside CRM and ITSM estates the buyer already owns [C], IFS (which also owns WorkWave) passed €1B of ARR and was valued at over €15B in April 2025 [B], and PTC paid about $1.46B for ServiceMax in January 2023 [B]. Between those ends the challengers are funded, not starved: Jobber (Edmonton) $176M [B], Housecall Pro $125M in one 2022 round [B], simPRO over $350M from K1 [B], BuildOps $127M at a $1B valuation in March 2025 [B], ServiceTrade $85M [B], Workiz about $60M [B]. The private-equity roll-ups — EverCommerce (Service Fusion, Kickserv), Xplor (FieldEdge), IFS/WorkWave — have already bought most of the independent long tail. Demand is not the problem; the price of reaching the buyer is. Every owner-operator plumber, cleaner and patrol company is already being sold to by a well-funded vendor with a payments line attached, so an entrant pays the same paid-search and trade-show toll without the installed base that makes the payments margin work. The only defensible positions are vertical — a workflow the generalists do badly (commercial service agreements and inspections, which is exactly where ServiceTrade and BuildOps went; regulated reporting; survey and utility crews tied to GIS) — and those belong on the vertical records, not here. The 238 anchor is navigational: the buyer is any business with a dispatcher.
Two funded incumbents plus aerial-measurement data licensing (EagleView) as a barrier. Insurance-claim workflow is the distinctive element and it is already the incumbents' core feature. Nothing is disclosed here. Every vendor named on this record is private, or sits inside a parent that does not break the line out, so no revenue floor can be built and the market size is genuinely unknown rather than estimated.
Structurally identical to the mechanical trades study and served by the same incumbent. Rather than duplicate it, the service-agreement wedge identified in 238220 should be tested across both trades at once — electrical contractors run the same maintenance-agreement model. Sourced update: ServiceTitan closed FY2026 at $961M, +24%. Electrical contractors are served by the same platform as the mechanical trades studied at 238220, so this market has no incumbent of its own to attack.
The category leader has priced itself out of the bottom two-thirds of its own market — but >95% retention means the play is selling beside it, never instead of it.
Finishing-trade software is three markets under one NAICS code, and the one with real lock-in has just been consolidated. The flooring dealer and installer side — sales orders, roll and cut inventory, installer calendars, job costing and accounting for carpet and flooring stores and flooring contractors — is now mostly in two hands. Cyncly (formerly Compusoft + 2020, backed by TA Associates) bought RFMS, a flooring ERP, CRM and measuring vendor with more than 35 years in the trade, in July 2022 [B], and agreed in December 2024 to buy Broadlume, which had already bought RollMaster (the flooring ERP that started as a family distributor's system in the late 1970s) and the flooring marketing agency FloorForce [B]; the combined unit trades as Cyncly Flooring and says Broadlume served thousands of dealers and manufacturers across North America [C]. rfms.com now redirects to Cyncly. The other consolidator is Canadian: Roomvo (Leap Tools Inc., Toronto), whose room visualiser says it sits on 7,000+ retailer websites [C], bought QFloors (South Jordan, Utah) in January 2025 [C]. Behind them, Measure Square (founded 2001 in Pasadena) sells flooring takeoff and estimating and says it has 4,500+ customers [C]. The painting side has no single owner: PaintScout (Calgary; says $4B+ of painting work has been sold through it [C]) leads on estimating, DripJobs (Ocala, Florida; 2,500+ painting and home-service contractors [C]) on follow-up CRM, Estimate Rocket (Newburyport, Massachusetts) on estimating and production, and the horizontal field-service tools — Jobber (Edmonton) and Housecall Pro (Denver; 45,000+ businesses [C]) — take the small crews that never buy a trade tool; they are researched at 238-fsm-field-service-management. The commercial drywall, acoustical and painting subcontractor side