Cost-per-square-foot is the most quoted and most misused number in residential construction. A builder hears that the local market is running at $240 per SF and uses it to set a price; a lender hears the same number and uses it to underwrite a loan; an appraiser uses it to triangulate value. The number itself is fine. The problem is that every party in that chain is computing it differently, applying it to a different baseline, and treating it as a target rather than a sanity check. Used well, $/SF benchmarks from NAHB and RSMeans are the cheapest sanity check a builder has. Used poorly, they are the source of the worst budgeting mistakes on a residential build.
The two published benchmarks worth using
Two industry sources anchor most residential cost discussions in the United States. Both are useful. Neither is sufficient on its own, and both require translation before the numbers become operational.
The NAHB Cost of Constructing a Home survey is released biennially by the National Association of Home Builders. It reports total construction cost, lot cost, financing cost, marketing cost, sales commissions, overhead, and profit, broken into eight major construction categories (site work, foundations, framing, exterior finishes, major systems, interior finishes, final steps, and other). The 2024 edition reports an average construction cost of around $162 per SF for new single-family homes nationwide on an average finished area of 2,647 SF, with construction cost representing roughly 60% of the total sales price. The figures lag the market by roughly twelve months because of the survey methodology.
RSMeans residential cost data, published annually by Gordian, is the unit-cost reference most estimators already trust. It ships with assemblies (a complete framing system priced per SF of floor), unit costs (a 2x6 stud priced per linear foot), and crew productivity factors. It also includes city cost indexes for several hundred metropolitan areas, which is the standard way to translate a national figure to a local one. Where NAHB tells you what a typical house costs in aggregate, RSMeans tells you what each component should cost in your specific market.
Per-trade ranges in 2026 dollars
These are the per-trade ranges we see for residential single-family construction in the southeastern United States in 2026, normalized to finished SF and stripped of land, financing, and developer overhead. They blend NAHB averages, RSMeans unit data adjusted with city cost indexes, and in-house data from active projects. Treat them as the reasonable band, not the target.
| Trade | Typical $/SF | Share of construction cost | Notes |
|---|---|---|---|
| Site work and excavation | $8–14 | 4–6% | Highly site-dependent; rocky lots push high |
| Foundation | $12–18 | 5–8% | Slab on grade lower; full basement pushes high |
| Framing (labor + lumber) | $14–22 | 8–12% | Roof complexity is the main variable |
| Roofing (cover only) | $4–7 | 2–3% | Architectural shingle baseline |
| Exterior finishes | $10–18 | 5–8% | Brick or stone pushes high; vinyl pulls low |
| HVAC | $5–9 | 3–4% | Heat pump baseline; dual-fuel pushes high |
| Plumbing | $4–7 | 2–4% | Fixture count is the swing factor |
| Electrical | $4–8 | 2–4% | Smart-home and EV-ready add 15–25% |
| Insulation and drywall | $8–12 | 4–6% | Vaulted ceilings push high |
| Interior finishes (cabinets, counters, flooring, paint) | $30–55 | 18–25% | The single largest swing in any benchmark exercise |
| Final steps and cleanup | $3–6 | 1–2% | Punch list, final clean, landscape touch-up |
Two things stand out. The interior finishes line is wider than every other trade combined, because finish levels span a roughly 2x range between a clean spec build and a high-end custom. And the swing in site work, foundation, and exterior finishes can move the total by $20 per SF before any of the trades that get most of the attention come into play.
Regional adjustment factors
Every published $/SF figure is a national average that has to be translated to a local one before it is useful. RSMeans publishes city cost indexes for several hundred metropolitan areas; the index uses the national average of 100 as a baseline, with each city expressed as a percentage. A city cost index of 92 means the local market runs about 8% below the national average; 118 means about 18% above.
The cost index is itself broken into materials and labor, because materials prices are roughly national and labor rates are sharply regional. A high-cost market like the San Francisco Bay Area runs a labor index near 145 with a materials index near 110, which means labor-heavy trades (framing, finish carpentry, drywall) are proportionally more expensive than materials-heavy trades (roofing cover, insulation). A lower-cost market like East Tennessee runs a labor index near 78 with a materials index near 96, which compresses the labor swing and pulls labor-heavy trades closer to materials-heavy trades on the cost ladder.
Why $/SF is misleading without context
The single largest mistake in benchmark use is comparing a small spec build to a large custom build on the same $/SF basis. The two have completely different cost structures, and the per-SF number is not comparable across them.
