Two methods, one bid form
Unit-cost estimating says a line item is worth quantity × unit price, where the unit price is pulled from a reference database — typically RSMeans Building Construction Costs, Gordian's facility cost data, or a firm's internal historical unit file. Assembly estimating says a line item is worth quantity × (material + labor-hours × composite-crew-rate + equipment), where the labor-hours come from a published trade standard: MCAA for mechanical, NECA for electrical, SMACNA for sheet metal, PCI for precast, NWFA for wood flooring. The first method is fast. The second is defensible.
The difference matters the moment you have to justify a number. RSMeans publishes national average unit prices for roughly 30,000 line items, updated annually, with city cost indices (CCI) for about 731 U.S. and Canadian locations. It is an excellent conceptual estimating tool at 15-25% accuracy on schematic documents. It is a dangerous hard-bid tool on a $5M mechanical package when your shop has a crew productivity 20 points below the union baseline that RSMeans assumes.
What RSMeans actually includes — and what it quietly does not
Open any RSMeans line item and you get bare-cost material, bare-cost labor, bare-cost equipment, total bare cost, and total including overhead and profit. The labor number is built from a crew definition (e.g., Crew B-5 = 1 labor foreman, 4 laborers, 1 crane operator, 1 dozer) at a blended hourly rate, multiplied by the daily output for the task. The productivity assumption is published — 1.2 CY/crew-day for hand excavation in sandy clay, for example — but it is a national average for "typical" conditions.
What the line item does not include: your job's specific mobilization, your escalation across the schedule, your general conditions loading, sales tax depending on jurisdiction, any productivity derating for overhead work, congestion, repetitive vs. non-repetitive conditions, shift premiums, or BIM coordination burden. For a 8,000 SF tenant improvement where all of those factors are roughly average, RSMeans with the right CCI will land the estimate inside 10%. For a 400,000 SF data center with dense rack work, restricted access, and 24/7 schedule, RSMeans alone will be 15-30% light on labor before you finish the first floor.
MCAA and NECA labor units: what trade assemblies look like
The Mechanical Contractors Association of America (MCAA) publishes the Labor Estimating Manual, which lists labor-hour units per piece for every major mechanical task: a 4" Schedule 40 black steel threaded joint is 0.45 manhours at 100% productivity, a 2" copper Type L soldered joint is 0.22 manhours, a 20-ton rooftop unit set-in-place is 18 manhours for the millwright plus 6 for the pipefitter. The National Electrical Contractors Association (NECA) publishes the Manual of Labor Units (MLU), with columns for Column 1 (favorable conditions), Column 2 (normal), and Column 3 (difficult) — a 3/4" EMT run with two couplings and a connector is 3.5 manhours per 100 LF in Column 2, 4.9 in Column 3.
Assembly estimating takes those labor units, multiplies by the piece count from the takeoff, applies a productivity factor (the "PF" — 1.00 is standard, 0.85 is an aggressive shop on repetitive work, 1.25 is a shop working above the ceiling on a congested retrofit), and multiplies by the composite crew rate. The composite crew rate itself is a buildup: base wage + fringe + payroll burden + small tools + a share of supervision. For a union Local 597 steamfitter in Chicago at roughly $52/hour base wage in 2025, the loaded composite crew rate lands around $108-$118/hour depending on the burden detail.
Applying RSMeans unit prices to a $12M hospital mechanical package without adjusting productivity for overhead work, infection-control barriers, and above-ceiling coordination. The bare labor in RSMeans assumes typical commercial conditions — a healthcare renovation routinely runs at 1.30-1.45 PF, meaning labor is 30-45% higher than the database hour count. Estimators who skip this adjustment lose the job on margin or win it and bleed out.
The accuracy delta: small jobs vs. large jobs
Published ASPE accuracy ranges for estimate classes (AACE 17R-97 Class 5 through Class 1) provide the benchmark. Class 5 order-of-magnitude estimates are -30% to +50%. Class 1 definitive estimates are -5% to +10%. Unit-cost estimating with a good database and correct CCI typically produces a Class 3 to Class 2 estimate (-10% to +20%). Assembly estimating with verified labor units and a calibrated PF produces Class 1 (-5% to +10%) on scopes you understand.
On small jobs below $250K, the accuracy spread of unit-cost estimating is narrower in absolute dollars. A 15% high read on a $180K tenant finish is $27K — inside the GC's contingency. On a $8.5M mechanical package, a 15% error is $1.275M, and the subcontractor writing that bid is either leaving money on the table or going bankrupt. The assembly method, with trade-standard labor units and a shop-specific PF, compresses that spread to the single-digit percentages that hard-bid work requires.
