Appendix C — Estimating Reference: CSI Divisions, Unit Costs, Productivity, Waste Factors

C.0 How to use this appendix — and the one thing you must not do with it

Chapter 12 taught you how to price one thing: quantity, productivity, rate, plus material, plus waste. Chapter 13 taught you how to assemble the whole number. This appendix is the lookup layer underneath both of them — the division map, the conversion between the two organizing systems, the crew compositions, the adjustment factors, the waste table, the burden template, the general-conditions checklist, and the two checklists you run before a number leaves the building.

And then, at the end, the section that matters more than everything above it: how to build your own historical cost database, which is the only source of numbers that will ever actually be right for your market and your crews.

⚠️ The standing warning

Every unit cost, production rate, crew composition, wage, burden percentage, waste factor, and adjustment factor in this appendix is illustrative teaching data. This is not a price book. Do not bid off it.

The numbers here are internally consistent with each other and with the worked examples in the book, so the arithmetic teaches something real. That is their entire job. They are not survey data, they are not adjusted to your market, and they do not know your crews, your foremen, your form system, your soils, your union agreement, your workers' compensation class rates, or what the ready-mix plant quoted you this morning.

Real numbers come from three places, in this order of reliability: your own company's historical cost data · current quotes from the market you are actually building in · published cost data adjusted for time, location, and size. Section C.11 is about generating the first one, which is the reason this appendix's numbers can afford to be illustrative.

You will see that warning restated at the head of every table of money in §C.5. That is deliberate.

A note on what is cited and how. MasterFormat and UniFormat are published by the Construction Specifications Institute and Construction Specifications Canada; both are periodically revised, and division and element titles change between editions. Published cost data — RSMeans is the most widely used in North America — and the cost indices published by Engineering News-Record are real, useful, and referenced here by name and function only. Nothing in this appendix quotes a line number, an index value, a location factor, or a published productivity table. Where a professional body has published extensively on a subject — AACE International on cost engineering and forensic analysis, the Mechanical Contractors Association of America and the National Electrical Contractors Association on labor productivity and inefficiency factors — this appendix says so and sends you to them. It does not invent their contents.

C.1 CSI MasterFormat, division by division

MasterFormat is the numbering standard that organizes the specification, and — because the specification is how the market sells work — it organizes the estimate, the bid packages, the cost codes, and the schedule of values as well. Chapter 7 covers it as a document system; Chapter 13 covers it as an estimating container. Its purpose in an estimate is narrow and enormous: completeness and non-duplication. An estimate is wrong in exactly two ways — something is missing, or something is in twice — and a division structure catches both, because every scope item has one obvious home.

The third column of the table below is the one you will not find elsewhere: the classic scope gap that lives at the boundary between this division and the next one. Money does not fall out of the middle of a division. Inside a division the subcontractor is expert, motivated, and will not miss much. Money falls off the edges, where two competent bidders each reasonably assume the other one has it. That is the whole argument of Chapter 16, and this table is its lookup form.

C.1.1 Procurement and general requirements

C.1.2 Facility construction subgroup — Divisions 02–19

C.1.3 Facility services subgroup — Divisions 20–29

C.1.4 Site and infrastructure subgroup — Divisions 30–39

C.1.5 Process equipment subgroup — Divisions 40–48

These are the divisions a building contractor almost never opens and an industrial contractor lives in. They exist because a refinery, a water treatment plant, a food processing line, or a generating station has scope that does not fit a building's vocabulary.

Three rules for using the division table.

One: an empty division is a documented zero, and you write it as a dash. A dash means I looked. A missing row means nothing at all, and six weeks later nobody can tell the difference between "we decided it was zero" and "we forgot."

Two: packages are not divisions. MasterFormat organizes the specification. Your bid packages organize the market. They almost never line up. Northgate's enclosure package spans Divisions 03, 05, 07, and 08 by itself. The division tells you where to file it; the scope sheet tells you who bought it — and the scope sheet is in Appendix D.

Three: confirm the edition. MasterFormat is revised periodically. Division and section titles move, and a project manual produced under one edition and amended under another is a real thing that happens. Read the table of contents in the front of your project manual and file to that, not to this table.

