Chapter 11 — Preconstruction Services: Conceptual Estimating, Budgeting, and Value Engineering

The Hook: Four Point Eight

Sixteen months before anybody would put a shovel in the ground at Northgate, I sat in a conference room on the fourth floor of Meridian Health System's administration building and watched a number ruin somebody's morning.

The number was $52,300,591.

Tomás Reyes, Kestrel's chief estimator, had put it on the screen himself. He does not hand off bad news. He had spent five weeks with the schematic design set from Halvorsen + Pike — forty-one sheets, a twenty-two-page outline specification, no details worth the name — and he had built an elemental estimate of the Northgate Outpatient Pavilion: four stories, 132,000 gross square feet, outpatient clinics, an imaging suite, ambulatory surgery, a café and a lobby, on a sloping 6.2-acre site with an active clinic on the other side of the north property line.

Priyanka Sethi, Meridian's owner's representative, looked at the number for about eight seconds. Then she looked at the other number, the one on the whiteboard behind her, which her board had approved in February and which the health system's capital plan, its bond covenants, and its interim clinic lease were all built on top of.

$47,500,000.

"So we're four point eight over," she said.

"Four million, eight hundred thousand, five hundred ninety-one," Tomás said. He is not being difficult when he does that. He is telling you the number is a real number and not a vibe.

Pri did the thing every owner does. I have watched maybe sixty owners do it and they all do it in the same tone of voice, somewhere between hopeful and already tired.

"Can you just value-engineer it?"

Here is what I said, and I want you to hear the whole thing, because the rest of this chapter is inside it.

"Yes. And no. If what you mean is can we find four point eight million dollars of cost that isn't buying you anything — yes, we can find a lot of that, and I'll show you where in nine days. If what you mean is can we take four point eight million dollars of building out and pretend nothing changed — no, and neither can anyone else, and anyone who tells you they can is going to hand you the difference back later as a change order with your name on it. Those are two different activities. One of them is my job. The other one is malpractice with a spreadsheet."

Pri wrote that down. Dale Whitcomb, the project architect, exhaled through his nose in a way that meant finally.

Nine days later we came back with a log of eighteen priced options. Meridian accepted thirteen of them and rejected five, and one of the five they rejected was the one that would have "saved" two million dollars in a single line and cost them four times that before the roof was on. Eleven months after that, one of the options they accepted came back and bit us for seventy-one thousand dollars, and I will show you that too, because a chapter that only shows you the wins is a brochure.

That nine days is what this chapter is about. It is the least visible work a construction manager does and it is worth more than anything they will ever do in the field.

Theme 3, and this is its home chapter: the project is built twice. Once on paper — in estimates, models, schedules, coordination drawings, and submittals — and once in steel and concrete and copper. The first build determines the second. By the time a crew is standing in front of the problem, the cheap fixes are all gone. A bad structural grid costs three weeks of Ruth Caldwell's time to fix in design and eleven weeks of Margo Deacon's life to fix in the field, and only one of those has a change order attached to it.

🏃 Fast Track: If you already estimate for a living, skim §11.3 (you know the methods) and read §11.2 on estimate classes, §11.4 on escalation to the midpoint of construction, §11.7 on the difference between value engineering and cutting, and §11.9 on the qualifications-and-assumptions page. The reconciliation table in §11.9 is the one thing in this chapter I would put on a wall.

🔬 Deep Dive: Unit costs, productivity, waste factors, and the CSI division structure live in Appendix C. The arithmetic conventions — compounding, indices, unit conversions — are in Appendix A. Detailed quantity takeoff is Chapter 12 and the full bid build-up is Chapter 13. The lean alternative to estimate-then-cut is Target Value Design, previewed in §11.8 and developed in Chapter 27.

Everything in this chapter — Kestrel, Meridian, Northgate, and every person in it — is a composite drawn from real projects and assembled so the numbers hang together. The numbers are illustrative. The mechanics are not.

11.1 What Preconstruction Services Actually Are

Preconstruction services are the design-phase deliverables a construction manager sells to an owner before there is a contract to build anything. On a CM at Risk (CMAR) job like Northgate — the delivery method we unpacked in Chapter 3 — the owner hires the construction manager early, during design, under a preconstruction agreement, and only later converts that relationship into a construction contract with a Guaranteed Maximum Price (GMP).

Most people outside the industry think a contractor's job starts at Notice to Proceed (NTP) — the owner's written instruction to begin work. It doesn't. On Northgate, Kestrel was under contract sixteen months before NTP, and by the time the first excavator rolled onto the site, we had already made most of the decisions that determined whether the job would make money.

Here is the actual menu. Not the marketing version — the deliverables that show up in a preconstruction agreement's scope of services.

Kestrel's preconstruction fee on Northgate was $185,000 — a fixed fee, paid monthly over the design period, credited against nothing. Meridian paid it out of the soft-cost line of a $61,000,000 total project budget. We will come back to that budget in §11.5, because a construction manager who does not understand the owner's whole budget gives confident, expensive, wrong advice.

Long-lead identification is a schedule service disguised as a procurement service

The service on that list students underrate most is long-lead identification — the earliest place the schedule and the budget become the same conversation, Theme 2 showing up before there is a schedule.

Those ranges have moved violently in both directions in recent years. Do not use them as prices or promises. Call the actual supplier, get a written lead time for the month you intend to release, and back-schedule from it — that exercise is Chapter 14, and the buyout that executes it is Chapter 16.

🏗️ From the field. We identified the steel lead time correctly on Northgate. We wrote it in the log. Then the anchor-bolt submittal sat in Kestrel's own office for eleven days before it reached Caldwell Structural, Ruth Caldwell took her full contractual fourteen days, Ironbridge Steel missed its mill rolling slot, and erection started August 27 instead of August 4 — twenty-three calendar days late, on the critical path, at $10,650 per calendar day of combined extended-general-conditions and liquidated-damages exposure. Identifying the long lead is necessary and nowhere near sufficient.

