Chapter 13 — Detailed Estimating: CSI Divisions, Subcontractor Quotes, Equipment Costs, and the Final Number

The Hook: 11:52 a.m.

Bid day. The Ridgeline Regional Library and Learning Center — $31.4 million, hard bid, lump sum, bids due at the Rivermont County Library District offices at 2:00 p.m. sharp. A bid that arrives at 2:01 is not a late bid. It is not a bid at all.

Tomás Reyes has the big conference room. He has had it since Monday. The long wall is covered with a projected spreadsheet forty feet wide, and Tomás is standing in front of it the way a conductor stands in front of an orchestra, except the orchestra keeps calling in changes. Three phones on the table. Two estimators taking quotes. Dani Okonkwo — six months out of school, on loan from Northgate for the day because Tomás believes everybody should see one of these before they are thirty — has been given exactly one job: write down every number, who called it in, what time, and what they said was excluded. Nothing else. Do not calculate. Do not judge. Write.

At 11:40 the mechanical page has three lines on it:

Kestrel's own in-house estimate for the HVAC package is $5,120,000.

At 11:52 the third phone rings. Dani takes it, writes, and turns around holding the pad up like it might be a trick.

"Ashfield Mechanical. Four million nine hundred seventy-eight thousand."

The room goes quiet for about a second and a half. Ashfield is $340,000 under Cardinal. On a $31.4 million bid where Kestrel expects to win or lose by less than one percent, $340,000 is not a rounding difference. It is the whole job.

Here is the thing I want you to notice, because it is the thesis of this chapter and I have watched it separate good estimators from expensive ones for twenty-two years.

Tomás did not smile.

He said, without looking away from the wall: "What did they miss?"

Not did they miss something. What. Then he looked at the clock — 11:52, two hours and eight minutes — and started the twenty minutes that decide the job.

"Dani, read me their exclusions. All of them, in order, slowly."

"It's a one-page proposal. It says: 'Excludes medical gas systems, excludes test and balance by independent agency, excludes duct cleaning and final filter change, excludes cutting and patching, excludes painting, excludes fire alarm interface wiring, excludes concrete pads and curbs, excludes seismic bracing calculations by others.'"

Tomás held up a hand. "Stop. Go back two. Medical gas."

The Ridgeline library has a small clinic suite — the county co-located a public health screening room into the building program at the sixty-percent design review. Medical gas is not what you expect in a library, which is exactly why it is easy to lose. In MasterFormat, medical gas piping lives in Division 22, Plumbing. In Kestrel's bid package, it was put with mechanical, because Kestrel wanted one throat to choke on that clinic suite.

Ashfield read the division. Kestrel wrote the package. Those are not the same document, and the difference was $268,000.

"Test and balance," Tomás said. "Who did we assume?"

"The mechanical sub. Package 23A, item 14."

"Does Cardinal carry it?"

An estimator, flipping pages: "Cardinal includes test and balance by an independent agency. Says so, page two."

"That's another eighty-something. Call Ashfield. Ask them three questions and do not tell them anybody's number."

That last clause — do not tell them anybody's number — takes ten seconds to say, and some contractors have never learned it. We will spend real time on it in §13.5, because it is the difference between negotiating and stealing.

At 12:26 Ashfield called back with a revised proposal: add $241,000 for medical gas through a partner, add $79,000 for independent test and balance, add $28,000 for duct cleaning and final filters. Revised number: $5,326,000.

Cardinal, at $5,318,000, was now $8,000 low. The $340,000 gap had never existed. It was a $383,000 hole with $340,000 of daylight showing through it.

A low number is a question, not an answer. That is the whole chapter. Everything that follows is machinery for asking the question fast enough, on a clock that does not care how tired you are.

Chapter 12 taught you how to price one thing — quantity times productivity times rate, plus material, plus waste. This chapter is about assembling the whole number: twenty-three divisions, forty-one subcontract packages, an equipment fleet, a staff, two winters of temporary heat, a bond, and a fee, reconciled into a single figure you will sign your name under and then live inside for the next nineteen months.

🏃 Fast Track: If you build bid tabs for a living, read §13.2 (the Northgate reconciliation), §13.3 (the itemized general-conditions build-up and why the burn rate is $5,150 per calendar day), the leveling worked example in §13.4, and §13.9 (where the number really gets set). Skim the rest.

🔬 Deep Dive: Unit costs, productivity factors, waste factors, and division-by-division reference data are in Appendix C. Blank scope sheets, bid tabs, and a general-conditions worksheet are in Appendix D. The estimate's afterlife as a control budget is Chapter 28; as a schedule of values it is Chapter 32.

13.1 The container: MasterFormat as an estimating structure

An estimate needs a filing system, and in North American commercial construction that filing system is CSI MasterFormat — a numbering standard published jointly by the Construction Specifications Institute and Construction Specifications Canada. You met it in Chapter 7 as the organizing spine of the project manual. Here it does a second job: it is the container that guarantees you counted everything once and only once.

That is the entire purpose. Not elegance. Completeness and non-duplication. An estimate is wrong in exactly two ways — something is missing, or something is in there twice — and a division structure catches both, because every scope item has one and only one obvious home.

Here is the working map, with who typically bids each division on a commercial building.

Divisions 15–19, 24, 29, 30, and 34–49 exist in the standard and get used in industrial, process, transportation, and utility work. On a typical commercial building they are empty — and an empty division is not a mistake. It is a documented zero. Write the line, put a dash in it, move on. 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."

The trap: packages are not divisions

This is the mistake that cost Ashfield the Ridgeline job, and it earns its own callout.

💡 Aha moment. MasterFormat organizes the specification. Your bid packages organize the market. They almost never line up, and the gaps between them are exactly where scope disappears.

Three examples from Northgate:

• Architectural precast panels are Division 03 by MasterFormat — they are concrete. But they are enclosure by function. They get bought against the curtain-wall schedule, coordinated with Division 05 embeds and Division 07 air barrier. The enclosure bid package spans Divisions 03, 05, 07, and 08 all by itself.
• Firestopping is Division 07. But the mechanical, electrical, and plumbing trades each firestop their own penetrations, so the Division 07 firestopping package covers only the drywall trade's penetrations and the perimeter slab-edge joint. Buy "all firestopping" from one sub and let each MEP sub carry their own, and you paid twice. Buy neither and you have an unsealed rated wall and a failed inspection two days before the certificate of occupancy.
• Medical gas is Division 22. Kestrel puts it with mechanical because the mechanical contractor already owns the clinic suite. Every bidder must be told this in writing, in the bid package, in the scope sheet. Ashfield was not told clearly enough — and that is at least half Kestrel's fault.

The rule that survives all of this: the division tells you where to file it; the scope sheet tells you who bought it. We build a scope sheet in §13.4.