C.2 UniFormat, and the crosswalk you actually need

MasterFormat organizes by work result — by trade, by product, by how the market sells. That is exactly what you want when you have a designed building and you are buying it.

It is exactly what you do not want when you have a program, a massing study, and an owner asking whether the enclosure should be precast or metal panel.

UniFormat organizes the same building by system and building element. It is available early, because you can say "exterior enclosure" long before you know it is precast. It keeps a building system whole, so a value-engineering comparison prices everything the change drags with it rather than just the visible half. This is the format Tomás Reyes reaches for in Chapter 11, and the one Chapter 7 introduces.

C.2.1 The Level 1 categories

C.2.2 When each one is the right tool

C.2.3 The crosswalk — UniFormat element to MasterFormat division to subcontract

This is the table readers ask for and rarely find. Read it left to right and you are moving from "what is this building made of" to "who do I call." Read it right to left and you are moving from a pile of subcontracts back to a system you can compare against an alternative.

The last column is the point. Every place a single UniFormat element decomposes into more than one subcontract is a place where scope can fall on the floor.

How to use the crosswalk in practice.

Going left to right — conceptual estimate to bid package. You have a UniFormat estimate at $/SF of element. Before you can buy any of it you must decompose each element into divisions and then group divisions into packages. Every place the "subcontracts" column reads 3 or more, write a scope sheet before you write a bid package.

Going right to left — bid results back to a system view. When an owner asks "why is the enclosure over budget," a division report cannot answer. Roll Divisions 03, 04, 05, 07, and 08 back up into B2010 and B2020, express it as $/SF of wall against 38,500 SF of curtain wall and 21,000 SF of precast, and now you have a comparison the owner can act on.

The teaching point, stated once. The translation between the two formats is where scope gaps are born. They are not born inside a specification section. They are born in the act of taking a building organized one way and buying it organized another way.

C.3 Illustrative unit costs by division

⚠️ Illustrative teaching data. Not a price book. Do not bid off this section.

Every figure below is a range, not a point value, and the ranges are wide because reality is wide. The right-hand column is the most useful part of every table: the single biggest driver of variation for that item. When your number falls outside a range, the driver column tells you where to look first.

Items marked † are the canonical worked values from the Northgate Outpatient Pavilion or the Cottonwood Creek Bridge Replacement, built from first principles in Chapter 12 and Chapter 13. They are internally consistent with the rest of the book. They are still illustrative.

C.3.1 Division 02 — Existing Conditions

C.3.2 Division 03 — Concrete

C.3.3 Division 04 — Masonry

C.3.4 Division 05 — Metals

C.3.5 Division 06 — Wood, Plastics and Composites

C.3.6 Division 07 — Thermal and Moisture Protection

C.3.7 Division 08 — Openings

C.3.8 Division 09 — Finishes

C.3.9 Divisions 10–14 — Specialties through Conveying

C.3.10 Divisions 21–28 — Facility services

MEP is a third of the money on a building like Northgate — Divisions 21 through 28 total $12,935,000, or 32.3% of the cost of work. It is also the family of divisions where $/SF of building is a far more reliable estimating and checking measure than any component unit price, because the systems are designed as systems.

C.3.11 Divisions 31–33 — Sitework

C.3.12 Whole-building benchmarks — the check, not the estimate

These are for sanity-checking a completed estimate, and for conceptual work before quantities exist. Use them the way §12.4.5 uses them: take off a quantity, extend it, then convert the answer into a unit you have history for and ask whether it looks like reality.

A caution about $/SF that is worth more than the table. A dollar-per-square-foot number is a statement about a building type in a market in a year. It is a check, not an estimate, and its accuracy is bounded by the completeness of the design, not by the estimator's skill. If your detailed summary says $268/SF on a building whose comparables run $310–$385, do not congratulate yourself — go find what fell out.

C.4 Productivity rates and crew compositions

These are the trades a construction manager most often self-performs or must independently price. For each: a representative crew, a production-rate range, and the man-hours per unit equivalent.

⚠️ Illustrative. These are planning values from a teaching text, not measured data.

C.4.1 Why man-hours per unit is the more stable measure

A production rate — "465 square feet of contact area per crew-day" — is three assumptions bundled into one number: how many people, how many hours in a day, and how much they install. Change any one of them and the number moves, even when nothing about the work changed.