💰 Money check — what preconstruction is worth. Kestrel's precon fee: $185,000. Of that, the constructability review was roughly 220 hours of estimator, superintendent, and VDC time — call it $38,000 of Kestrel's actual cost. That review produced eight findings (§11.6) with a combined avoided exposure of $435,000 and up to 21 calendar days. Twenty-one days at $10,650/CD is another $223,650 of schedule exposure. So a $38,000 slice of a $185,000 fee bought roughly $658,650 of avoided cost and risk — about 17 to 1 on the review alone, and about 3.6 to 1 on the entire preconstruction fee. That ratio is why owners buy preconstruction, and it is the argument you make when someone asks why they are paying a contractor before the contractor builds anything.

🔄 Check your understanding. An owner tells you: "I'll bring you on at 100% construction documents. Design is the architect's job; I don't need a contractor's opinion until there's something to price." Name two specific, dollarized things that owner has already lost.

11.2 Design Phases and Estimate Classes: Accuracy Is a Function of Information

Here is the single most important idea in conceptual estimating, and it is the one that gets estimators fired when nobody has explained it to the owner:

An estimate's accuracy is a function of how complete the design is, not of how hard the estimator worked.

You cannot out-effort missing information. If the drawings do not say what the exterior wall is, no amount of care tells you what the exterior wall costs. You can price the range. You can price your best judgment of where in the range it will land. You cannot price a thing that has not been decided.

The industry has a formal way of saying this. AACE International — the association for cost engineering — publishes a cost-estimate classification system that most of the construction and process industries follow, sorting estimates into Class 5 (earliest, least information) through Class 1 (latest, most information), and tying each class to a range of expected accuracy. The specific ranges differ between the building-construction and process-industry versions of the recommended practice, and they get revised. Look up the current AACE recommended practice for the authoritative definitions. What follows is a plain-language version calibrated to building work, and the percentages are illustrative typical ranges, not rules.

DESIGN COMPLETENESS ──────────────────────────────────────────────────────►
  0%          2%          15%          40%           75%          100%
   │           │            │            │             │             │
   ▼           ▼            ▼            ▼             ▼             ▼
PROGRAM     CONCEPT        SD           DD         50–90% CD    100% CD / BID
   │           │            │            │             │             │
Class 5     Class 5/4     Class 4      Class 3      Class 2       Class 1
   │           │            │            │             │             │
 $/SF        $/SF +     Elemental    Elemental +   Detailed     Full takeoff
capacity   functional   (UniFormat)   assemblies   quantities   + sub quotes
 factor       unit           │            │             │             │
   │           │            │            │             │             │
 -30/+50%   -25/+40%    -15/+25%     -10/+20%     -7/+15%      -3/+10%
   │           │            │            │             │             │
DESIGN CONTINGENCY TYPICALLY CARRIED:
  15%         10%          7%           5%            3%            0%
   │           │            │            │             │             │
   └───────────┴────────────┴────────────┴─────────────┴─────────────┘
        the band narrows because the DRAWINGS get better,
                  not because the estimator does

📊 Diagram (described). Read that ladder left to right as a funnel. Each step right moves decisions out of the "unknown" column and into the "priced" column. The design contingency in the bottom row is the price of the remaining unknowns, and it shrinks for exactly one reason: there are fewer of them.

Now go back to the hook. Tomás's schematic-design estimate is a Class 4 estimate: $52,300,591, with an honest band of roughly −15% to +25%.

• Low end: $52,300,591 × 0.85 = $44,455,502
• High end: $52,300,591 × 1.25 = $65,375,739

Stare at that for a second. The band around the estimate is $20.9 million wide. The gap everyone is arguing about is $4.8 million. The gap is smaller than the estimate's own uncertainty.

That is not a reason to shrug. It is a reason to be precise about what you are actually doing. You are not "correcting an error." You are steering a distribution — moving the whole band down and narrowing it, decision by decision, until the day you have to put a guaranteed number on it and the band collapses onto a single value that Kestrel owns.

🔍 Why this works. Why can a parametric estimate at 12% design land within ±20% when the estimator has never seen a detail? Because buildings are not arbitrary. The cost of a building system is tightly coupled to a few measurable parameters that are known early: gross area, floor-to-floor height, number of stories, perimeter, skin-to-floor ratio, structural bay, program mix, and code occupancy. A four-story steel-framed outpatient building with a 0.45 skin ratio has a cost structure much like other four-story steel-framed outpatient buildings with a 0.45 skin ratio, whatever the door hardware turns out to be. On top of that, errors partially cancel: across thirty-five elemental lines some rates are high and some are low, and the aggregate is steadier than any single line. Which is exactly why the method fails catastrophically when the building is unusual — a one-off atrium, a vibration-sensitive imaging platform, contaminated soil. Then the errors stop cancelling and start compounding, and you must go find a real quantity.

11.3 Five Conceptual Estimating Methods, Worked on Northgate

11.3.1 Cost per square foot — the crudest method and the one everybody uses

Divide a comparable project's cost by its area, multiply by your area. Done.

At programming — eighteen months before NTP, before there was a floor plan — Tomás gave Meridian $355 per square foot for a 132,000 SF outpatient pavilion.

$355/SF × 132,000 SF = $46,860,000

Meridian's board rounded, added a little, and approved $47,500,000. That is where the budget in the hook came from: one multiplication, done well, done honestly, and then treated by everyone downstream as if it had been carved into a mountain.

Now here is the danger. A dollar-per-square-foot number is meaningless until you know what is inside it. Before you use anybody's $/SF figure — including your own from a project two years ago — answer all four: What area? (gross, net assignable, rentable, or footprint — gross versus net can differ by 30% in a clinical building). What is included? (sitework, parking, utilities to the property line or the building, demolition, design fees, FF&E, owner-furnished equipment, escalation, contingency, fee). What was excluded and never mentioned? (Northgate's owner-furnished imaging equipment is $3,640,000 and is not in the $360/SF; somebody will eventually assume it was.) When, where, and into what market? The same building bid into a hungry market and a booked-solid one can differ by 10–20%.