The Northgate cost of work by division

Here is the whole thing. This is the estimate summary Tomás handed Nadia Haddad and Pri Sethi at the Guaranteed Maximum Price (GMP) submission for the Northgate Outpatient Pavilion — 132,000 gross square feet, four stories, structural steel frame, curtain wall and architectural precast enclosure, on a 6.2-acre urban-edge site with an active clinic staying open along the north property line.

Percentages are rounded and may not total exactly 100%.

Read that table the way a doctor reads a chart. The three biggest numbers on a four-story medical outpatient building are HVAC (12.98%), Finishes (12.25%), and Metals (12.13%). If those three are right, the estimate is probably right. If any one of them is off by ten percent, you have moved roughly a half-million dollars, and no amount of care in Division 10 will get it back.

Notice also that Divisions 21 through 28 — mechanical, plumbing, fire protection, controls, electrical, communications, and security — total $12,935,000, or 32.3% of the cost of work. One dollar in three, bought from six or seven subcontractors who must all fit their systems into the same ceiling cavity above the same corridor. That is the arithmetic behind the Chapter 10 lesson that MEP, not structure, sets the interior schedule. It is also a third of your money.

🔄 Check your understanding. Kestrel's bid package for the enclosure asks glaziers to price "the exterior wall." A bidder returns a number covering curtain wall and storefront only. Which scope did that bidder almost certainly leave out, and roughly how much money is sitting in it on Northgate?

13.2 The reconciliation: from $40,000,000 to $47,500,000

The cost of work is not the price. Between them sit five lines that most people outside the industry — and a discouraging number of people inside it — cannot correctly name or order.

Here is the Northgate GMP reconciliation. Memorize the shape of it; you will build one of these on every job you ever price.

Check the arithmetic yourself, because I want you in the habit:

• $40,000,000 + $2,900,000 + $900,000 + $1,320,000 = $45,120,000
• $45,120,000 × 0.040 = $1,804,800
• $45,120,000 + $1,804,800 + $575,200 = $47,500,000

And the derived rate that will follow you through the rest of this book:

$47,500,000 ÷ 132,000 SF = $359.85/SF, which the whole Northgate team rounds to $360/SF in conversation.

What each line is actually for

Cost of work is the physical building. It is what a subcontractor invoices and what a truck delivers. On a GMP contract it is the auditable portion — Meridian Health System's auditors can and will ask to see the subcontracts and the material invoices behind it.

General conditions is the cost of having a project rather than the cost of building the building. Nothing in this line becomes part of the finished structure. Every dollar of it is spent to make the building possible: the superintendent who sequences the work, the trailer they sit in, the dumpster the trades throw into, the temporary heat that lets the drywall mud cure in January. §13.3 builds it line by line.

Insurance and bonds is priced as a rate on contract value, not on scope, which means it moves when the contract value moves and is therefore the one line in the estimate that is genuinely mechanical.

Contingency is not padding. Chapter 6 made this a threshold concept: contingency is a priced, owned, drawn-down reserve for identified risk. On Northgate it is 3.0% of $43,800,000 — $1,314,000 — carried at a round $1,320,000. On this contract, unused contingency at the end splits 75% to Meridian and 25% to Kestrel, which is the single most powerful incentive alignment in the whole agreement. Kestrel gets paid a quarter of every dollar of risk it manages away.

Fee is Kestrel's compensation for management, overhead, and profit. It is a markup — 4.0% applied to $45,120,000 — and we will pull that apart in §13.8, because "4% fee" and "4% margin" are two different amounts of money.

Escalation covers the price movement between the day you sign the GMP and the day you actually buy the work. At GMP execution Kestrel had firm quotes covering about 34% of the cost of work. The other 66% — $26,400,000 — was still exposed to the market. Escalation is priced against that exposure, not against the whole job.

💰 Money check. Escalation, worked.

Unbought scope at GMP: $40,000,000 × 0.66 = $26,400,000, split into material-heavy and labor-heavy exposure:

What it means for the job: Kestrel has bought itself about a 2% cushion on two-thirds of the work. If steel moves 6% before the mill order goes in, that cushion is gone by itself and the contingency starts absorbing the rest. This is why the buyout calendar in Chapter 16 is a risk document, not a purchasing document: every week you delay buying is a week you are still exposed.

⚖️ What the contract says. A GMP is not a lump sum. Under a cost-plus-fee-with-a-GMP agreement of the AIA A102/A133 family, the owner pays the actual cost of the work plus the fee, capped at the guaranteed maximum. If Kestrel spends $44,300,000 on the cost of work, Meridian pays $44,300,000, not $45,120,000 — and the savings split applies to the unused contingency. If Kestrel spends $46,000,000, Kestrel eats the overage. That asymmetry is the whole point of the word "guaranteed," and as Chapter 4 established, it guarantees the contractor's exposure, not the owner's cost. Scope changes still move the number — CO #14 will prove that expensively in Chapter 31.

13.3 Division 01 is a cost, not a formality

In Chapter 7 I told you that Division 01 is where the money is. New estimators skip it because it reads like administration — submittal procedures, closeout procedures, temporary facilities, quality requirements. It reads like paperwork.

It is $2,900,000.

That is 7.25% of the cost of work and 6.1% of the GMP, and unlike almost every other number in the estimate, it is not driven by quantity. It is driven by time.

What the specification actually requires you to pay for

Read Division 01 with a pen and a running total. On Northgate the requirements that carry real cost are:

• Temporary facilities and utilities — offices, power, water, sanitary, heat, lighting, and the meters and permits behind them.
• Temporary protection — floor and stair protection, dust partitions at the active clinic on the north line, an accessible path of travel maintained around the site at all times.
• Progressive and final cleaning — the specification says "broom clean daily and remove debris from the site"; that sentence is $124,200 of dumpsters and hauling.
• Safety program — orientation, personal protective equipment, fall-protection hardware, guardrail systems, first aid, the safety manager's time.
• Quality control and testing — Kestrel's own quality management, mockups, and pre-installation conferences. (The independent testing laboratory is owner-retained on Northgate; on many jobs it is not, and that single sentence in the supplementary conditions is worth six figures.)
• Submittals and scheduling — the project engineer, the scheduler, the software, the reproduction.
• Permits and fees — those not paid directly by the owner.
• Surveying — control, benchmarks, robotic total station work, and the as-built survey.
• Waste management — including the diversion and sorting the sustainability section demands.
• Closeout — as-built drafting, operation and maintenance manual compilation, warranty administration, and owner training coordination. The cost of the closeout process itself is a Division 01 line, and it is the one every rookie forgets, because it happens after the interesting part is over.

The itemized Northgate general-conditions estimate

Here it is, built the way Tomás builds it: staff first, because staff is the majority; then facilities; then services. Nineteen months of field duration on a 565-calendar-day contract, with the project manager and superintendent carried a little longer at each end for preconstruction and closeout.

