Sixty-eight percent of the budget blowups I've watched destroy general contractors over 20 years started in the dirt, not the building. Site work is typically 5-15% of total project cost on a clean suburban pad and can hit 25-30% on a sloped, wooded, rocky, or wet site — yet most estimators spend 80% of their takeoff time on the structure and maybe an afternoon on the earthwork. My shop lost $47,000 on a single retail pad in 2019 because we trusted a geotech report with four borings on six acres. I haven't made that mistake since, and this article is the checklist I built so I never would.
If you want the short version: price every line item you can see, carry 10-20% contingency on everything you can't, and bid unit prices on anything underground. Here's the long version, with numbers.
Why Site Work Blows Budgets: You're Bidding What You Can't See
Every other trade bids from drawings. Site work bids from a guess about what's under 6 to 20 feet of soil that nobody has fully inspected. That asymmetry is the whole problem, and it's why site work change orders average 8-12% of the earthwork contract value industry-wide, versus 2-4% for interior trades.
The underground unknowns that actually cost money
In rough order of how often they've burned me, and what they cost when they do:
- Rock. Ripping rock runs $15-$40 per cubic yard; blasting runs $40-$100+ per CY once you add seismic monitoring and permits. One unmapped granite shelf on a 10,000 CY cut can add $80,000 overnight.
- Unsuitable soils. Undercutting soft clay or organic material and importing structural fill costs $25-$45 per CY for the round trip — excavate, haul off, import, place, compact. A 2-foot undercut across a 30,000 SF building pad is roughly 2,200 CY, or $55,000-$99,000.
- Groundwater. Dewatering a wet excavation runs $5,000-$15,000 per month for well points and pumps, and it wrecks your compaction schedule on top of the direct cost.
- Buried debris. Old foundations, septic tanks, stumps from a 1970s clearing job. Demolition and disposal of surprise concrete runs $200-$400 per CY hauled.
- Utility conflicts. An unmarked 6-inch water main relocation costs $8,000-$25,000 and two weeks of schedule.
Soil borings are the cheapest insurance you'll ever buy
A geotechnical investigation with borings on a 100-foot to 150-foot grid costs $3,500-$8,000 per acre depending on depth and lab work. That's 0.2-0.5% of a typical commercial site package. My rule after the 2019 loss: one boring per 10,000 SF of building pad minimum, plus borings at every detention pond and any cut deeper than 5 feet. When an owner refuses to pay for adequate borings, I price the risk instead — my contingency on that line goes from 10% to 20%, and I say so in the bid letter. About half the time the owner suddenly finds the boring budget.
The Line Items: What Site Work Actually Costs in 2026
Here's my current pricing framework, calibrated against 2026 subcontractor quotes in the Southeast and mid-Atlantic. Your region will vary 15-25% either direction — West Coast and Northeast run high, the rural Midwest runs low.
| Line Item | Typical 2026 Range | Unit | Notes |
|---|---|---|---|
| Clearing and grubbing (light brush) | $1,000-$2,500 | per acre | Mowable scrub, few trees |
| Clearing and grubbing (heavy timber) | $3,000-$5,000+ | per acre | Add stump disposal $150-$300/stump for hardwoods over 24" |
| Topsoil strip and stockpile | $1.50-$3.50 | per CY | Assume 4-8" depth; verify with borings |
| Mass excavation (cut to fill on site) | $2.50-$6.00 | per CY | Machine-only moves under 1,000 ft |
| Import structural fill | $12-$25 | per CY placed | Haul distance is everything |
| Export/haul-off | $8-$18 | per CY | Add $50-$150/load tipping if contaminated |
| Rock ripping | $15-$40 | per CY | Blasting $40-$100+ |
| Fine grading building pad | $0.35-$0.75 | per SF | Laser or GPS grade control |
| Storm drainage (RCP 15"-36") | $45-$120 | per LF installed | Structures $2,500-$6,000 each |
| Sanitary sewer (8" PVC, 6-10 ft deep) | $40-$90 | per LF | Depth drives cost more than pipe size |
| Water main (8" DIP) | $55-$110 | per LF | Add fire hydrants $5,500-$8,000 each |
| Erosion control | $2,500-$6,000 | per acre | Silt fence, inlet protection, construction entrance |
| Compaction testing | $1,200-$2,500 | per week | Third-party lab, one tech |
Two things about this table. First, notice how wide the ranges are — a 2x spread on most lines. That's not sloppy estimating; that's site work. Second, the cheap lines kill you through quantity errors, not unit price errors. Being $0.50/CY wrong on 40,000 CY of mass excavation is a $20,000 miss.