⚠️ Treat that table as orders of magnitude, not as a price list. These are illustrative ranges for U.S. building construction in recent years, excluding land, design fees, FF&E, and owner's soft costs, and they move with region, year, market, and program. For current, defensible numbers use RSMeans cost data (a commercially published construction cost database, updated annually, with location adjustment factors), the Engineering News-Record (ENR) cost and material indices, and — better than either — your own company's completed-project database, the only source that reflects how your company actually builds.

The most useful thing about $/SF is not the estimate. It is the audit. When Tomás's elemental estimate came out at $52,300,591, the first thing he did was divide:

$52,300,591 ÷ 132,000 SF = $396.22/SF

and ask whether $396 per square foot was a believable number for this building. Against a $280–$420 outpatient range with a full imaging suite and ambulatory surgery in it, at the top but not off the map. If that division had produced $610/SF, he would have gone looking for the double-counted line before he showed it to anybody. Always divide. The ratio catches the blunder the addition hides.

11.3.2 Cost per functional unit

Owners don't think in square feet. Hospital executives think in beds and exam rooms and operating rooms. School boards think in student stations. Hotel developers think in keys. Speak the owner's units and you will be trusted faster.

Northgate has 72 exam rooms, 4 operating rooms, 6 imaging bays, and (after the value engineering you are about to read) 188 surface parking stalls.

| Metric | At the SD estimate ($52,300,591) | At the final GMP ($47,500,000) | |---|---|---| | Per gross square foot | $396.22 | $359.85 — the canonical $360/SF | | Per exam room (whole-building cost ÷ 72) | $726,397 | $659,722 | | Per surface stall (site paving element only) | $3,292 (212 stalls) | $2,766 (188 stalls) |

One caution, and it matters. "Cost per exam room" divides the whole building — lobby, café, imaging, mechanical — by the exam rooms. That is fine as a benchmark as long as everyone knows it, and a catastrophe when somebody uses it to price four additional exam rooms, because the marginal cost of one more exam room inside an existing shell is a fraction of the average. Average cost and marginal cost are different numbers, and confusing them is how owners get talked into things. Past programming, treat the functional-unit method the way a pilot treats a second altimeter: a cross-check, never the primary instrument.

11.3.3 Parametric / elemental estimating with UniFormat — the method that actually works at SD

This is the centerpiece. At schematic design you have systems but not details: you know it is a steel frame, you know roughly how much curtain wall, you know the mechanical concept. UniFormat — a classification system published jointly through CSI and CSC that organizes construction by building element (substructure, shell, interiors, services) rather than by trade — is built exactly for that state of knowledge. Contrast it with CSI MasterFormat, which organizes by work result and is what you use for specifications and detailed bids (Chapter 7 and Chapter 13).

The difference matters: at SD nobody has decided whether the partition is 3⅝-inch or 6-inch stud, but everybody knows there are interior partitions, and you can measure roughly how many linear feet.

Here is the actual Northgate schematic-design estimate. Note that several quantities below are the SD-phase quantities, before value engineering — the structural steel, the curtain wall, the precast, and the earthwork export all change in §11.7, and the post-VE values are the ones that appear everywhere else in this book.

Northgate Outpatient Pavilion — Schematic Design Estimate (Class 4) 132,000 gross square feet · priced at current-day costs · 16 months before NTP

Then the markups, which is where a Class 4 estimate earns or loses its credibility.

Two things in that table deserve a paragraph each.

Why 6% design contingency and not 7%? The ladder in §11.2 suggests roughly 7% at SD. Tomás carried 6% and wrote down why: Meridian's program was unusually well defined, the health system had built two nearly identical clinics in five years, and H+P had designed both. That is a defensible judgment — and exactly the kind that becomes a lie if you don't record it, because in eight months somebody will ask why the contingency was thin and "I felt good about it" is not an answer. Every contingency percentage on every estimate you issue should carry a written reason. That is Chapter 6's rule applied here: contingency is a priced, owned reserve for identified risk, not padding.

The burden multiplier. Look at the ratio of the bottom line to the cost of construction:

$52,300,591 ÷ $43,566,690 = 1.2005

Equivalently: (1 + 0.06 + 0.03 + 0.0643) × 1.04 = 1.2005.

Every dollar you take out of the cost of work takes about $1.20 out of the bottom line, because design contingency, construction contingency, escalation, and fee all ride on top of it. Tomás says it like this in every VE meeting: "Don't chase four point eight million dollars of work. Chase four million dollars of work, and the markups will hand you the other eight hundred thousand."

$4,800,591 ÷ 1.2005 = $3,998,920 of actual work to remove

Hold onto that. It changes what the target feels like.

🔄 Check your understanding. At the design development estimate the design contingency drops from 6% to 4% and escalation is re-forecast from 6.43% to 5.10%, with the fee unchanged at 4%. What is the new burden multiplier, and what does that do to the value of a $250,000 raw saving?

11.3.4 Assembly estimating — where the elemental rate comes from

An elemental rate like "$69.00/SF of architectural precast wall" looks like a guess. It isn't. It is an assembly: a bundle of components, priced together, per unit of the thing you can measure.

Assembly: architectural precast wall with punched aluminum windows — per SF of wall

That is where the $69.00 in element B2010 came from. Not a feeling — a build-up you can defend line by line, and, more importantly, argue about line by line when Dale Whitcomb says the window-to-wall ratio is going to be 30% instead of 22%.

Assembly: interior partition, 3⅝" studs at 16" o.c., 10'-0" high, 5/8" Type X both faces — per LF

That is element C1010. And notice the cross-check the assembly gives you for free: 18,600 LF × 10 ft × 2 faces × a modest factor for double-layer and shaft walls comes out right around 412,000 SF of gypsum board — the canonical Northgate quantity, which is also what the finish contractor will use to price wall finishes. When two independent paths produce the same quantity, you have probably not made a gross error. When they don't, stop and find out why before you send the estimate.

You can build an assembly for almost anything repetitive: a single-occupant accessible staff restroom at roughly $18,400 each (fixtures, partitions, accessories, tile, plumbing rough-in and trim, exhaust, lighting, door, finishes), a nurse station, a patient toilet, an exam room, a typical hotel key. Once you have the assembly, conceptual estimating becomes counting.