Project staff (rates are fully burdened monthly — base salary plus payroll taxes, insurance, and benefits):

Facilities, services, and consumables:

Staff is $1,578,000 of $2,900,000 — 54.4%. That ratio is worth carrying in your head. On most commercial buildings, project staff runs somewhere between half and two-thirds of general conditions. If you build a GC estimate where staff comes out at 30%, you have either forgotten somebody or double-counted the facilities.

The burn rate, and why it decides everything

$2,900,000 ÷ 565 calendar days = $5,132.74 per calendar day.

Kestrel and Meridian negotiated the contractually agreed extended-general-conditions rate to a round $5,150 per calendar day. That is the number that appears in the supplementary conditions, and it is the number that gets multiplied by every compensable delay day for the rest of the project.

🔍 Why this works. Look back at the itemized table and sort it in your head into two buckets: things you buy once, and things you buy per unit of time.

Buy once: trailer delivery and setup ($16,000 of the trailer line), fencing, temporary roads, final cleaning, the mockups, the as-built survey.

Buy per month: every single staff line. Trailer rent. Temporary power. Dumpsters. Portable toilets. Vehicles. Security. Photography.

Now count: roughly $2.28 million of the $2.9 million is time-dependent. It does not care how much work got done this month. It only cares that the month happened.

That is the mechanism behind Theme 2 — the schedule and the budget are the same conversation. A schedule slip is not a schedule problem that might later become a cost problem. It is already a cost problem, automatically, the instant it happens, at $5,150 a day, before anybody argues about who caused it. Nobody has to decide to spend that money. It spends itself.

And on Northgate there is a second meter running next to the first one. Liquidated damages are $5,500 per calendar day. Add them:

$5,150 + $5,500 = $10,650 per calendar day of total exposure to slipping substantial completion past September 18, Year 2.

💰 Money check. When Ironbridge Steel's anchor-bolt submittal sat in Kestrel's office for 11 days and Caldwell Structural then took its full contractual 14-day review, Ironbridge missed its mill rolling slot and steel erection start slipped from August 4 to August 27, Year 1 — 23 calendar days, on the critical path.

23 CD × $10,650/CD = $244,950.

Two hundred forty-five thousand dollars, generated by a stack of paper sitting on a desk. Not by bad welding. Not by weather. By eleven days of nobody moving a submittal. That is a Division 01 failure with a Division 05 price tag, and it is the reason Chapter 25 exists.

🔄 Check your understanding. Kestrel's general conditions are $2,900,000 over 565 calendar days. Suppose the owner adds scope that extends the contract by 40 calendar days, with no acceleration and no change to the staffing plan. Before you compute anything: is Kestrel entitled to 40 days of the full $5,150, or something less? Why?

13.4 Subcontractor quotes and the bid tab

On Northgate, about 86% of the cost of work is bought from somebody else. Kestrel self-performs cast-in-place concrete and rough carpentry; everything else arrives as a subcontractor's price. Which means the accuracy of a $40 million estimate is mostly the accuracy of forty-one other companies' estimates, assembled by you, under time pressure, from documents of wildly varying quality.

The bid tab is how you do that without losing money.

Coverage: how many bidders is enough

The working rule in most commercial markets is three qualified, competitive bidders per trade. Not three names. Three companies that actually intend to bid, are capable of the work, and are not too busy to care.

When a trade has one bidder, you have three honest moves, and none of them is "hope."

1. Estimate the trade yourself, in detail, from quantity and unit cost, and use your number as the check on theirs. If the sole bidder is 8% over your detailed build-up, you have a negotiating position. If they are 40% over, you have a scope misunderstanding or a market problem.
2. Break the package up. One elevator bidder is a fact of life; one drywall bidder means your package is too big, your schedule is too tight, or your terms scared people off. Split it by area or by system and see if smaller firms come to the table.
3. Carry a qualified allowance and say so out loud. Tell the owner: "We have one bidder on Division 14. We are carrying $780,000 as an allowance, and here is the exposure." An allowance disclosed in the basis of estimate is a managed risk. An allowance hidden inside a division total is a lie with a timer on it.

🏗️ From the field. Early in my career I had one roofing bidder on a school addition, and I carried their number because it was close to my in-house estimate and I was tired. What I did not do was ask why nobody else bid. The answer, which I got eight months later from a competitor in a parking lot, was that our specification called for a warranty term that only two manufacturers in the region would write, and one of them had a backlog. My "competitive" number was a monopoly number with good manners. It cost about $61,000 more than it should have, and every dollar of it was available to me on bid day for the price of one phone call I did not make.

Scope sheets: the most valuable document in buyout

A scope sheet is a written, itemized list of every element of work in a bid package, prepared before bids come in, against which every bidder's proposal is checked. It is the single most valuable document in the entire buyout process, and it costs an estimator maybe two hours per trade.

Here is a real one — the Northgate drywall and framing package, which carries a Kestrel budget of $2,760,000 within the $4,900,000 Division 09 line. The three bidders are Meridian Interiors, Cordova Wall Systems, and Bright Line Drywall. (All three are illustrative composites, as is every company in this book.)

Note item 12. All three bidders exclude lead-lined board. That is not a leveling adjustment — it does not change the comparison between them at all. But it is absolutely a scope gap, and it must be covered somewhere: on Northgate it goes into Division 13 with the radiation-shielding contractor. An exclusion shared by every bidder is invisible on the bid tab and lethal on the job. Circle those separately.

Note item 23. Cordova and Bright Line acknowledged Addendum 4; Meridian acknowledged only through Addendum 3. Addendum 4 added six exam rooms on level 3 — about 1,180 LF of additional partition. Meridian's number is a price for a different building.

🧩 Productive struggle

Before you read the next table, take five minutes and try this yourself.

You have three base bids: Cordova at $2,512,000, Meridian at $2,684,000, Bright Line at $2,748,000. Cordova is $236,000 low and Bright Line is the most expensive of the three. Kestrel's own budget is $2,760,000.

Kestrel's carried values for the gaps are: shaft wall $118,000 · in-wall blocking $46,000 · firestopping $88,000 · impact-resistant board $38,000 · hoisting and distribution $52,000 · daily cleanup $31,000 · temporary protection $17,000 · patching after other trades $26,000 · mockup $9,000 · access panels $12,000 · off-hours work at the clinic corridor $18,000 · Addendum 4 $68,000 · Level 5 finish (Meridian's own alternate price $34,000; Kestrel's carry for Cordova $41,000). A subcontractor bond, where not included, costs 1.15% of the leveled subtotal.

Who is actually low, and by how much? Write your answer down before you look.

Bid tabulation and leveling

Leveling means adjusting every bidder's number so all of them cover the same scope, so that the comparison is between prices rather than between paragraphs. You add back what a bidder excluded, subtract what they duplicated, adjust for alternates and addenda, and apply the same bond and insurance treatment to everybody.