Run your quantities through our free excavation calculator to sanity-check cut volumes, and use the grading and slope calculator to verify your fall across the pad before you trust the civil drawings.
Cut and Fill Math: Where Estimators Lose Real Money
Cut/fill analysis sounds simple — compare existing grades to proposed grades, subtract, done. The mistakes hide in the conversion factors, and they're worth real dollars.
Shrink and swell will eat 10-25% of your volume
Soil changes volume when you move it. Bank (undisturbed) soil swells 10-30% when excavated into a truck, then shrinks 10-25% below bank volume when compacted to 95% Standard Proctor. Practical translation: 10,000 bank CY of typical clay compacts down to roughly 8,500-9,000 CY of fill. If your site needs 10,000 CY of compacted fill and you assumed 1:1 from your cut, you're short 1,000-1,500 CY — at $15-$25/CY imported, that's a $15,000-$37,500 hole in your bid.
My defaults when the geotech report doesn't give me factors: 0.85 shrink for clays, 0.90 for sandy soils, and I never assume better than 0.92 without lab data.
Haul distance changes the unit price more than the machine does
A dozer push under 300 feet costs $2.50-$3.50/CY. The same dirt moved 1,500 feet by scraper or artic truck costs $4.50-$6.50/CY. Off-site haul adds $8-$18/CY depending on trip time. When I review a site plan, the first thing I mark up is the mass haul diagram: where's the cut, where's the fill, and how far apart are they? A "balanced site" with cut in the northeast corner and fill in the southwest corner isn't balanced in dollars.
Compaction testing failures cost more than the testing
Budget $1,200-$2,500 per week for a materials testing tech, and expect testing to run the full duration of earthwork. A failed proctor means re-work: rip, dry or moisture-condition, re-compact, re-test. Each failed lift on a building pad costs $3,000-$8,000 in machine time and delay. Cheap insurance: pay your operator a $500 bonus for a month of zero failed tests. Costs me $6,000 a year, saves me five figures.
Sizing the Contingency: 10% Is a Floor, Not a Target
Here's my contrarian take: the standard 10% site work contingency is a marketing number, not a risk number. It's what estimators carry because it's what owners will tolerate without questions. The actual risk on a given site is knowable, and it's often 15-20%.
I score every site on five factors, adding contingency for each:
- Geotech quality: Full report with adequate borings, +0%. Thin report or none, +5%.
- Rock indicated or suspected: +3-5%.
- Water table within 5 feet of deepest excavation: +3%.
- Previously developed site (demolition debris, old utilities, possible contamination): +3-5%.
- Cut/fill balance dependent on on-site material quality: +2-4%.
A clean greenfield with a good geotech report gets 10%. A redevelopment site with two borings and a creek on the property line gets 20%, and I'll walk from the job before I bid it at 10%. Over the last six years my shop's average site work overrun against estimate is 4.1% — which means the contingency absorbs it and the margin survives. The GCs I know running 10% flat across everything show great margins on paper and then give one bad job a year all of it back.
For the full framework on how contingency interacts with markup and overhead recovery, see our construction cost estimating guide — site work contingency stacks on top of, not instead of, your normal estimating buffers.