11.3.5 Historical cost data and the three adjustments

Your own completed projects are the best cost data you will ever have. They are also wrong the moment you try to use them, in three specific ways. Adjust for all three, in this order, and show your work.

Adjustment 1 — Time (escalation). Prices moved between then and now. Compound forward.

Adjustment 2 — Location. Labor rates, productivity, code, market, and delivery costs differ by metro. Both RSMeans and ENR publish city cost indices for exactly this. Multiply by the ratio of your city's index to the source city's index. (Use real, current published indices. The index values in the example below are illustrative placeholders so you can follow the arithmetic — do not use them for anything real.)

Adjustment 3 — Size (economy of scale). Bigger buildings generally cost less per square foot, because fixed costs — the general conditions, the crane, the mobilization, the core — spread over more area, and because repetition improves productivity. A common rule of thumb is a 1.5% reduction in unit cost for every 10% increase in floor area, over a modest range. It is a rule of thumb. It breaks down badly at extremes and it does not apply at all if the program mix changes with the size.

Here is the whole thing, worked, on a real comparable.

Comparable: the Fairhaven Ambulatory Care Center — 104,000 SF, outpatient clinics plus imaging plus ambulatory surgery (the same program mix as Northgate), final construction cost $36,600,000, construction midpoint 34 months ago, in a metro with a published city cost index of 99.2 against Rivermont's 103.8. Assumed escalation over that period: 3.8% per year.

Against the elemental estimate of $52,300,591, that is 0.87% apart.

💡 Aha moment. Two independent methods — one bottom-up from thirty-five elemental quantities, one top-down from a single historical project — landed less than one percent apart. That does not mean the number is right. Both can be wrong in the same direction: same market assumptions, same estimator, both anchored on healthcare projects built the way Kestrel builds them. What convergence tells you is that you have not made a blunder. Divergence is information; convergence is only the absence of one kind of alarm. Tomás's version: "Agreeing with yourself is not evidence."

And notice the detail that trips up most people: escalate from the comparable's construction midpoint, not its completion date. A project's cost is dollars spent across its duration, so the "as of" date is roughly the middle of construction. Fairhaven completed 26 months ago and took 16 months, so its midpoint is 26 + 8 = 34 months ago. Using completion instead would have escalated only 26 months and understated the adjusted cost by about 2.4% — roughly $1.2 million on a Northgate-sized project. Nobody notices that error. It just quietly makes you low.

📋 Try it — adjust a historical project to Northgate.

You have a comparable and no time. Produce an adjusted budget estimate, state an honest accuracy range, and write down three assumptions.

Given: - Historical project: $38,200,000 final construction cost, 118,000 SF outpatient clinic, comparable program mix - Substantially complete 3 years ago; construction duration was 20 months - Located in a metro with a published city cost index of 94.8; Rivermont's is 103.8 - Escalation to apply: 3.8% per year - Size adjustment guideline: 1.5% reduction in unit cost per 10% increase in floor area - Target: 132,000 SF in Rivermont, today

Produce: (a) the adjusted $/SF, (b) the total, (c) an honest accuracy range with the estimate class named, and (d) three assumptions you would write on the cover sheet.

11.4 Escalation: Putting a Price on Time

Escalation is the increase in construction cost over time — materials, labor, equipment, and market conditions moving together. It belongs on its own line in every estimate, and it belongs there for two reasons.

First, it is not a contingency. Contingency covers risks that might happen. Escalation covers a thing that will almost certainly happen and whose only question is how much. Burying escalation inside the unit rates makes both numbers unauditable.

Second, whoever owns escalation risk is a contract question, and if it is invisible you cannot negotiate it. On Northgate's GMP, Kestrel owns escalation on everything inside the guaranteed number — which is why the escalation line survives all the way into the final GMP build-up at $575,200.

Forecast to the midpoint of construction, not to NTP

This is the mistake that quietly makes estimates low, and it is entirely avoidable.

You will not spend the money on the day of Notice to Proceed. You will spend it across the whole construction period. The weighted average date of the spend on a typical building — where the cost curve is an S-shape, slow at the start, heavy in the middle — is close enough to the midpoint of construction that the midpoint is the industry's standard convention.

Northgate, worked:

• Estimate is priced at current-day costs.
• Time to NTP: 16 months.
• Contract time: 565 calendar days ÷ 30.44 days/month = 18.56 months. Half of that: 9.28 months.
• Time from today to the midpoint of construction: 16 + 9.28 = 25.28 months = 2.107 years.
• Escalation assumption: 3.0% per year — Tomás's judgment, written down, with his reasoning attached.

Escalation factor = (1.030)^2.107 = 1.06425 → 6.43% Escalation amount = $43,566,690 × 0.0643 = $2,801,338

What it means for the job: we are carrying $2.8 million because we intend to buy this building an average of two years and one month from now, and we believe money buys about 6.4% less construction then than it does today.

💰 Money check — the cost of escalating to the wrong date. Suppose you escalate only to NTP: 16 months = 1.333 years.

(1.030)^1.333 = 1.04020 → 4.02% $43,566,690 × 0.0402 = $1,751,381

The difference: $2,801,338 − $1,751,381 = $1,049,957.

Escalating to NTP instead of to the midpoint of construction would have understated the estimate by roughly $1,050,000 — about 22% of the entire gap everyone is about to spend nine days arguing over. It would not have looked like a mistake. It would have looked like a better number. That is what makes it dangerous.

🔄 Check your understanding. A project is priced today. NTP is 10 months out and the contract time is 400 calendar days. Your escalation assumption is 4.0% per year and the escalatable base is $22,000,000. What escalation do you carry, and what would you have carried if you had escalated to NTP?