Man-hours per unit strips out crew size and shift length and leaves only the labor content of the work.

                          crew size × hours per day
   MH per unit  =  ————————————————————————————————————
                     units produced per crew-day

Work the formwork crew both ways:

   Crew of 5, 8-hour day, 465 SFCA per day
     Production rate     = 465 ÷ 8      = 58.125 SFCA per crew-hour
     Man-hours per unit  = (5 × 8) ÷ 465 = 0.0860 MH/SFCA

Now put six people on it and get 540 SFCA a day. The production rate rises to 67.5 SFCA per crew-hour, which looks like an improvement. The man-hours per unit rises too — (6 × 8) ÷ 540 = 0.0889 MH/SFCA — which tells you the truth: you added 20% more labor and got 16% more output. The crew got less efficient per hour of labor, and only the man-hour measure showed it.

Three reasons the profession budgets in man-hours:

1. It survives crew-size changes. A foreman who adds two people has not changed the labor content of the work.
2. It survives shift-length changes. Ten-hour days do not make a wall cheaper to frame.
3. It is the unit your cost report already collects. Timecards record hours, not crew-days (Chapter 28), so man-hours per unit is the measure your own history will produce without any extra effort — which is the entire point of §C.11.

The practical rule: budget in man-hours, schedule in crew-days. The estimate tracks labor content; the schedule needs a duration, and a duration requires you to decide a crew size.

C.4.2 Crew compositions and production rates

Burdened rates in the crew tables are the canonical Northgate rates built in §12.7.1.

Earthwork — mass excavation and haul

Driver: haul cycle time governs mass excavation completely. A 4-minute change in cycle time can move production 25% without anyone working differently.

Formwork

Driver: reuse cycles. A form system used eight times and a form system used twice have the same labor content per cycle and wildly different amortized material cost — and the second one has more setup labor per SFCA because there is less repetition to learn from.

Reinforcing steel

Driver: bar size. A ton of #4 bar is many more pieces, ties, and lifts than a ton of #9. Small bar is expensive per ton and cheap per pound of structure.

Concrete placement and finishing

Driver: truck rotation. A slab crew is a queueing system. If the plant cannot keep a truck in front of the pump, the crew's production is set by the batch plant, not by the crew.

Masonry

Driver: scaffolding and material staging. Masonry production is a materials-handling problem wearing a trowel. Tender-to-mason ratio and how far the block travels decide the day.

Structural steel erection

Driver: tons per piece. A frame averaging 3 tons per piece erects several times faster per ton than one averaging 0.8 tons per piece, because the crane cycle is per pick, not per ton. Northgate's 985 tons is a moderately complex four-story frame.

Metal-stud framing

Driver: opening density and access. Ray's most expensive estimating mistake lives in the second and third rows of this table: 1,180 SF/day of warehouse framing became 540 SF/day in an occupied hospital corridor, and the mistake was not the rate — it was the undocumented rate.

Gypsum board

Driver: sheet size versus wall height, which is also the driver of the waste factor. A crew hanging 4×12 board vertically on 10'-6" walls throws away eighteen inches off every sheet and handles every sheet twice.

Rough carpentry

Driver: piece count, not quantity. General trades and rough carpentry are the two scopes where a countable physical unit is hardest to define, which is why Kestrel codes 06-1000-L in man-hours rather than a physical unit — an honest choice that costs you the ability to compute a productivity factor. Accept that trade openly rather than inventing a fake denominator.

C.5 Productivity adjustment factors

⚠️ Read these as planning heuristics with explicit caveats. They are not measured constants.

The direction of every effect below is well established across the industry literature. The magnitude on any specific job is not, and several are actively contested. Industry bodies including AACE International, the Mechanical Contractors Association of America, the National Electrical Contractors Association, and the Lean Construction Institute have published extensively on labor inefficiency and its measurement; go to them for the underlying studies and their limits. This appendix does not reproduce their tables and does not invent numbers with their authority attached.

Use these to price a decision. Use measured data from your own job to prove one.