Side-by-side: 1,200 LF of 4" Schedule 40 carbon steel pipe
Take a real scope — 1,200 LF of 4" Schedule 40 ERW black steel pipe, threaded joints, on pipe racks at 14 feet, Chicago jurisdiction, union labor, January 2026 pricing. Both methods below target the installed cost.
| Line | Unit-Cost Method | Assembly Method |
|---|---|---|
| Source | RSMeans 22 11 13.74 1600, CCI Chicago 1.14 | MCAA LEM + Local 597 composite rate |
| Quantity | 1,200 LF | 1,200 LF = 120 joints @ 10' nominal |
| Material unit | $18.40/LF bare | $17.85/LF + fittings/hangers line |
| Labor hours | built into $22.15/LF bare labor | 0.45 mh/joint × 120 = 54 mh + 0.03 mh/LF support × 1,200 = 36 mh = 90 mh |
| Crew rate | Crew Q-6 blended @ $119/hr | Composite steamfitter $112/hr loaded |
| Productivity factor | not applied | 1.15 (elevated work, above 12 ft) |
| Material subtotal | $22,080 × 1.14 = $25,171 | $21,420 (+ $3,850 fittings/hangers) |
| Labor subtotal | $26,580 × 1.14 = $30,301 | 90 × 1.15 × $112 = $11,592 |
| Subtotal before OH&P | $55,472 | $36,862 |
| Installed estimate | $66,567 | $44,234 |
The $22,333 gap is not a database error — it is the fact that RSMeans pipe labor is built around an assembly that includes fittings, hangers, and tests baked in, while an assembly buildup separates them cleanly. On a bid you control, the assembly number is closer to truth because you can point at each hour. On a concept estimate you will revise three times before GMP, the RSMeans number is defensible as "database Class 3" and nobody asks harder questions.
When unit-cost actually wins
Unit-cost estimating beats assembly estimating when the project is below roughly $500K, when documents are schematic, when the scope is standard commercial repetition, or when you have zero productivity history and the MCAA/NECA manual number is the best guess you have. It also wins on owner's-rep and conceptual work where speed matters more than dollar accuracy.
Use RSMeans with its Historical Cost Index (HCI) to escalate older jobs forward, pair it with the relevant CCI for the city, and never, ever leave the overhead-and-profit column untouched if you intend to submit that number as a hard bid. The OH&P in RSMeans is generic 10%/5% — if your firm runs 8% overhead and needs 7% fee, strip it out and rebuild.
When assembly estimating pays for itself
Assembly estimating wins on anything above $2M, on any scope where your shop's productivity differs meaningfully from the national average, on any project with unusual access or repetition (refrigerated warehouse, data center, healthcare, clean room, high-rise MEP), and on any bid where the GC or owner will benchmark your number and ask you to defend it.
The cost of assembly estimating is time. Building a complete MCAA-based plumbing assembly for a 120,000 SF medical office building takes an estimator two to three full days before pricing. The payoff is a bid that holds together under scrutiny, a buyout target that the PM can actually manage, and a productivity loop where every completed job updates the PF database for the next bid.
Composite crew rate: the number the whole method sits on
A composite crew rate that is wrong by $8/hour on a 6,000-manhour job is $48,000 of bid error that will not surface until labor cost reports hit month three. Build it carefully:
- Base wage from the current CBA or Davis-Bacon wage determination.
- Fringe package — health, pension, training, vacation, annuity — typically $18-$36/hr for a union trade.
- Payroll burden — FICA, FUTA, SUTA, workers comp, general liability — usually 22-42% of base wage.
- Small tools allowance — commonly 2-4% of base labor.
- Supervision loading — a foreman at a crew ratio of 1:8 or 1:10 baked into the rate.
That buildup generates the number you multiply manhours by. Get it wrong and every line in the assembly is wrong by the same factor.
The practical answer: run both
Run the unit-cost number as a sanity check on every assembly estimate, and run an assembly buildup on every scope over $1M that came out of a unit-cost database. Where they disagree by more than 12%, dig in — one of them is modeling the job wrong, and you want to find out before bid day, not at the buyout meeting. The estimators who win consistently are the ones who use RSMeans as a parametric anchor and assemblies as the actual bid, not either one in isolation.
PILRS builds the takeoff quantity that both methods depend on, and ties MCAA and NECA labor units directly to the counted objects so your assembly number is ready the moment quantification is finished. See pricing and run it against your next bid set.