Look at what happened. The rankings completely inverted. The apparent low bidder is $196,822 more expensive than the apparent high bidder. The bidder whose base number looked worst — Bright Line, at $2,748,000 — is the actual low at $2,778,000, and lands within 0.65% of Kestrel's own budget.

That last fact is the quiet vindication in the table. Kestrel's estimator built a $2,760,000 number from quantities and unit costs, without seeing a single quote, and the true market price came in $18,000 away. That is what a good detailed estimate does: it tells you which quote to believe.

And notice what Bright Line's number really was. It was not high. It was complete. Cordova's $2,512,000 was a perfectly honest price for a smaller scope, clearly qualified in writing. Cordova did nothing wrong. If Kestrel had carried Cordova's number and awarded Cordova's subcontract, the missing $429,000 would have arrived over eighteen months as change orders, backcharges, and scope-gap arguments — mostly at a worse price than $429,000, because nobody negotiates well in February with the ceiling grid already up.

Reading a quote for what is missing

Train your eye. These are the exclusions that show up over and over, and each one is a real number:

⚖️ What the contract says. The defense against all of this is a subcontract that establishes order of precedence — the subcontract and its scope exhibit govern over the subcontractor's proposal. Write the scope sheet into the subcontract as Exhibit A, list any accepted qualifications explicitly in Exhibit B, and include a clause stating that qualifications, exclusions, and conditions contained in the subcontractor's proposal are not part of the agreement unless expressly listed in Exhibit B. Without that clause, a sub can attach their two-page proposal with fourteen exclusions and argue the exclusions came along with the price. With it, the only exclusions that exist are the ones you knowingly accepted. This is the same order-of-precedence logic you learned for the prime contract documents in Chapter 7, applied one tier down.

Evaluating a bidder beyond price

The leveled low number is a candidate, not a decision. A subcontractor who cannot perform is more expensive than one who costs 4% more, because a default in month eleven costs you the schedule, the rebid premium, and the surety's timeline.

Prequalification — doing this work before the bid rather than during it — is the subject of Chapter 19, and the mechanics of converting a leveled bid into an executed subcontract are Chapter 16. For now, hold this: you are not buying a price. You are buying a company's next nine months.

🔄 Check your understanding. A drywall sub's leveled number is $61,000 higher than the low bidder's. Their EMR is 0.71, they have worked four Kestrel jobs with no backcharges, and their named foreman ran your last hospital fit-out. The low bidder is new to you, has an EMR of 1.34, and their proposal requires payment within 15 days of invoice. What is the $61,000 actually buying?

13.5 The ethics of the bid tab: shopping, peddling, and the line

This is a required conversation, and I am going to be precise, because vagueness here is how people talk themselves into things.

Bid shopping is when a general contractor, having received subcontractor quotes, discloses or characterizes one subcontractor's price to another and invites a lower number for the same scope. It can happen before the prime bid ("Ashfield is at 4.9 — can you get under that?") or after award ("we're carrying $2,512,000, beat it and the job is yours"). The tell is that nothing about the work changed. Only the knowledge of somebody else's price changed.

Bid peddling is the mirror image, initiated by the subcontractor. A sub who did not submit, or who submitted and was not low, approaches the general contractor after the prime bids are opened and offers to beat the number the GC carried. The tell is the same: no scope change, just a price chase against information that was supposed to be confidential.

Both are prohibited or restricted on most public work. Many states require prime bidders on public projects to list their subcontractors at bid time and forbid substituting a listed sub without a stated, approved reason. The specific mechanism, the dollar thresholds, and the penalties vary substantially by state and change over time, so check your jurisdiction's public contracting code and the bid documents themselves rather than assuming. Both the Associated General Contractors of America and Associated Builders and Contractors address the practice in their ethical standards. And on private work, where no statute reaches, plenty of owners write anti-shopping language into the prime contract anyway, because owners have figured out what it does to their buildings.

Why it is corrosive, in numbers

Bid shopping does not lower the cost of a building. It moves money and quality around, and it takes a cut on the way through.

1. Subcontractors price a known shopper higher. Every market has a whisper network. When a sub believes their number will be shopped, they add a hedge — call it two to four percent — or they simply do not bid. On $40,000,000 of subcontracted work, a three percent "shopper's premium" is $1,200,000 a year against you, forever, on every job.
2. Your best subs stop bidding. The firms with the deepest benches and the best foremen have more work than they need. They do not participate in auctions where the last number wins. You are left with the firms who must bid, and the correlation between "must bid" and "is in financial trouble" is not zero.
3. A shopped price gets recovered in scope. A sub forced down $80,000 does not absorb $80,000. They find it: thinner crews, a cheaper foreman, substituted product, and a proposal full of exclusions written to create change orders. You will pay the $80,000 back, plus markup, plus the argument. Curtis Boone's Rivermont Elementary job is the worked example, and it is in case-study-02.md.
4. It poisons the field. A subcontractor who believes they were cheated in buyout behaves differently at 6:30 a.m. in month fourteen. They will not do you the small favors — the two-hour resequence, the Saturday, the "we'll work around them and come back" — that keep a job on schedule. Those favors are worth more than the $80,000, and you cannot buy them back.

Where the line actually is

Here is the distinction that matters, stated as plainly as I can:

Notice that the left column is not a soft version of the right. It is a different activity. On the left, the scope or the terms change and the price follows. On the right, only the price moves. That is the test, and it works in every situation I have ever been in.

Back to the hook. When Tomás told his estimator "ask them three questions and do not tell them anybody's number," the three questions were: Is medical gas in or out? Is independent test and balance in or out? Are duct cleaning and final filters in or out? Ashfield came back $348,000 higher on their own scope, because they had genuinely missed scope. Kestrel never revealed Cardinal's number, and Cardinal was awarded on merit at $5,318,000. That is what the left column looks like at 12:26 p.m. on bid day.

🏗️ From the field. Margo Deacon puts it in one sentence, which she has said to me more than once: "Every dollar you squeeze out of a sub in the trailer, you buy back in the building — with interest, and at a time of their choosing."

13.6 Self-perform work

Kestrel self-performs cast-in-place concrete, rough carpentry, and general trades. On Northgate that is $2,200,000 of cast-in-place work inside the $3,500,000 Division 03 line, plus roughly half the $760,000 Division 06 line.

Estimating self-perform work is a different act from tabulating a subcontract bid, and the difference is not technique. It is ownership.

When you take a subcontractor's number, you have bought a price. If their crew is slow, that is their loss. When you self-perform, you have bought a plan — a crew composition, a productivity rate, and a duration — and if the crew is slow, that is your loss, dollar for dollar, with nobody in between.

Worked: the Northgate slab on grade

33,000 SF of slab on grade at 5 inches — 510 cubic yards — placed by Jamal Foster's crews in five placements.

The estimate carries slab on grade at $8.15/SF, or $268,950. The $2.02/SF difference — $66,680 — is self-perform supervision (Jamal's allocated time, which is charged to the work and not to general conditions), self-perform small tools, a productivity-risk contingency, and Kestrel's self-perform markup.