C.5.1 The factor table

C.5.2 Four cautions that matter more than the numbers

One: the factors are not independent, and you must not simply multiply them. Stacking four trades and running sixty-hour weeks is not 0.78 × 0.84. The conditions overlap — the congestion is part of why the overtime is unproductive — and multiplying them stacks the same loss twice. When you use more than one factor, apply the largest, then adjust modestly for the others, and state in your basis of estimate exactly what you did.

Two: know what your baseline was. A factor is a ratio to a base. If your historical rate was collected on a job that already had two trades in the area, applying a trade-stacking factor to it double-counts. This is why §12.8 says the rate is the smaller half of the answer and the written basis is the larger half.

Three: a factor is a planning aid, not an entitlement. The measured-mile method (Chapter 31) — comparing an unimpacted period of the same work by the same crew to an impacted period — proves disruption. A table of factors does not, and a claim built on published factors alone will be attacked and usually beaten.

Four: the asymmetry. A production loss costs more than it looks like, because material does not care how long you took. Only the time-based portion of the unit cost inflates:

   Production drops 15%  →  time-based cost rises by 1 ÷ 0.85 = 17.6%

On the Chapter 12 formwork example, a 15% production loss raised the unit cost 14.3% and the labor-and-equipment portion 17.6%. Losses compound against you; they do not scale one for one.

C.5.3 Overtime, worked

Because this is the decision people price wrong most often. Baseline: the crew of 5 at $257.46/hr, 58.125 SFCA per crew-hour. Go to 6 × 10 = 60 hours, sustained, at the mid-range factor of 0.84.

   Adjusted rate = 58.125 × 0.84 = 48.825 SFCA/crew-hr

   Baseline week:  58.125 × 40 hr = 2,325.0 SFCA
   OT week:        48.825 × 60 hr = 2,929.5 SFCA
   Additional output:                  604.5 SFCA   (+26.0%)

   Baseline weekly labor: 40 × $257.46                = $10,298.40
   OT weekly labor:       60 × $257.46 + 20 × $104.69 = $17,541.40
   Additional cost:                                      $7,243.00

   $7,243 ÷ 604.5 SFCA = $11.98 per additional SFCA

Baseline labor unit cost is $4.43/SFCA. The marginal square foot bought with overtime costs nearly triple. Sustained overtime is not a productivity strategy. It is a time purchase, and you should know its price before you buy it — which on Northgate is why acceleration and absorption came out within about $13,050 of each other and the decision got made on Meridian's interim clinic lease instead.

⚠️ And the cost that is not in the arithmetic. Sustained sixty-hour weeks raise incident rates. On Northgate, the acceleration after the steel delay produced trade stacking, a rework event on deck-edge detailing, and a near-miss spike in weeks 34 through 36 — 22 near-misses against a running average of about 2.5 a week. An estimate carrying an optimistic production rate is a schedule-pressure generator, and schedule pressure is a hazard exactly like an unguarded edge. Price the rate you can achieve safely.

C.6 Waste, shrinkage, and loss factors

Every material quantity has two versions: the quantity that ends up in the building and the quantity you have to buy. The gap is waste, and it is a real budgetable cost, not a rounding error.

C.6.1 The waste table

These are not universal constants. They vary by crew, by layout, by the stock sizes available in your market, by weather, and by how tidy your site is. A company that tracks its own waste from its own cost reports replaces every number in this table within about three years, and its estimates get better for it.

C.6.2 Earthwork is a different problem: swell and shrinkage

Dirt is not wasted; it changes volume. Three states, three numbers for the same material.

These ranges are directional planning values. Use the project geotechnical report, and on a large earthwork job, a test strip. Do not treat a published range as a number.

C.6.3 The arithmetic of applying them

Material waste — inflate the quantity:

   Quantity to buy = neat quantity × (1 + waste factor)

Northgate footings: 1,216.43 CY neat × 1.019 = 1,240 CY carried. The estimate does not carry the geometry; it carries the geometry plus a named, quantified allowance.

Bank to loose — inflate for haul:

   LCY = BCY × (1 + swell)

32,000 BCY × 1.25 = 40,000 LCY. At $18.50 per loose cubic yard hauled and dumped, that is $740,000 — an equivalent bank-measure rate of $23.13/BCY. Extend the bank quantity against the loose rate and you are $148,000 short on one line.