Be honest about what escalation forecasting is

It is a judgment call, and recent years have made that plain: escalation has run in the low single digits for long stretches and then, in some markets and material categories, in the high teens or worse over a single year, with steel, copper, aluminum, and electrical distribution equipment all doing dramatic and largely unforecast things. Industry forecasts have been badly wrong in both directions within recent memory. So do this:

1. Name your rate and your source — ENR's cost and material indices, published producer-price data, your own bid history, and what your subcontractor market is telling you.
2. State the effective date the estimate is priced to, on the cover sheet.
3. Escalate to the midpoint, always, and show the calculation.
4. Do not pretend to precision. "3.0% per year is our assumption; a 1.0-point change moves this estimate by roughly $920,000" is far more useful than a number carried to the dollar.
5. Escalate only un-bought scope. This is the one people miss, and it is why Northgate's escalation line falls from $2,801,338 at SD to $575,200 at GMP. At SD nothing is bought and everything escalates. At GMP, 84% of the work is under subcontract at fixed prices — those subcontractors now own their own escalation — and the allowance covers only what is left.

⚖️ What the contract says. Escalation risk allocation is a contract term, not a market fact. A GMP normally puts escalation on the construction manager inside the guaranteed number; a cost-plus contract without a GMP normally leaves it with the owner. Some contracts add a price-escalation clause or a materials-index adjustment on named commodities, sharing the risk above a threshold; some public owners forbid them outright. In a volatile market a well-drafted escalation clause on two or three named commodities is often cheaper for the owner than the contingency the contractor would otherwise carry — the contractor must price the worst case, while the owner pays only the actual case. That is Chapter 4's territory, and a conversation worth having with your attorney before you sign, not after.

11.5 The Owner's Whole Budget — and Why You Need to Know It

Construction cost is not project cost. On Northgate, the construction contract is $47,500,000 and the owner's total project budget is $61,000,000. If you only understand the first number, you will give Meridian bad advice with total confidence.

Construction is 77.9% of it. The other 22.1% is where three of the most common CM mistakes live.

Mistake 1 — advising on the wrong number. A VE item that saves $180,000 of construction but adds $60,000 to commissioning and delays occupancy three weeks has not saved $180,000. Meridian's interim clinic lease expires October 1, Year 2; three weeks past that carries a rent number and an operations number that dwarf your $180,000.

Mistake 2 — forgetting who is buying the equipment. The imaging equipment — $3,640,000 — is owner-furnished. It is not in the GMP; it sits in the owner's budget, procured by Meridian's imaging vendor on Meridian's schedule. That single fact is the seed of the most expensive change order on the job. The vendor selected a different MRI unit after the GMP was set, requiring a deeper depressed slab, added structural framing, more RF shielding, and a larger electrical feed. Pri gave a verbal go-ahead on a Thursday; Kestrel's assistant superintendent let the concrete crew build it Monday, with no written directive, no agreed price, no time-impact analysis, and no time-and-material tickets for four days. Owner's understanding of the cost: "about $60,000." Kestrel's actual cost: $186,400. Substantiated with contemporaneous records: $121,000. Settled eight weeks later at $142,750, leaving Kestrel $43,650 short, with 4 of 9 claimed calendar days granted. That is CO #14, taken apart properly in Chapter 31. Its origin is a line in the owner's budget that a lot of construction managers never look at.

Mistake 3 — not knowing where the owner's flexibility is. Sometimes the cheapest way to close a construction gap is not in the construction line at all: an owner's contingency with unspent room in it, an FF&E package that can be phased. You do not get to spend the owner's money — but you do get to show them the whole board. The most valuable thing a CM ever did for Pri Sethi was show her a trade-off nobody else had.

This is a direct callback to Chapter 2: the owner's budget exists because a pro forma exists — a financial model in which this building generates enough patient volume, revenue, or mission value to justify the capital. The budget is not a preference. It is a solved equation. When you ask an owner to spend $400,000 more, you are asking them to re-solve it.

🔄 Check your understanding. Meridian's construction budget is $47,500,000 and the total project budget is $61,000,000. You find a VE option that reduces construction cost by $240,000 but requires Meridian to buy a piece of equipment themselves that was previously in the contractor's scope, at an owner cost of $205,000. Is this a saving? What do you tell Pri?

11.6 Constructability Review — Reading the Drawings Like Somebody Has to Build Them

A constructability review is a systematic examination of the design documents by people who build, asking one question in seven different ways: can this actually be built, in this sequence, with this access, to these tolerances, with materials that exist, and maintained afterward?

It is not a design review. You are not second-guessing the engineer's calculations. You are the only person in the process whose job is to imagine the building being assembled by human beings in a real order on a real site.

The seven lenses:

1. Buildability — can it be built with normal means and methods, or does it need something exotic?
2. Sequence — does the order of assembly the drawings imply actually work? Can each trade get in, do its work, and get out?
3. Access — can the material, the equipment, and the people physically reach the work? Will the largest piece fit through the opening that exists at the time it must be installed?
4. Tolerance — do the tolerances of adjacent systems add up? Steel is fabricated to one tolerance, curtain wall to another, and somewhere there has to be a joint that absorbs the difference.
5. Availability — do the specified products exist, in this market, at a lead time the schedule can absorb, from more than one source?
6. Maintainability — can somebody service this in year seven without a lift, a demolition permit, or a prayer?
7. Safety — can this be built and maintained without exposing people to hazards the design could have removed?

Eight findings from the Northgate SD/DD constructability review

Margo Deacon, Grace Lindqvist, Tomás, and I spent about 220 hours across two review cycles. Here is the log, with the cost of not catching each one.

Every one of those is a redline at DD and a change order in the field. CR-02 is the classic: it is nobody's mistake in particular, which is exactly why nobody catches it. The architect drew the embeds because the curtain wall needs them. The structural engineer didn't draw them because the curtain-wall system wasn't selected yet. Both were behaving reasonably. The steel still ships without embeds.