Now the lesson, and it is a hard one.

That $66,680 is Kestrel's to win or lose. If the crew places 6,600 SF a day as planned, Kestrel keeps it. If the crew averages 5,300 SF a day — a 20% productivity miss, which is an ordinary bad month, not a catastrophe — the job takes 6.2 days instead of 5, labor goes from $23,100 to $28,644, and the pump goes from $10,750 to $13,330. That is $8,124 gone from a single small placement. Scale that miss across $2,200,000 of self-perform concrete and you have eaten the entire self-perform margin and started on the fee.

When you subcontract, that risk belongs to the subcontractor. When you self-perform, it belongs to you. Self-perform is not a way to make money. It is a way to take a risk you believe you can manage better than the market can price it.

The honest reason contractors self-perform

Ask around and you will be told self-perform is about margin. In my experience that is the second reason at best.

The real reason is control of the critical path. Foundations and the slab feed steel; steel feeds deck; deck feeds enclosure. When that early chain belongs to a subcontractor with three other jobs, you can only ask. When it belongs to your own superintendent, you can decide — you can add a crew Thursday, you can place on a Saturday, you can resequence a pour because the rebar inspection slipped. On Northgate, that ability was worth more than the concrete margin, and everyone on the job knew it.

The secondary reasons are real too: you keep a core craft workforce employed between jobs (which is how you still have a foreman when the market turns), you retain genuine knowledge of what work actually costs (which is how you know a subcontractor's number is wrong), and you have a credible in-house benchmark for every concrete bid you will ever level.

⚠️ Safety alert. Look back at the drywall scope sheet, item 15: hoisting and material distribution. Two of three bidders excluded it. Now imagine an estimator who levels the bid, misses that line, awards the subcontract without it, and hands the field a job where nobody owns hoisting.

What happens is not that the material stops moving. What happens is that crews improvise — an overloaded personnel hoist, a pallet of board on a scaffold platform never designed for it, a trade borrowing another trade's equipment without training. In week 34 on Northgate, on the north elevation at level 3, a frame scaffold was partially modified overnight by a trade that had not erected it, to run conduit. A plank was lifted and not re-secured. Milo Serrano, a mason tender, stepped onto it at 7:20 a.m. and went down onto the platform, catching himself on the top rail. No injury — that time.

Bea Salgado's investigation found three failures, and the third was a crew running behind after the steel acceleration under an unwritten pressure to make it up. But the first two were ownership failures: no competent-person inspection tag for that shift, and a scaffold modified by a trade that did not erect it and did not re-inspect it.

Ownership of temporary equipment starts in the estimate. If scaffold, hoisting, and material handling are not explicitly assigned to a named party in a priced line, they are assigned to whoever is closest and most desperate at 7:20 in the morning. That is not a scope gap. That is a hazard you bought. Chapter 22 covers the technical requirements; Chapter 24 covers why the system, not the rulebook, produces safety.

13.7 Equipment costs

Equipment is the part of the estimate that most reliably separates people who have run a job from people who have not, because equipment cost is not a price you look up. It is a rate you build, and the rate depends on facts about your company that no catalog knows.

The two halves: ownership and operating

Ownership cost accrues whether the machine runs or sits in the yard:

Operating cost accrues only when the machine runs:

Operator cost is separate from both and is usually carried with the labor for the activity, not with the machine. State clearly in your basis of estimate which convention you used — the single most common equipment estimating error I see is a rate that includes the operator being applied to an activity whose labor already includes the operator. You just paid for that person twice.

Mobilization and demobilization is its own line. A lowboy for a 45,000-pound excavator is not free, and on an urban site with a tight north property line it may also involve a permit, a police detail, and a Sunday morning.

Standby and idle cost is the one nobody carries and everybody pays. A rented machine on your site during a two-week weather delay costs full rent and produces nothing. An owned machine costs ownership cost — depreciation is hour-based, but interest, taxes, insurance, and storage keep accruing on the calendar. Standby is exactly the kind of exposure the construction contingency exists to absorb, and it is exactly the kind of exposure people forget to name when they write the risk register in Chapter 6.

Worked: own or rent a telehandler for Northgate?

Northgate needs a 10,000-pound, 55-foot rough-terrain telehandler for roughly 14 months and 1,150 machine hours — material distribution from mobilization through the interiors phase. Kestrel owns three. Should the job take one, or rent?

Step 1 — build the owned rate.

Machine: purchase price $198,000; useful life 8 years or 8,000 hours; salvage $52,000; Kestrel's fleet averages 1,000 hours per machine per year; cost of capital 7.5%; taxes, insurance, and storage 4.0% of average value.

Average value = ($198,000 + $52,000) ÷ 2 = $125,000

Kestrel's internal rental rate — the rate the equipment division charges a job — is the owning-and-operating cost plus a load for shop overhead, transport between jobs, and the weeks the machine sits idle between assignments. Kestrel uses 15%:

$57.42 × 1.15 = $66.03/hr, published as $66.00/hr.

Step 2 — price both options over the real duration.

Owning is $14,054 cheaper — about 15%.

Step 3 — find the break-even, and then find the assumption that controls it.

Per month, owned costs about $5,420 and rented costs about $6,521, a $1,101/month advantage to owning. Owning starts $1,350 behind on mobilization ($2,800 versus $1,450). Break-even: $1,350 ÷ $1,101 = 1.23 months. Past about five weeks of use, owning wins.

Which sounds decisive, and is not, because the entire result depends on one assumption buried in step 1.

🔍 Why this works — and where it stops working. Re-run the owned rate at 600 hours per year of fleet utilization instead of 1,000. Depreciation does not change, because it is charged per hour of use. But interest, taxes, insurance, and storage are annual costs spread over fewer hours:

At 600 hours a year, owning costs $91,408 against renting's $92,754. The $14,054 advantage collapses to $1,346 — inside the noise of any of these assumptions.

Utilization, not purchase price, decides own versus rent. The machine costs the same either way. What changes is how many hours you have to spread the annual carrying costs across. This is why a contractor's fleet decision is a company decision made across all jobs, not a project decision made on one — and why the job that takes the machine out of the yard for fourteen months is doing the fleet a favor that never shows up on that job's cost report. Chapter 21 takes this all the way through crane selection, fleet strategy, and matching equipment to the work.

🔄 Check your understanding. Your excavation subcontractor's schedule slips and your rented 45,000-pound excavator sits on site for eleven working days doing nothing. It is rented at $9,400 per month. What does the standby cost, and who pays it?

13.8 Markups, assembled properly

You have a cost of work, general conditions, and equipment. Now you turn cost into price. This is where estimates most often go quietly wrong, because each markup has its own base and its own arithmetic, and applying the wrong base to the right rate produces a number that looks right and is not.