Bank to compacted — deflate for fill:

   CCY = BCY × (1 − shrinkage)

12,000 BCY × 0.88 = 10,560 CCY. If the grading plan needs 12,000 CCY in place, you are 1,440 CCY short and must import 1,440 ÷ 0.88 = 1,636 BCY of borrow.

C.6.4 The two rules that keep waste from costing you twice

Rule one: pick a place and put it there. Either inflate the quantity or inflate the unit price — never both. Kestrel's convention is quantity, because a quantity you can see is a quantity you can audit. Doing both on Northgate's footings alone manufactures $4,414 of phantom cost. Scale that across forty divisions.

Rule two: waste is a material factor, not a labor factor. Your crew does not install the waste. Applying a waste factor to labor hours budgets crew time to install concrete that went home in the drum. (The exception: material handled twice — cut-off drywall carried to a dumpster — is a real labor cost, but you capture it in the productivity rate, not in a waste factor.)

C.7 The burdened labor rate — a build-up template

The wage is not the cost. The burdened rate is the cost, and on commercial work the gap runs roughly 35% to 55% of the base wage, and higher on high-hazard classifications and union work.

C.7.1 The template, blank and worked side by side

Fill the middle column for your own market. The right-hand column is the canonical Northgate carpenter from §12.7.1, carried through so you can see the shape.

C.7.2 Six notes that decide whether the template is right

The taxable wage base is not the same as wage plus fringes. Bona fide contributions to a benefit plan are generally not taxable wages; vacation and holiday pay generally is. Compute payroll taxes on line B, not on line A. Get this backwards and you over-burden every hour in the estimate. Tax treatment of fringes varies and changes — confirm the current treatment with your controller, not with a textbook.

FUTA has a low annual wage cap and SUTA varies enormously by state and by your own experience rating. The blended figure in any template is company-specific and year-specific. "Look up the rate" is the wrong instinct; "ask accounting for last year's effective rate" is the right one.

Workers' compensation is priced by class code, and the spread is brutal. Manual rates are set by state rating bureaus, vary by jurisdiction, and change. Below is the shape of the spread, expressed as a multiple of a carpentry rate — not as rates, because rates are jurisdictional:

That table is the answer to "why is an ironworker's burden a multiple of a carpenter's." It is not a judgment about the person. It is the insurance market pricing the observed frequency and severity of injury in that class of work. Two consequences for you: use the correct class code for each classification in each crew — Northgate's laborer burdens to $39.51 in a framing crew and $40.74 in a trench crew, same person, same wage, different work — and understand that if your estimating system carries one burdened laborer rate for the whole job, it is wrong somewhere.

Your experience modification rate multiplies all of it. Kestrel's EMR of 0.82 reflects a good loss history and is worth (1.00 − 0.82) × manual rate × total payroll on every hour the company works. That is the most direct financial argument for a safety program you will ever see, and Chapter 24 makes it properly.

Small tools and consumables — blades, bits, screws, string, chalk, gloves, layout paint — are real and are usually carried as 2–4% of labor because tracking them per unit is not worth the effort.

Union and prevailing-wage work is a different build-up with the same structure. The fringe package is defined by the collective bargaining agreement or the wage determination, contribution rates are specified rather than negotiated, and certified payroll makes the reporting non-optional. Prevailing- wage requirements, thresholds, and reporting vary by jurisdiction and by funding source — the federal Davis-Bacon framework and state "little Davis-Bacon" statutes are not the same and do not always reach the same work. Check the solicitation, not your memory.

C.7.3 The canonical Northgate burdened rates

Use these for internal consistency when working the book's examples.

Three different rates, three different jobs — do not mix them up. An individual burdened rate (above) prices one person's hour. A crew composite — Kestrel's carpentry composite is $58.40/MH and its concrete-crew composite is $54.00/MH — is the weighted average cost of a man-hour from a mixed crew including its foreman, and it is what you use to price a change order. A site-wide blended rate — Northgate's is $66.00/MH across everybody from a laborer to a sheet-metal journeyman — is what you use to price a site-wide disruption, and it is meaningless for any single trade.