A 1,784 SF spec build amortizes a fixed cost (a single kitchen, two and a half bathrooms, one HVAC system, one electrical service, one roof, one foundation) over 1,784 SF. A 4,500 SF custom build amortizes a similar fixed cost (still one kitchen, often three bathrooms, often two HVAC systems, one electrical service, one roof, one foundation) over 4,500 SF. The fixed cost per SF on the larger house is roughly 40% lower because the denominator is bigger. The custom house also carries variable costs the spec house does not, which is why the custom can come in at $310 per SF and the spec at $240 per SF on the same materials and labor pool. The numbers tell two different stories.
The other common error is including or excluding land, developer overhead, financing, and sales costs inconsistently. The NAHB headline figure of $162 per SF is construction cost only. The MLS-listed sales figure of $268 per SF includes lot, financing, marketing, sales commissions, and builder profit. The two numbers are forty-six percent apart on the same house, and any benchmark exercise that mixes them is comparing different things.
Backing out land and overhead
A clean benchmark comparison strips the budget down to construction cost only. Land cost, lot improvements outside the building footprint, developer overhead, financing carry, and sales commissions all come out before the per-SF calculation. What is left is the amount that flows to subs, materials, equipment, and direct project supervision. That is the figure NAHB reports, the figure RSMeans estimates, and the figure that is comparable across projects.
For a builder doing a sanity check on a pricing exercise, the order of operations is to compute total construction cost (sum of trade bids and direct costs), divide by finished SF (excluding garage, unfinished basement, covered porches, decks), and compare to a regionally adjusted benchmark. Anything within roughly 10% of the benchmark is in the normal band. Anything 15% above or below the band deserves a written explanation before the budget is locked.
The right way to use a benchmark
A benchmark is a sanity check, not a target. Used as a sanity check, the benchmark catches the line items that are wildly off (the framing budget that comes in at $9 per SF on a complex roof, the HVAC budget that comes in at $14 per SF on a simple spec). Used as a target, the benchmark drives behavior in two unhelpful directions. Either the builder pads the budget to hit the published average and prices themselves out of the local market, or the builder cuts corners to match a low published figure that did not include the finishes the buyer is actually paying for.
The right cadence is to compute the benchmark variance during the estimating phase, not after the budget is locked. Each trade line gets compared to the regionally adjusted range, and any line that falls outside the band gets a written note from the estimator explaining why. If the framing budget is high because the design has a hip roof and three dormers, the note is the audit trail the builder needs when the lender asks about the variance later.
Worked example: 926 Stratford
926 Stratford is a 1,784 SF single-story spec build in Sweetwater, Tennessee, contracted at $430,250. The lot was acquired for $58,000; the construction budget is $342,000. Direct construction divided by finished SF gives $191.70 per SF, which falls inside the southeast regional band of roughly $175–215 per SF for a 1,500–2,000 SF spec at this finish level.
Per-trade, the framing budget comes in at $32,500, which is $18.22 per SF and falls cleanly inside the $14–22 band. HVAC at $12,800 is $7.18 per SF and falls inside the $5–9 band. Electrical at $9,200 is $5.16 per SF and falls inside the $4–8 band. Interior finishes at $68,500 are $38.40 per SF and fall on the low end of the $30–55 band, consistent with a clean spec rather than a custom.
The one line that falls outside the band is exterior finishes at $32,000, or $17.94 per SF, which is on the high end of the $10–18 band. The estimator’s note explains that the elevation calls for a brick water table on three sides with painted cementitious panel above, which prices roughly 25% above a vinyl baseline. The note becomes part of the audit trail. When the lender asks about the exterior line during underwriting, the answer is on the same page as the budget.
Building a private benchmark
For a builder with five or more closed projects, the most accurate benchmark is the builder’s own portfolio. The published sources are useful as a backstop for new builders and a cross-check for unusual project types. The internal benchmark is built by tagging every closed project with finished SF, finish level, region, and roof complexity, and computing per-trade $/SF for each one. Five projects produces a tighter and more relevant range than any published source. Twenty projects produces a benchmark accurate enough to use as a pricing input for new builds.
The internal benchmark also lets the builder spot drift. If framing $/SF has trended upward across the last six builds, that is a labor market signal worth pricing into the next bid. If interior finishes have trended downward, that is a vendor relationship paying off and a candidate for sharing with the buyer as savings on a future build.
How BuilderGrid uses benchmarks
Each project carries a benchmark profile (region, finish level, finished SF, project type), and the validation engine compares each trade line to the regionally adjusted range during budget setup. Lines outside the band surface on the validation queue with the expected range and a request for a written explanation. As projects close, the actuals feed back into the builder’s private benchmark, and the bands tighten over time. The published NAHB and RSMeans figures anchor the system for new builders and as a cross-check; the builder’s own history takes over as the primary reference once five or more projects are closed.