Bond premium: a rate on contract value, on a sliding scale

Surety bond premium is quoted per thousand dollars of contract value, and the rate declines as the contract gets larger — the surety's underwriting cost does not scale linearly with exposure. Rates vary by surety, by contractor's financial strength, and by market, so treat this scale as illustrative of the structure, not as a quote.

Effective rate: $425,000 ÷ $47,500,000 = 0.895% of contract value.

Note the base: the bond is priced on the full contract value, including your own fee. Bond, price, fee, and bond again is a small circular reference that estimating software handles and hand-built spreadsheets often do not. Two practical consequences: if the contract value moves, the bond premium moves, and a change order that adds $1,000,000 of work adds roughly $8,250 of bond premium that you must remember to include in the change-order price.

The Northgate insurance and bonds line, itemized

That deductible reserve is worth a sentence. Builder's risk carries a deductible — often substantial, and often larger for water damage and wind. If a pipe lets go on level 3 in February and destroys $340,000 of finishes, insurance pays the loss above the deductible and you pay the deductible. Carrying a named reserve for it is the difference between a managed risk and a surprise. Note also that in some jurisdictions and on some projects the owner purchases builder's risk; read the supplementary conditions before you carry it, and if the owner carries it, read the policy anyway to learn what the deductible is and who owes it.

Contingency and escalation

Contingency on Northgate is 3.0% of the $43,800,000 of cost of work plus general conditions plus insurance and bonds, carried at $1,320,000. It is drawn down against named risks from the risk register, with a log, and what is left at the end splits 75/25 with Meridian.

Escalation is $575,200, built in §13.2 against the $26,400,000 of scope that was not yet under a firm quote when the GMP was signed. The rule is: escalate what you have not bought, for the period until you buy it. Escalating the whole cost of work would have double-counted the third of the job already locked by firm quotes — a $190,000-ish error in the owner's disfavor that a competent owner's estimator will find and that will cost you credibility you need later.

Markup versus margin — the same distinction, at full scale

Chapter 12 drew this on a single line item. Here it is on a $47.5 million contract, where the difference is real money.

Markup is a percentage added to cost. Margin is that same money expressed as a percentage of price. They are never equal, and margin is always the smaller number.

Northgate's fee is a 4.0% markup on the $45,120,000 subtotal:

$45,120,000 × 0.040 = $1,804,800, giving a price of $46,924,800 before escalation.

As a margin on that price: $1,804,800 ÷ $46,924,800 = 3.846%. As a margin on the full GMP of $47,500,000: $1,804,800 ÷ $47,500,000 = 3.80%.

Now suppose Nadia Haddad had said "I want a 4% margin, not a 4% markup." Then:

Price = Cost ÷ (1 − margin) = $45,120,000 ÷ 0.96 = $47,000,000, and the fee is $1,880,000.

$1,880,000 − $1,804,800 = $75,200.

💰 Money check. Seventy-five thousand two hundred dollars, on one word. That is more than the entire Northgate final-cleaning line ($81,840) and about the same as the site security budget for the whole job. If you and your executive are using the words "markup" and "margin" interchangeably in the final review meeting, one of you is about to be surprised, and it will be discovered in month sixteen when the fee does not cover what somebody thought it covered.

The markup stack, in order

📊 Diagram (described). Picture the estimate as a stack of layers, each one sitting on a specific base. Reading up from the ground:

  ┌──────────────────────────────────────────────────────┐
  │  GUARANTEED MAXIMUM PRICE            $47,500,000     │
  ├──────────────────────────────────────────────────────┤
  │  + Escalation      $575,200   (base: unbought scope   │
  │                                $26,400,000)           │
  ├──────────────────────────────────────────────────────┤
  │  + Fee @ 4.0%    $1,804,800   (base: $45,120,000)     │
  ├══════════════════════════════════════════════════════┤
  │    SUBTOTAL                    $45,120,000            │
  ├──────────────────────────────────────────────────────┤
  │  + Contingency   $1,320,000   (base: $43,800,000)     │
  ├──────────────────────────────────────────────────────┤
  │  + Ins. & bonds    $900,000   (base: contract value,  │
  │                                incl. the fee)         │
  ├──────────────────────────────────────────────────────┤
  │  + Gen. conds.   $2,900,000   (base: 565 days)        │
  ├──────────────────────────────────────────────────────┤
  │    COST OF WORK               $40,000,000             │
  │    (base: quantities × unit costs, and 41 subcontract │
  │     packages)                                          │
  └──────────────────────────────────────────────────────┘

The teaching point is in the parentheses. Every layer has a different base. General conditions rest on time. Insurance and bonds rest on contract value, which includes the fee sitting above them. Contingency rests on cost. Fee rests on the subtotal. Escalation rests on the unbought portion only.

Get the rate right and the base wrong and you have an error that survives every review, because the rate is defensible and nobody re-derives the base. I have seen a bond premium computed on the cost of work instead of the contract value on a $60 million job. The error was $62,000, it lived in the estimate for four months, and it was found by a twenty-four-year-old who asked what number the surety would actually invoice against.

13.9 Bid day

Everything above is craft. Bid day is a process-control problem, and the failures on bid day are almost never estimating failures. They are clerical failures at high speed, and they are enormous.

The war room

📊 Diagram (described). Kestrel's bid-day layout, which has not changed in fifteen years because it works:

        ┌───────────────────────────────────────────────┐
        │   PROJECTED BID SUMMARY  (the wall)           │
        │   Div | Package | Budget | Low | Carry | Δ    │
        └───────────────────────────────────────────────┘
                            ▲
                            │ ONE person updates this.
                            │ Only one. Ever.
   ┌────────────┐   ┌───────┴────────┐   ┌────────────┐
   │  PHONE 1   │   │  BID CAPTAIN   │   │  PHONE 2   │
   │  quote     │──▶│  (Tomás)       │◀──│  quote     │
   │  intake    │   │  decides what  │   │  intake    │
   └────────────┘   │  gets carried  │   └────────────┘
   ┌────────────┐   └───────┬────────┘   ┌────────────┐
   │  PHONE 3   │──────────▶│◀───────────│  SCRIBE    │
   │  quote     │           │            │  (Dani):   │
   │  intake    │           ▼            │  time,     │
   └────────────┘   ┌────────────────┐   │  name,     │
                    │ FINAL REVIEW   │   │  number,   │
                    │ Nadia Haddad,  │   │  exclusions│
                    │ T-45 minutes   │   └────────────┘
                    └───────┬────────┘
                            ▼
                    ┌────────────────┐
                    │  BID DELIVERY  │
                    │  T-30 on site  │
                    └────────────────┘

Three rules are doing all the work in that diagram.

One person touches the summary. Quote takers write on paper and hand it over. They do not type into the master. A spreadsheet with four people in it on bid day is a spreadsheet that will contain a number nobody can explain.