C.8 General conditions checklist

General conditions are the cost of having a project rather than the cost of building the building. Nothing in this list becomes part of the finished structure, and on Northgate it is $2,900,000 — 7.25% of the cost of work.

The TD column is the one that matters: time-dependent items are billed per unit of time and convert directly into a daily burn rate. One-time items do not.

C.8.1 Project staff

Staff is $1,578,000 of Northgate's $2,900,000 — 54.4%. Carry that ratio in your head. On most commercial buildings project staff runs between half and two-thirds of general conditions. If your GC estimate puts staff at 30%, you have either forgotten somebody or double-counted the facilities.

C.8.2 Field office and utilities

C.8.3 Temporary facilities and site services

C.8.4 Safety

C.8.5 Cleanup and waste

C.8.6 Equipment and small tools

C.8.7 Communication and technology

C.8.8 Quality, testing, and permits

C.8.9 Closeout

C.8.10 Converting the checklist into a burn rate

   General-conditions burn rate = total general conditions ÷ contract duration in calendar days

Roughly $2.28 million of Northgate's $2.9 million is time-dependent. It does not care how much work got done this month. It only cares that the month happened. A schedule slip is already a cost problem, automatically, the instant it happens, before anybody argues about who caused it.

⚖️ The honest caveat on extended general conditions. A stipulated daily rate is an average, and extended general conditions are really a marginal rate at the end of the job — by then the assistant superintendent has rolled off, the trailers are down to one, and temporary heat is finished. A true marginal build-up might be materially lower; conversely, an extension landing at peak staffing can cost more than the stipulated rate. A stipulated rate trades accuracy for the elimination of a fight. That is usually a good trade, and it is a trade, which means somebody is slightly better off. Know which side you are on before you agree to the number.

C.9 The pre-bid estimate review checklist

Two people, about ninety minutes on a job of Northgate's size, before any number leaves the building.

Quantities

1. Does every quantity cite the sheet and detail it came from?
2. Was every quantity checked by a second method — count against schedule, area against $/SF, tonnage against SF of building?
3. Are units labeled on every line? (Yes, every line.)
4. Have all addenda been incorporated, and has the drawing log been reconciled to the issued-for-bid revision of every sheet?
5. Do quantities from the enlarged plans, details, and schedules agree with the overall plans?
6. Is every conceptual factor (lb/CY of rebar, $/SF checks) flagged for replacement when real information arrives?

Unit costs

7. Is every labor rate burdened, and does the burden use current rates from accounting?
8. Is the workers' compensation class code correct for each classification in each crew?
9. Does every production rate have a stated basis and stated conditions?
10. Has waste been applied exactly once — in the quantity or in the price, not both?
11. Are earthwork quantities in the right measure (bank, loose, or compacted) for the rate they are extended against?
12. Is freight in? Is sales tax in, and is the treatment correct for this jurisdiction and this project type?
13. Is equipment priced with mobilization, demobilization, and idle time?

Subcontracts and quotes

14. Has every quote been read to the end, including the exclusions on the second page?
15. Does the bid tab compare scope-corrected numbers rather than face amounts?
16. Is there an independent in-house number for each major trade, and has any large variance been explained?
17. Is every quote logged with its validity date, with escalation carried for anything bought after that date?
18. Have shared exclusions — items every bidder excluded — been circled separately and assigned somewhere?
19. Does each package have at least two real bidders, or a flagged and loaded plug?

Indirects and markups

20. Do the general conditions match the current schedule duration, and were they re-run after the last schedule change?
21. Is contingency tied to a risk register with named risks?
22. Is escalation applied to the correct scope and the correct window — the unbought portion only?
23. Is the bond computed on the final price, using the divisor method?
24. Is the markup the markup that produces the margin you intend?
25. Are gross-receipts or contractor's excise taxes, where they apply, computed on the price and included in it?

The document

26. Is there a basis-of-estimate page, and could a stranger price this job from it?

C.10 The bid-day checklist

Bid day failures are almost never estimating failures. They are clerical failures at high speed, and they are enormous.

Three rules do most of the work: one person touches the summary; every quote is logged with time, company, number, and stated exclusions before it is evaluated; and the bid leaves early. A bid that is technically finished at 1:58 and physically outside the building at 2:00 is worth nothing.