⚠️ Safety alert — design out the hazard while it is still free. CR-01 is a safety finding wearing a logistics costume, and it is the one that would have hurt somebody: a crawler crane swinging loads over a route ambulances are required to use in a hurry is a suspended-load exposure, not a scheduling inconvenience. NIOSH's Prevention through Design initiative exists on exactly this premise — the cheapest and most effective place to eliminate a construction or maintenance hazard is in the design, not in the field with a rule and a sign. Put these four questions in every constructability review:

• Can this be erected without an unprotected leading edge? If the deck or roof geometry forces exposure, ask for a detail change, not a harness policy.
• Are permanent fall-protection anchors designed in for future roof, facade, and equipment maintenance — or is somebody in year six going to tie off to a vent pipe?
• Can rooftop equipment be serviced without crossing an unguarded edge? Add the guardrail now; the retrofit costs many times more and the alternative is unpriceable.
• Does the sequence require work over occupied space? On Northgate: yes, north elevation, continuously. That belongs in the logistics plan and the site-specific safety plan long before it belongs in a toolbox talk.

A hazard removed in a design meeting costs a drawing revision. The same hazard removed in the field costs a stop-work, a resequence, and — sometimes — a person.

🔄 Check your understanding. Constructability finding CR-08 — the 3,000 A switchboard section will not pass through the only door on its path once the CMU walls are laid — costs $21,000 if missed. But CR-08 is really about time, not doors. What is the general question CR-08 is an instance of, and name two other pieces of equipment on a typical building where you would ask it?

🧩 Productive struggle. Before you read the next section, spend three minutes on this. Meridian's finance office has sent Pri a suggestion: reduce the design contingency from 6% to 3% and the construction contingency from 3% to 1.5%. On a $43,566,690 cost of construction, that is a stroke-of-the-pen "saving" of $2,038,921 with the fee — 42% of the entire gap, achieved before lunch, with no change to the building.

Write down: (1) what has actually changed about the project, (2) who now owns the risk that the contingency was covering, and (3) what you would say to Pri, in two sentences, without being condescending.

11.7 Value Engineering, Properly — Closing the $4.8 Million Gap

11.7.1 What value engineering actually is

Value engineering (VE) — sometimes value analysis or value management — is the disciplined pursuit of value, where

Value = Function ÷ Cost

You can raise value two ways: deliver the same function for less cost, or deliver more function for the same cost. What you cannot do and still call it value engineering is deliver less function for less cost. That is a scope reduction. It may be the right decision — owners choose it all the time and legitimately — but it must be called by its name, decided by the owner with open eyes, and written down as what it is.

The core move is function analysis: for each element, ask what does this actually do? — in a verb and a noun, deliberately abstract — and then ask whether there is a cheaper way to do that.

That is the whole trick. The moment you write "unitized curtain wall" in the function column instead of "exclude weather, admit daylight," you have foreclosed every alternative and you are no longer doing value engineering. You are shopping.

11.7.2 The Northgate value engineering log

Nine days. Two half-day workshops with Tomás, Dale Whitcomb, Ruth Caldwell, Sofia Marchetti of Cardinal Mechanical, Devlin Achebe of Halcyon Electric, Grace Lindqvist on the model, Margo on constructability, and Pri in the room the entire time — which is the single most important logistical fact about the whole exercise. An owner who is not in the room does not own the decisions, and a VE log the owner did not participate in is a document you will be arguing about for the rest of the job.

Every savings figure below is burdened — multiplied by the 1.2005 factor from §11.3.3, so it is the effect on the bottom line, not on the raw cost of work. Tomás says this at the start of every VE meeting: "Every number in this log is burdened. I've already done that math. Don't do it twice."

Accepted burdened savings: $3,027,200 — 63.1% of the $4,800,591 gap.

Revised schematic design estimate: $52,300,591 − $3,027,200 = $49,273,391, or $373.28/SF.

Still $1,773,391 over budget. Hold that thought; §11.9 finishes the story, and the way it finishes is the most important thing in this chapter.

Notice what the accepted items have in common. Not one of them makes the building worse at what it is for. VE-01 changes how lateral load is resisted; the building resists it just as well. VE-02 changes what the wall is made of; it excludes weather and admits daylight to the same performance numbers. VE-09 changes how a copper joint is made; the water still gets to the sink. VE-16 is the odd one out, and it is labeled as the odd one out: it is a real reduction in what the café can do, chosen deliberately by the owner who will operate it, recorded as a program decision.

11.7.3 False value engineering — the four traps, named

Trap 1 — Deferring scope that will come back as a change order. "Let's shell the fourth floor and fit it out later." Sometimes that is a correct phasing decision. Often it is arithmetic with the future deleted. Coming back into a finished, occupied healthcare building means a second mobilization, a second permit, infection-control risk assessment protocols, off-hours work, and MEP tie-ins into live systems — a premium that commonly runs 30–40% over what the same work would have cost in sequence. If deferring is genuinely right, it will still look right after you write down the return cost. Write it down.

Trap 2 — Cheapening what the owner lives with every day. That is VE-12. The specified sheet vinyl with heat-welded seams was chosen for infection control and a 20-year wear layer; the substitute has seams and a 10-year wear layer. It "saves" $228,100 and buys Meridian a full corridor re-floor in year eleven, in an occupied clinic, at escalated cost, with the clinic closed for it. Function did not stay constant — function got worse. The tell is always the same: the life-cycle column is empty or hand-waved.

Trap 3 — Removing contingency. That is VE-14, and it is the trap I want you unable to fall into after this chapter. It "saves" $2,038,900 — 42% of the gap — while changing nothing about the project.

Contingency is not money. It is an admission. It is the priced, owned reserve for risks you have identified and named — Chapter 6's definition, and the one that matters. Deleting the reserve does not delete the risk. It deletes the funding and transfers the exposure to whoever is standing closest when it lands. A construction manager who lets an owner "value-engineer" the contingency has done that owner harm, even though the owner asked for it, even though it made the meeting easier, and even though the number on the page got better. Say no in writing, explain what the contingency is funding, and if the owner insists, document the decision, the reasoning, and the exposure in a letter. Case study 2 is what happens when nobody does.

Trap 4 — Deleting the verification. That is VE-15: delete the full-size enclosure mockup and the water-penetration testing, save $86,400. Notice the trap inside the trap — Kestrel had just accepted VE-02, a facade change worth $508,400, and the mockup is precisely what proves the new system performs. Deleting the verification of a change you made to save money is how a $508,400 saving becomes a seven-figure water-infiltration remediation in Year 3. The verification is part of the option, not an add-on to it.