estimates off plans in ConstructConnect's On-Screen Takeoff and PlanSwift (41,000+ estimators claimed for PlanSwift [C]) or STACK (Cincinnati; $7M from Level Equity in 2018, doubled in 2022 [B]). Incumbent vulnerability decides it. The flooring ERP holds a dealer's inventory, purchase orders, installer schedules and books, so it is the stickiest software here, and it has just been bought by two consolidators in under three years; painting is a crowded, cheap CRM-and-estimate fight beneath horizontal tools with $100M+ of funding; and plan takeoff is owned by the commercial takeoff vendors. The buyer base is real but fragmented: 26,183 Canadian building-finishing establishments with employees (Statistics Canada, December 2023), 70% with one to four employees, and 127,033 US establishments with 847,233 employees (County Business Patterns, 2022) [A] — the small crew that buys a $50-a-month app, not an ERP. How this differs from its neighbours: 2383-building-finishing-contracting is the contractor as a business (defensibility decides it, no software); 238-fsm-field-service-management is the horizontal dispatch layer (Jobber, Housecall Pro, ServiceTitan) and is not re-researched here; 238160-roofing-contractor-software is insurance-claim and aerial-measurement software for roofers; 2361-home-builder-and-remodeler-software is builder and remodeler project management (Buildertrend, Builder Prime), which overlaps only where a remodeler also sells flooring or paint. A Canadian wedge is not open: two of the category's own leaders, PaintScout and Roomvo, are already Canadian.
And what every business buys · 33 generic categories
Sold to every industry rather than this one, so they are filed against the software industry's own code. The same few vendors recur across most of them.
Companies in this industry · 107
Every company this research names that is filed here or beneath — the operators, and the vendors that sell to them — largest disclosed revenue first. The rank is within the company’s own six-digit industry.
| Company | Filed under | Revenue | Rank |
|---|---|---|---|
| CRHPrivate | Highway, street and bridge construction2373 | $37.4B | 1/4 |
| D.R. HortonNYSE:DHI | Residential building construction2361 | $34.3B | 1/30 |
| Quanta ServicesPrivate | Utility system construction2371 | $28.5B | 1/3 |
| EmcorNYSE:EME | Plumbing, heating and air-conditioning contractors238220 | $17.0B | 1/4 |
| MastecNYSE:MTZ | Utility system construction2371 | $14.3B | 2/3 |
| Comfort Systems USANYSE:FIX | Plumbing, heating and air-conditioning contractors238220 | $9.1B | 2/4 |
| Dycom IndustriesNYSE:DY | Utility system construction2371 | $5.5B | 3/3 |
| Construction PartnersPrivate | Highway, street and bridge construction2373 | $2.8B | 2/4 |
| Forestar GroupNYSE:FOR | Land subdivision2372 | $1.7B | 1/4 |
| ServiceTitanPrivate | Electrical contractors and other wiring installation contractors238210 | $961M | 1/3 |
| Great Lakes Dredge & DockDelisted | Other heavy and civil engineering construction2379 | $888M | 1/4 |
| Badger Infrastructure SolutionsPrivate | Site preparation contractors238910 | $832M | 1/7 |
| JobberPrivate | Building finishing contractors2383 | — | 1/16 |
| Housecall ProPrivate | Building finishing contractors2383 | — | 2/16 |
| ECI Software SolutionsPrivate | Residential building construction2361 | — | 2/30 |
And 92 more on the companies page.
Who works here
The occupations employed in Construction, most concentrated in it first. The share is measured against the whole sector, not this industry — the published cross-tabulation stops there.
Tagged to this industry
Concentrated in this sectorA
These jobs are mostly done here. An operator in this industry is competing for them against others in the same industry, not against the whole economy.
And the jobs every business has
Found across at least fourteen of the twenty sectors. But note the shape of this industry: 81% of establishments have fewer than ten employees, and at that size most of these roles are one person wearing several hats, or bought in from outside.
Inside this industry
3 rows sit directly beneath 23, and 70 in all once every level is counted. Each has a base report of its own.