And the "what did we forget" ritual is not a formality. Everybody speaks — the quote takers, the scribe, the estimator who did earthwork, the person who has been quiet for four hours. "Go around. One thing each. If you have nothing, say nothing." A twenty-three-year-old field engineer saying "we never got a number for the temporary chain-link along Ferry Street" saved $71,000.

C.11 Build your own historical database

Everything above this line is a starting point. This section is the destination.

The reason the numbers in this appendix are illustrative is not modesty. It is that your own data beats anyone's book, and it beats it by a wide enough margin that a company which generates its own history and one which does not are playing different games. Published data is compiled from broad surveys and adjusted by location factor. It cannot know your crews, your foremen, your form system, your subcontractor pool, your site conditions, or your supervision ratio. Your cost report knows all six.

Twenty completed jobs of your own work is a better predictor of your cost than any published table. The catch is that the history only exists if somebody deliberately builds the machinery that produces it — and that machinery is designed before notice to proceed, not at closeout.

C.11.1 Structure cost codes so they generate history

This is Chapter 28's subject, and it is an estimating subject before it is an accounting one.

Kestrel's code has three parts:

        03    -    3100    -    L
        ▲          ▲            ▲
    DIVISION    ACTIVITY     COST TYPE
   CSI Master-  the countable   L = labor
   Format div.  unit of work    M = material
                                E = equipment
                                S = subcontract
                                O = other / indirect

Five requirements. Miss any one and the code produces no usable history.

The single sentence to remember: a company that codes its costs sloppily is not just failing at cost control — it is destroying its own estimating database, and it will not notice for three years.

C.11.2 What to capture at job close

Held two weeks after substantial completion, while the team still exists — not after final completion, when everyone has scattered (Chapter 40).

The estimator and the project accountant sit down for a day and produce one table. Here is what a row must contain:

The last two rows are where most companies destroy their own data. Northgate's structural steel erection came in at $712/ton against an estimated $655/ton — and the entire variance is the acceleration premium and the second crew after the steel delay. Drop $712 into the unit-cost library and you have permanently overpriced steel erection on every future bid because of one job's one-time event. The right entry is $655, plus a documented note about what acceleration costs when it happens.

A variance without a diagnosis is a number. A variance with a diagnosis is an estimating standard.

Capture four other things in the same session:

• Productivity by cost code in MH/unit, with the conditions.
• Waste factors actually experienced, which within about three years will replace every number in §C.6.
• General conditions actual versus estimated, item by item, which is how you learn that final cleaning is three cleans and commissioning support is 1.5 staff-days a week rather than 0.5.
• Subcontractor performance evaluations, honestly — including the failures that were yours. These feed the next prequalification.

C.11.3 Normalize before you reuse — time, location, size

A rate from a job three years ago in another city on a building twice the size is not your rate. Three adjustments, applied in this order, before the number enters the library or comes out of it.

1. Time — escalation via a published cost index.

Engineering News-Record publishes construction cost indices — its Construction Cost Index and Building Cost Index are the two most widely used in North America — that track the movement of a fixed market basket of labor and materials over time. Trade associations and government agencies publish others. Use one consistently; do not switch indices mid-library.

                                       index at target date
   Adjusted cost = historical cost × ————————————————————————
                                     index at historical date

Worked: a footing labor rate of $332.26/CY from a job three years ago, with the index up 14.2% since:

   $332.26 × 1.142 = $379.44 / CY

Two cautions. Escalate labor and material separately where you can, because they move independently — a general index blends them and will be wrong on a material-heavy or a labor-heavy scope. And do not escalate what you have already bought. Northgate escalates only the $26,400,000 that was not under firm quote at GMP execution; escalating the whole cost of work would double-count the third of the job already locked and hand a competent owner's estimator a $190,000 error to find.

2. Location.

Published cost data ships with location factors that adjust a national average to a specific city. Use them for published data. For your own data, the more useful adjustment is a direct wage and material comparison: what does a carpenter cost there against here, and what does ready-mix cost there against here? A location factor is a blend; a wage comparison is a fact.