11.7.4 Two more rejections worth understanding

VE-06 — reduce floor-to-floor from 15'-0" to 14'-4", $384,200. On paper it is beautiful: less curtain wall, less precast, less steel, less stair, less everything. Sofia Marchetti killed it in four minutes with Grace's model on the screen. In the clinic wings the above-ceiling zone carries a 26-inch main supply duct with its insulation, a sprinkler main below it, medical gas, cable tray, structural depth, and lights — and exam-room ceiling height is not negotiable under Meridian's own standards. Take eight inches out and the duct main gets reshaped and rerouted around structure at every bay, which costs more than the eight inches saves and buys a coordination nightmare that surfaces as RFIs eleven months later.

💡 This is Chapter 10's threshold concept arriving early: MEP coordination, not structure, sets the interior. The reason we killed VE-06 in four minutes instead of discovering it in the field is that Grace had a coordinated model at schematic design. The value engineering was the model.

VE-13 — delete spare conduit and the spare 400 A feeder provision, $117,600. I argued against this one, and the reasoning is different from the others. There is nothing wrong with the item; the spare pathway is genuinely unnecessary for the building as designed. What made it a bad trade was CR-03 — the imaging vendor had not been selected. When you know a decision is still open, the cheapest thing you will ever own is the option to accommodate it. $117,600 to hold an option on a $3,640,000 equipment package nobody has chosen yet is not a cost. It is insurance at a price you should be delighted to pay.

And here is the honesty that makes this chapter worth your time: on the same day, for the same reason, we accepted VE-04 — the imaging structural simplification, $267,700 — carrying exactly the same open risk. Eleven months later that decision cost roughly $71,300 in demolition and rebuild inside CO #14. VE-04 still netted about $151,700 and was still, on the information available, a defensible bet. But a good decision and a good outcome are not the same thing, and neither are a good decision and a consistent one. Case study 1 takes that room apart.

11.7.5 A decision test you can actually use

                 A "SAVING" IS PROPOSED
                          │
              Does the FUNCTION change?
                    ┌─────┴─────┐
                   NO           YES
                    │             │
     Does LIFE-CYCLE cost         │
     stay level or improve?       │
        ┌───────┴───────┐         │
       YES              NO        │
        │                │        │
   Does it add    ─── it's a  ── Did the OWNER choose it,
   SCHEDULE or        FALSE      knowing what they give up,
   QUALITY risk?       VE        with it written down?
    ┌────┴────┐                    ┌────┴────┐
   NO        YES                  YES        NO
    │          │                   │          │
 GENUINE   GENUINE VE          LEGITIMATE   SCOPE CUT
   VE      with a named        PROGRAM      IN DISGUISE
           risk — put it       DECISION     ── say so ──
           IN THE LOG          (label it)

📊 Diagram (described). Four questions, and the order matters. Function first, because if function changes it is not VE no matter how good the number is. Life-cycle second, because a first-cost saving that raises operating cost is a transfer, not a saving. Schedule and quality risk third, because those are real costs the estimate does not show. The owner's informed choice last, because a scope reduction the owner makes deliberately is legitimate and a scope reduction the owner discovers later is a betrayal.

⚖️ What the contract says. Under a CM at Risk agreement, value engineering before the GMP is set is simply designing to a budget: the number moves and nobody owes anybody anything. After the GMP is set, everything changes — post-GMP VE is a change to the contract documents, requiring a change order to reduce the GMP, and the standard forms (the AIA A133 family for CM at Risk, and comparable ConsensusDocs and EJCDC forms) address how savings are handled and whether the contractor shares them. On Northgate, unused GMP contingency at final completion splits 75% owner / 25% Kestrel. Know before the meeting starts whether a "saving" reduces the GMP dollar for dollar, returns to the owner at closeout, or gets shared — that determines who is actually motivated to find it.

🔄 Check your understanding. A subcontractor proposes substituting a different manufacturer's rooftop unit that meets the specified capacity and efficiency, saves $46,000, and has a lead time 8 weeks shorter. The specification names three acceptable manufacturers and this is not one of them. Is this value engineering? What do you do?

11.8 Life-Cycle Cost, First Cost, and the Incentive Problem

Every VE decision has two prices: what it costs to build and what it costs to own. A construction manager is paid to care about the first and is trusted to care about the second.

A worked life-cycle comparison

Northgate's central plant, at design development. Two chiller options:

Step 1 — first-cost delta. $468,000 − $412,000 = $56,000 raw. Burdened at 1.2005: $67,226.

Step 2 — annual energy. Design cooling load 380 tons; 1,850 equivalent full-load hours per year (a typical assumption for this climate and occupancy — get the real number from the mechanical engineer's model).

380 tons × 1,850 hours = 703,000 ton-hours Base: 703,000 × 0.62 = 435,860 kWh/yr High-efficiency: 703,000 × 0.50 = 351,500 kWh/yr Saving: 84,360 kWh/yr

Step 3 — annual dollars. At a blended commercial rate of $0.132/kWh:

84,360 kWh × $0.132 = $11,136 per year

Step 4 — simple payback.

$67,226 ÷ $11,136 = 6.0 years

Step 5 — with the utility incentive. The local utility's commercial efficiency program offered $14,000 for this equipment class.

Net first cost: $67,226 − $14,000 = $53,226 Payback: $53,226 ÷ $11,136 = 4.8 years Energy saved over a 20-year chiller service life, at flat rates: $222,710

What it means for the job: Meridian spends $53,226 net and gets it back in under five years, then keeps roughly $11,000 a year for fifteen more. Add maintenance, refrigerant, and the fact that energy prices historically rise, and this is not close.

Simple payback ignores the time value of money, escalating energy prices, maintenance differences, and equipment life. For decisions above roughly $250,000, ask for a proper life-cycle cost analysis with a discount rate and a defined study period. But simple payback is the right tool in a VE meeting, because every person in the room can check it in their head.

Now the uncomfortable part

Meridian will own this building for forty years. Kestrel will own it for 565 calendar days.