Unit-Price vs. Lump-Sum: Match the Contract to the Unknown
The bidding format matters as much as the numbers. My rule: lump-sum what you can measure, unit-price what you can't.
Where lump-sum works
Clearing, fine grading, erosion control, paving, and utilities shown on approved drawings — these are quantifiable from the plans within 5%, so lump-sum is fine and owners prefer it. Carry your normal 10-15% markup and move on.
Where unit prices save you
Rock, undercut/unsuitable soils, dewatering, and imported fill should be unit-priced with an estimated quantity and a contract clause that actual quantities govern. Structure it like a DOT bid: "Rock excavation: 500 CY estimated @ $32/CY; measured in place by survey." If you hit 2,000 CY, you're paid for 2,000 CY. If the owner insists on lump-sum for everything, that's fine — but the unknowns get priced at the 80th percentile instead of the mean, and my number goes up 8-12%. I tell them exactly that. Sophisticated owners take the unit prices every time because it's cheaper for them too; they only pay for rock that exists.
One more contract clause worth fighting for: a differing site conditions clause (the federal FAR 52.236-2 language is the model). Without it, in most states you own whatever you find. With it, materially different subsurface conditions become a change order. It's two paragraphs that are worth more than any single line item in your bid.
Check the whole stack before you submit
Before any site-heavy bid leaves my office, I run three checks: quantities through the excavation calculator, pad drainage through the grading and slope calculator, and the assembled package through the cost estimator to confirm the site work percentage lands in a sane band for the project type — 5-15% for flat commercial, 15-30% for hillside or heavy-utility work. When my number falls outside that band, one of us is wrong: me or the site. Worth finding out which before contract signing, not after.
Frequently Asked Questions
What percentage of total construction cost is site work?
Typically 5-15% on flat, previously graded commercial sites, and 15-30% on sloped, wooded, rocky, or infill sites with demolition. On raw land residential development, site work and infrastructure can reach 25-35% of total cost. If your estimate lands below 5% on anything but a pre-graded pad, re-check your quantities.
How much does land clearing cost per acre in 2026?
Light brush and scrub runs $1,000-$2,500 per acre; heavily wooded land runs $3,000-$5,000+ per acre, with hardwood stumps over 24 inches adding $150-$300 each for removal and disposal. Burning on site, where permitted, can cut disposal cost 30-50% versus grinding and hauling.
How much contingency should I carry for site work?
Carry 10% as a floor on clean greenfield sites with a solid geotechnical report, and scale up to 15-20% for sites with suspected rock, high groundwater, prior development, or thin soils data. Price the risk you can identify; contingency covers the risk you can't.
Is unit-price or lump-sum bidding better for earthwork?
Use lump-sum for quantifiable scope (clearing, grading, utilities per plan) and unit prices for subsurface unknowns (rock, undercut, imported fill, dewatering) with estimated quantities and field measurement. If forced to lump-sum unknowns, price them at pessimistic quantities — typically adding 8-12% to the site package.
What does compaction testing cost?
A third-party materials testing technician runs $1,200-$2,500 per week, with proctors at $150-$300 each and density tests at $25-$50 each. Budget testing for the full earthwork duration. The testing is cheap; the re-work from failed lifts ($3,000-$8,000 per failure) is what you're actually managing.
How do I estimate cut and fill volumes?
Compare existing topo to proposed grades on a grid (or with takeoff software), then apply shrink/swell factors: bank soil typically shrinks to 85-90% of its volume when compacted to 95% Proctor. A site that "balances" on paper at 1:1 is usually 10-15% short of fill in reality.
Your Action Item for This Week
Pull the last three site-heavy jobs your shop completed and compare final earthwork cost against your original estimate line by line. Find the single line item with the biggest miss — for most contractors it's undercut, import fill quantity, or rock. Then add one specific defense to your next bid: a boring-count minimum, a shrink factor, or a unit-price clause for that exact item. One fixed leak per quarter beats a new estimating system you'll never finish implementing.
