On a building tender, the spec is the rulebook and the drawings are the picture, and reading both closely gets you most of the way to a defensible number. Site work breaks that model. The document that actually decides your cost exposure isn't written by the design team — it's a geotechnical consultant's investigation report, usually a PDF appended to the spec or dropped into a "Reference Documents" folder, written first for the structural engineer sizing footings, not for the estimator pricing earthwork. Read it the way most bids get built — skimmed once for bearing capacity, then filed — and you've priced the drawings' assumptions about the ground instead of the ground itself.
That's backwards, and specifically so: rock quantities, what "unsuitable" material actually triggers, dewatering, compaction, haul distances, and how much of that survives a winter schedule are all decided by data sitting in the geotech report, not the civil drawings or Division 31/32/33. An estimator who reads "the spec" in the normal sense — trade sections, general requirements — never opens the one document actually carrying this information, because it isn't filed as a spec section at all.
Rock excavation: the spec has the definition, the report has the actual line
Every rock excavation dispute starts with the contract's definition of rock — excavatability by machine class versus a blast-line survey produces very different quantities from an identical hole in the ground. But that definition only tells you how rock gets measured once you hit it; it says nothing about where it is. That's in the boring logs: auger refusal depths per borehole, and, where the investigation went deeper, RQD (rock quality designation) values describing how fractured it is. A grading plan's general notes might carry a single-line rock allowance sized off nothing more than the geotechnical summary's opening paragraph. The actual logs, read hole by hole, routinely show a rock surface pitching sharply across a site rather than sitting at one clean elevation — so the real quantity is a function of how far apart the boreholes were drilled, not the tidy allowance on the drawings.
Unsuitable material: what the word actually triggers
"Unsuitable material" sounds like a field judgment call, but the geotechnical report is usually where the judgment is defined in advance — organic content above a stated percentage, moisture content above optimum, a soil classification that pumps under proof-rolling. Removal, disposal and replacement is typically allowed for with a quantity allowance and a unit rate, but the report's own strata descriptions and moisture-content results are what tell you, before excavation starts, how much of the site is likely to actually meet that definition. Triggering "unsuitable" isn't just a removal-and-haul cost either — it usually means engineered fill brought in to replace it, a subgrade re-proof-roll, and sometimes a request for the geotechnical engineer to re-assess bearing capacity before founding can proceed, a schedule cost as real as the material cost.
Dewatering: a recommendation buried in narrative prose
Groundwater table depths are recorded per borehole, on the date each hole was drilled — a fact worth noting on its own, covered below. What often gets missed is that the report's dewatering guidance usually lives inside a narrative "Site Preparation and Earthworks Recommendations" section, several pages in, written as prose rather than a numbered clause — "wellpoint dewatering should be anticipated below elevation X" reads nothing like a spec requirement an estimator is trained to scan for. If the contract's own dewatering clause just says "in accordance with the geotechnical report," and nobody chases that reference back, the whole basis for the dewatering line in the bid was never read.
Compaction and testing: numbers the spec doesn't restate
Proctor percentages by material zone, lift thickness, and subgrade acceptance criteria (CBR values, proof-roll pass/fail thresholds) are frequently specified in detail in the report's recommendations and only referenced, not repeated, in Division 31 of the spec. A compaction clause reading "per the geotechnical report" is common, and it means the actual numbers governing your compaction plan and testing frequency sit in a document most bidders treat as reference material rather than as the spec itself.
Import/export and haul distance: reuse suitability decides the balance
A grading plan states a cut/fill balance assuming excavated material can be reused as engineered fill. Whether it actually can is answered by classification, moisture content and gradation data in the same report, not on the drawing. Material that fails reuse criteria turns an assumed balanced cut/fill into import fill plus disposal of everything excavated, and the haul distance to a stated disposal site, where one is identified, is often only in the report's appendix.
Winter conditions: the assumptions were probably made in a different season
Boreholes get drilled once, on a specific date, and the groundwater and moisture readings recorded that day are a snapshot, not a year-round guarantee. A report drilled in July says little about frost depth or groundwater behaviour on a schedule that runs earthwork through January, and neither the drawings nor the spec will flag that mismatch. If the schedule has any real chance of crossing freezing months, the report's own drilling date is worth checking before trusting its groundwater and compaction assumptions unchanged.
What a missed clause actually costs
A realistic version of how this plays out: a site package carries a rock excavation allowance of roughly 200 m³, sized off a single sentence in the civil drawings' general notes. The geotechnical report, four boreholes across the building pad, shows auger refusal ranging from 1.1 m to 3.4 m below grade — a rock surface pitching sharply across the site rather than sitting flat. Read against the actual boring logs rather than the drawing's summary line, the probable volume is closer to four or five times the stated allowance. The bidder who opened the report priced the exposure, or qualified the number against the allowance explicitly. The bidder who priced off the drawings' allowance carries a five- or six-figure gap the moment excavation starts — a fair unit rate never protects against a wrong quantity.
The twenty-minute pass that actually reads a geotech report
You don't need to read it cover to cover. Hunt these, in this order, and you've pulled out what actually changes the bid:
| Find | What it tells you |
|---|---|
| Borehole location plan | How many holes, how far apart, relative to your actual excavation footprint — sparse or uneven coverage is a risk signal on its own. |
| Boring logs (refusal depths) | The real rock line, borehole by borehole — not the narrative summary's single-line estimate. |
| Groundwater readings + drilling date | Actual table depth on the day drilled, and whether that date represents the season you'll actually be excavating in. |
| Soil classification + moisture content, by stratum | The real answer to what's reusable as fill and what's going to get called unsuitable. |
| Site Preparation / Earthworks Recommendations section | Dewatering method, compaction requirements by zone, subgrade acceptance criteria — the numbers the spec only references. |
| Disposal site / haul distance, if stated | Often only in the report's appendix, not the civil package. |
Then treat every one of those six as a line item, a qualification, or a call to the geotechnical engineer before the number leaves the office — not a fact you noted and moved past. A geotechnical report that was actually read cover to cover is rare enough on a site work tender that doing it is, by itself, a competitive advantage.