Every quote is logged with time, company, number, and stated exclusions, before it is evaluated. Dani's job in the hook was not clerical busywork. It is the audit trail that lets you reconstruct at 1:15 p.m. why the electrical carry moved.

The bid leaves early. T-30 on site, minimum. Traffic, elevators, a locked door, a receptionist at lunch — a bid that is technically finished at 1:58 and physically outside the building at 2:00 is worth nothing.

The last-minute quote problem

Subcontractors submit late for the same reason contractors do: to limit the time available for shopping. A quote that lands at 1:47 p.m. for a 2:00 p.m. bid is a business decision, not a discourtesy, and you have to be able to handle it.

You handle it with a plug number — a carried value for a package you do not yet have a real, leveled quote for. Plugs are legitimate and universal. They are also how people get hurt, so:

Spreadsheet discipline

I am going to be blunt: more money has been lost on bid day to a spreadsheet error than to a bad unit price. A transposed digit that turns $1,850,000 into $1,580,000 is a $270,000 mistake that no amount of estimating skill will catch, because the number looks fine.

The controls that catch it:

1. Sum every column two ways. Total the divisions and, separately, total the subcontract packages. They must agree. When they do not, you have a package assigned to no division or to two.
2. Print the summary and check it against paper. The eye catches on paper what it slides over on a screen. This sounds antique. Do it anyway.
3. Lock the formula cells. Every catastrophic bid-day error I know of involved somebody typing a number into a cell that used to contain a formula.
4. Reconcile against $/SF. $47,500,000 ÷ 132,000 SF = $360/SF. If your comparable outpatient buildings run $310–$385/SF, you are in the band. If your summary says $268/SF, do not congratulate yourself — go find what fell out. This is the same sanity check as the conceptual estimate you built in Chapter 11, used in reverse: the conceptual number now audits the detailed one.
5. Reconcile against the previous estimate. If the design-development estimate was $44.1 million and the construction-document estimate is $47.5 million, you must be able to explain the $3.4 million in a short list of named causes. "The design got more expensive" is not a cause. "Curtain wall went from stick-built to unitized at the owner's direction, $1,180,000" is.

The bid-day checklist

That "what did we forget" ritual on the second-to-last line is not a formality. Everybody in the room speaks — the quote takers, the scribe, the estimator who did earthwork, the person who has been quiet for four hours. Tomás runs it the same way every time: "Go around. One thing each. If you have nothing, say 'nothing.'" I have watched a twenty-three-year-old field engineer say "we never got a number for the temporary chain-link along Ferry Street" and save $71,000.

The final review, and the decision to move the number

Here is the honest part of this chapter, and the part textbooks usually skip.

The estimate produces a cost. It does not produce a bid. Between them sits a judgment about market, competition, backlog, and appetite — and that judgment belongs to somebody with the authority to lose the company's money. At Kestrel it is Nadia Haddad, Vice President of Operations. Nobody else moves the number. Not the estimator, not the PM, not the person on the phone at 1:52.

The legitimate reasons to move a number down:

Every one of those has a name and a mechanism. That is finding savings.

And then there is the other thing.

Buying the job is reducing the number without a corresponding change in scope, risk, or expectation — taking cost out of a line you still intend to spend, because you want the work. It is not fraud, it is not illegal, and it is extremely common. It is also, in my experience, the most reliable way to lose money in this industry.

💰 Money check — what buying work actually costs. Curtis Boone's Rivermont Elementary School #12 bid is the worked example, and case-study-02.md walks it in full.

Curtis's built-up number was $22,865,000. He and his team "found" $465,000 in the final review and bid $22,400,000. He won by $312,000 over the second bidder. Here is what the four shaved lines cost during construction:

The shave was $465,000. The cost was $998,000 — $2.15 back out for every dollar taken.

Now put it against the fee. Rivermont Elementary carried an anticipated fee of 3.5% on $22,400,000 = $784,000.

$998,000 − $784,000 = $214,000 lost. Not "made less than we hoped." Lost. Kestrel finished a two-year job, tied up a superintendent, a PM, a bonding line, and a piece of its aggregate capacity, and paid $214,000 for the privilege.

And the recovery arithmetic is the part that should stay with you. To earn back $214,000 at a 3.5% fee, Kestrel must go win and successfully perform:

$214,000 ÷ 0.035 = $6,114,286 of additional work — at full fee, with no overruns.

A $465,000 shave in a forty-minute meeting created a $6.1 million hole to dig out of. That is the real price of buying work, and it is why Nadia Haddad's most valuable skill is not knowing when to sharpen a number. It is knowing when to say "we're not the low bidder on this one, and that's the right answer."

Unbalanced bidding and front-loading

Two related practices sit on the same ethical boundary, and you will meet both.

Unbalanced bidding is loading price into some line items and stripping it out of others while keeping the total unchanged. On a unit-price contract like Cottonwood Creek Bridge, this can be genuinely predatory: if you believe the actual quantity of structural excavation will far exceed the engineer's estimate, you load that unit price and strip the ones you expect to under-run. Your total bid looks competitive; your actual revenue is materially higher. Some owners' bid documents expressly reject materially unbalanced bids, and public owners routinely reserve the right to do so.

Front-loading is the schedule-of-values version: assigning more value to early activities — mobilization, foundations, underground — and less to late ones, so you get paid faster than you spend. On Willow Street's $6.8 million, moving $180,000 of value from finish activities into mobilization and foundations might improve your financing position by roughly $30,000–$35,000 over the job. It is also a representation to the owner that mobilization is worth what you said it is worth, and it is not.

⚖️ What the contract says. Most owner-favorable agreements require the schedule of values to be submitted for the architect's approval and to "allocate the contract sum to the various portions of the work" — language that a materially front-loaded SOV does not satisfy. Architects reject them routinely. More seriously, the practice leaves an unbilled liability at the end of the job: you have collected for value you have not created, so the last 10% of the work has 4% of the money behind it, and you finish the punch list on your own dime with a subcontractor who has already been paid. When jobs go bad, a front-loaded SOV is also the first exhibit in the owner's argument that your billing was never trustworthy — which will follow you into every claim you file. Chapter 32 builds an honest schedule of values and shows what happens to the dishonest one.

There is a legitimate neighbor to this practice, and it is worth naming so you can tell them apart: genuine early costs belong in early line items. Mobilization really does cost money. Bonds are paid up front. Long-lead deposits are real. Putting real early costs in early line items, and being able to substantiate every one of them, is correct estimating. Putting finish money in early line items is something else.

13.10 The estimate's second life

The estimate does not end on bid day. Roughly two weeks after award it comes back to life as three different documents, and how much pain that transition causes was decided months earlier by how you structured the estimate.