                                    target-market burdened wage
   Adjusted labor = historical × ————————————————————————————————
                                  source-market burdened wage

And note what a location factor cannot tell you: subcontractor availability, market appetite, backlog, weather, code differences, and permitting duration. Those are not cost adjustments. They are schedule and risk adjustments, and they belong in the risk register and the general-conditions duration, not in a multiplier.

3. Size.

Cost per square foot does not scale linearly. Larger buildings generally cost less per square foot — mobilization, general conditions, and setup spread over more area — but the effect flattens, and above a certain size complexity starts adding it back. Small projects carry a genuine size penalty that is mostly general conditions and mobilization, not unit costs.

Two practical rules:

• Do not adjust unit costs for size. A cubic yard of footing concrete does not know how big the building is. Adjust the general conditions and the mobilization, which is where the size effect actually lives.
• Adjust element quantities, not element rates. A bigger building has a lower ratio of exterior wall to floor area, which is a quantity effect that a UniFormat estimate (§C.2) captures automatically and a $/SF estimate hides.

C.11.4 A worked normalization

You are bidding a 96,000 SF outpatient building in a different metro. Your best data is Northgate: 132,000 SF, three years old, footings at an actual $385.48/CY including a documented cold-weather event.

   Step 1 — clean the number.
     Strip the one-time cold-weather event noted in the diagnosis field.
     Cleaned base rate:                                    $352.00 / CY

   Step 2 — time. Index up 14.2% over three years.
     $352.00 × 1.142                                     = $401.98 / CY

   Step 3 — location. Target market carpenter burdened rate is 8% lower;
     ready-mix is 6% higher. Labor is ~20% of this unit cost, material ~73%.
     Labor portion:     $401.98 × 0.20 × 0.92            =  $73.96
     Material portion:  $401.98 × 0.73 × 1.06            = $311.05
     Equipment portion: $401.98 × 0.07 × 1.00            =  $28.14
     Adjusted                                             = $413.15 / CY

   Step 4 — size. NO adjustment to the unit cost.
     Instead, re-run general conditions against the new duration and
     the new staffing plan, and re-run mobilization.

   Carry:  $413 / CY, basis documented.

Now write down the basis, because the number is worthless without it: "Footing concrete placed and finished, $413/CY. Basis: Kestrel historical actual, Northgate Outpatient Pavilion, cost code 03-3100, 1,240 CY, cleaned of a documented cold-weather event; escalated 14.2% by published cost index over 3 years; adjusted for target-market wages (−8% labor) and ready-mix (+6% material). Assumes day shift, single crew, clear access, no trade stacking, gang forms on repeat footings."

That paragraph is worth more than the number, because it converts a guess into a managed condition that a second person can check, argue with, and improve.

C.11.5 Where published data legitimately fits

None of this means published data is useless. It means it has a job, and the job is not "be your primary basis on work you self-perform."

And whichever you use, note in the basis of estimate which source, which edition, and which adjustments you applied. An estimate that cannot say where its numbers came from is not an estimate; it is a rumor with a dollar sign.

C.12 If you remember nothing else from this appendix

1. The division tells you where to file it; the scope sheet tells you who bought it. Money falls off the edges of divisions, not out of their middles.
2. UniFormat for conceptual, MasterFormat for detailed — and the translation between them is where scope gaps are born.
3. Man-hours per unit is the stable measure. Production rates bundle crew size and shift length into a number that moves when nothing about the work changed.
4. Adjustment factors price a decision; measured data proves one. Do not multiply them together and do not present them as entitlement.
5. Apply waste exactly once, to the quantity, and never to labor.
6. The wage is not the cost. Build the burden from class codes and current accounting rates, and remember that the same laborer costs different amounts in different crews.
7. General conditions are a rate, not a quantity. Sort every line into time-dependent or one-time, divide by the duration, and you have the number that turns every schedule conversation into a cost conversation.
8. Every number in this appendix is illustrative. The numbers that will make you money are the ones your own cost report generates — which is why §C.11 is the only section here you cannot get from anybody else.

Related: Appendix A — Construction Math Reference · Appendix D — Forms and Templates Library · Appendix H — Software and Tools Directory · Chapter 11 · Chapter 12 · Chapter 13 · Chapter 16 · Chapter 20 · Chapter 28