Here is a slice of the Northgate mapping, so you can see the shape:

Three rules make this transition survivable, and all three are decisions you make while building the estimate, not after:

1. Estimate at the level of detail you intend to control at — no finer, no coarser. A cost code you cannot get accurate field data against is worse than no code, because it produces confident garbage. If nobody is going to separate "interior partitions, level 2" from "interior partitions, level 3" on a timecard at 3:30 p.m. on a Friday, do not create the code.
2. Keep self-perform work coded by labor, material, and equipment separately. Subcontracts can be one line — they are a fixed price and you either owe it or you do not. Self-perform work is where you find out whether you are making money, and you cannot find that out from a lump.
3. Make the schedule of values and the estimate reconcilable. Not identical — the SOV is negotiated with the owner and is organized for progress measurement, so it gets subdivided by area and floor. But you must be able to walk from any SOV line back to the estimate lines behind it. When the owner asks in month nine why interior partitions are billing at 62% while your cost report says 71% spent, you need to answer in ten minutes, not three days.

🔄 Check your understanding. Two estimators build the same $40 million cost of work. One produces 180 line items; the other produces 2,400. Which is better?

📋 Try it: level three masonry bids

The Northgate masonry package carries a Kestrel budget of $890,000 (Division 04). Scope: CMU shaft and stair enclosures, interior CMU at mechanical and electrical rooms and the loading dock, brick veneer at the entry canopy, cast stone sills and copings, masonry reinforcing and grout, flashing and weeps, control joints, cleaning and sealing, one mockup panel, and burnished CMU at the lobby stair.

Three bids arrive:

Kestrel's carried values for the gaps: scaffolding $22,000 · hoisting and distribution $12,000 · flashing and weeps $22,000 · cast stone $28,000 · cold-weather protection $37,000 · mockup $11,000 · cleaning and sealing $16,000 · daily cleanup $14,000 · layout $6,000 · saw-cutting $9,000 · burnished CMU $23,000. Addendum 3 added CMU at the new generator enclosure — 1,850 SF, valued at $52,000. A subcontractor bond, where not included, costs 1.2% of the leveled subtotal.

Your tasks:

1. Level the three bids into a comparable table.
2. Identify the true low bidder and the variance to Kestrel's $890,000 budget.
3. Name the three scope gaps most likely to have become change orders if the apparent low bid had simply been carried.
4. State what you would write into the subcontract to close them.

Spaced Review

Answer these from memory before you check anything.

From Chapter 12: What are the three factors in a unit-cost build-up for labor, and why is a productivity rate the most dangerous of the three?

Also from Chapter 12: State the markup-versus-margin distinction, then apply it: Northgate's fee is a 4.0% markup on $45,120,000. What is it as a margin on the $47,500,000 GMP?

From Chapter 11: What is a basis of estimate, and what does escalation actually cover?

Deep callback to Chapter 4: On a GMP contract, what is the difference between the cost of the work, general conditions, and the fee — and which of the three is the owner allowed to audit?

Project Checkpoint: The Willow Street Detailed Estimate

In Chapter 12 you produced a quantity takeoff for the Willow Street Community Center — sitework, concrete, masonry, and structure, with waste factors applied. Now you turn quantities into a number.

Your deliverable: a complete detailed estimate for the Willow Street Community Center, 24,000 SF, two stories, wood-framed second floor over a structural steel and CMU first floor, design-bid-build lump sum, 425 calendar days, prevailing wage, 100% payment and performance bonds, liquidated damages $1,200 per calendar day. The full project package — program, drawings described, specification excerpts, and quantity data — is Appendix K.

Build these five components.

1. An estimate summary by CSI division. Every division from 02 through 33 gets a row, including the empty ones with a dash. Price your Chapter 12 self-perform quantities at unit cost; carry the rest at package level.

2. Bid tabs for five subcontracted trades. Choose five with real money in them — earthwork, masonry, structural steel, roofing, mechanical, electrical are the obvious candidates. For each: write the scope sheet first, invent three bidders with genuinely different inclusions and exclusions, then level them. Do the leveling arithmetic in a table exactly as §13.4 does. Write the scope sheet before you invent the bids. Doing it in the other order teaches you nothing, because you will unconsciously write bids that level cleanly.

3. An itemized general-conditions estimate with a derived burn rate. Staff first, with duration and monthly rate; then facilities, services, and consumables. Then divide by 425 calendar days and state the daily rate you would put in the contract. For scale: a $680,000 general-conditions estimate on Willow Street produces exactly $1,600 per calendar day, and adding the $1,200/CD liquidated damages gives a total exposure of $2,800 per calendar day for slipping substantial completion. Compare that to Northgate's $10,650/CD and notice that the ratio to contract value is not far off — that is not a coincidence.

4. All markups, in the right order, on the right bases. Bond premium on the full contract value using a sliding scale. Insurance as a rate. Contingency on cost. Fee as a markup — and state the resulting margin separately, so you know both numbers. Escalation on the unbought portion only. Reconcile to a final bid number.

Here is a target skeleton. Yours will differ; if it differs by more than about 8% on any line, go find out why:

$6,800,000 ÷ 24,000 SF = $283.33/SF. Sanity-check your own number against that.

5. A basis-of-estimate page. One page. Documents and their dates, addenda acknowledged, quantities and their source, inclusions, exclusions, allowances (with dollar values), assumptions about schedule and construction duration, assumed market conditions, and the named risks carried in contingency. If you can only produce four of the five components this week, produce this one and the general conditions. An estimate without a basis of estimate is a rumor with a dollar sign.

Next chapter you build the CPM schedule — activity list, logic, durations, forward and backward pass, critical path, and Gantt. Keep your general-conditions duration assumption where you can reach it. The schedule you build in Chapter 14 will either confirm the 425 days you just priced, or it will not, and finding that out on paper costs nothing.

Chapter Summary

A reference card for the next time you are standing in front of a forty-foot spreadsheet at 11:52 in the morning.

The eight things that decide whether a detailed estimate is any good:

The Northgate reconciliation, for reference:

The four-question test before any number leaves your desk:

1. Is every division on the summary? Including the ones with a dash.
2. Is every package covered by at least two leveled bidders, or by a flagged and loaded plug? Say the plug total out loud.
3. Does every markup sit on the correct base? General conditions on time. Bond on contract value including fee. Escalation on unbought scope only.
4. Can you defend every line to somebody who wants it to be lower? If a line cannot survive the question "why is this here," it will not survive the final review — and if it disappears there without a named reason, you did not find a saving. You bought the job.

The ethical line, in one sentence: on the legitimate side, the scope or the terms change and the price follows; on the other side, only the price moves.

What's Next

You have a number. It rests on an assumption you have not tested yet: that this building can be built in 565 calendar days. Chapter 14 makes you prove it — activity list, logic, durations, the forward and backward pass, and the critical path, which is a calculated result rather than a management opinion. That chapter also puts a price on the steel delay you have been hearing about since Chapter 1, using the $10,650 per calendar day you derived here. Then Chapter 15 takes the estimate and the schedule together into the bid decision itself — including how to know when the right answer is to walk away.