Kestrel's fee is 4% of cost, so the high-efficiency chiller adds about $2,586 to Kestrel's fee. On this decision, a Kestrel optimizing for fee should push the expensive chiller; a Kestrel desperate to close a gap should push the cheap one. Neither incentive has anything to do with what is right for Meridian.

So name it out loud: "For what it's worth, our fee is a percentage, so we make about $2,600 more if you buy the better chiller. That's not why I'm recommending it. I'm recommending it because it pays back in under five years and you're going to own it for forty." An incentive you have named is manageable. An incentive operating silently erodes trust over eighteen months in ways nobody can point to.

This is also where sustainability and cost stop being different conversations — energy codes, efficiency targets, and the life-cycle case for better systems are Chapter 36.

Target Value Design — the alternative to estimate-then-cut

Everything in this chapter so far describes a loop: design it, price it, discover it costs too much, cut it. That loop is normal, it is what most projects do, and it is wasteful — every cycle spends design fees producing something that gets thrown away.

Target Value Design (TVD) inverts it: instead of estimating the design, you design to the estimate. Cost is an input to design, not an output.

1. Establish the allowable cost — what the owner's business case can fund. On Northgate that is $47,500,000, straight from Meridian's pro forma.
2. Validate that the program can be delivered for that cost before design begins, using benchmarks and conceptual estimates.
3. Break the target into cluster targets by system — structure, enclosure, mechanical — and give each cross-functional cluster team its own number.
4. Design within the target continuously, estimating in near-real-time, with the estimator in the design meeting rather than receiving its output six weeks later.
5. Treat an over-target condition as a design problem to solve now, in that cluster, not as a VE exercise in eight months.

When it works, TVD lands projects on budget without a demolition-derby VE phase, and it is one of the practices associated with the Lean Construction Institute's body of work.

Be honest about what it requires: an owner willing to state a real, fixed allowable cost early and not move it; a design team paid in a way that does not punish collaboration, often a multi-party or integrated agreement rather than a conventional fee; a contractor on board during programming, not schematic design; continuous estimating capacity, which costs money; and everybody accepting that some design decisions get made earlier and more irreversibly than architects are used to. Meridian was not set up for it, and many owners aren't. That is not a moral failing; it is a description of how most projects are procured. Full treatment in Chapter 27.

11.9 Converting to a GMP — Where the Advisor Becomes the Risk-Taker

Up to this moment, everything Kestrel has done is advice. Estimates are opinions with arithmetic attached. The moment you set a Guaranteed Maximum Price, that stops. You are now guaranteeing that the cost of the work plus your general conditions plus your fee will not exceed a number, and every dollar above it is yours.

What must be true before you set a GMP

Chapter 4's threshold concept, restated where it hurts

Chapter 4 told you that a "guaranteed" maximum price guarantees the contractor's exposure, not the owner's cost. The operational meaning: the GMP is guaranteed against the scope defined in the GMP documents, and every change to that scope moves the number. So the definition of the scope is the entire ballgame — and that definition is not the drawings. It is the drawings plus the qualifications page.

The single most valuable page in any GMP proposal is the qualifications and assumptions page. It is also the page most people write in twenty minutes at 11 p.m. the night before it is due. Here is what it should look like.

Northgate GMP — Qualifications, Clarifications, Assumptions, and Allowances (extract)

This GMP of $47,500,000 is based on the following. Where any of these is not correct, the GMP is subject to adjustment by change order.

1. Documents. Based on H+P drawings and specifications dated [date], 100% construction documents, including Addenda 1 through 4. Later revisions are not included.
2. Contract time. 565 calendar days from NTP, with NTP no later than March 3, Year 1. A later NTP entitles Kestrel to review escalation and winter-condition impacts.
3. Work hours. Priced on a single shift, Monday through Friday, 6:30 a.m. to 4:30 p.m. Premium time is not included except as specifically listed. Any work-hour restriction imposed by Meridian or the adjacent clinic beyond those in the contract documents is a change.
4. Adjacent clinic operations. Priced on continuous access to the north work zone during working hours, with two pre-scheduled 8-hour shutdown windows for utility tie-ins. Additional shutdowns or restricted windows are a change.
5. Subsurface. Based on the geotechnical report dated [date] and its recommendations. No allowance for rock excavation, contaminated soil, dewatering beyond normal surface control, or undocumented buried structures. Quantities and unit prices for each are attached at Exhibit C.
6. Owner-furnished equipment. Imaging and medical equipment is owner-furnished and owner-installed. The GMP is based on the equipment cut sheets and structural, electrical, and shielding requirements received [date] for the units listed at Exhibit D. A change in unit, manufacturer, or model is a change in scope.
7. Permits. Building permit fees are by Meridian. Kestrel includes the cost of obtaining trade permits and inspections. Impact fees, utility connection fees, and plan-review fees are excluded.
8. Design contingency is not carried. The GMP is based on complete documents. Construction contingency of $1,320,000 is carried for construction-phase risk as defined in the risk register at Exhibit E, and is drawn down against written entries only.
9. Escalation. An allowance of $575,200 is carried for the un-bought portion of the work, approximately 16% of the cost of work as of the GMP date. Buyout is assumed complete within 120 days of NTP.
10. Allowances (each a defined scope, not a lump for the unknown): signage and graphics $128,000; landscape planting material $86,000; specialty lighting at the lobby and café $74,000; owner-directed casework revisions $95,000. Allowances are reconciled to actual cost at closeout.
11. Unit prices for added or deleted footing excavation, undercut and structural fill, rock removal, and additional pile or pier work are attached at Exhibit C and are valid for the contract duration.
12. Owner decisions. The GMP assumes owner decisions on the items listed at Exhibit F by the dates listed. Decisions later than those dates may impact cost and time.
13. Weather. Priced with 14 days of anticipated adverse-weather delay per the contract's weather provision. Days beyond that are a time extension per the contract.
14. Exclusions. FF&E, IT and AV equipment and cabling beyond the pathways shown, medical equipment, moving, owner's testing and inspection, commissioning agent, builder's risk deductible above $25,000, hazardous-material abatement, and off-site utility work beyond the property line.

Fourteen items. About a page. Every one of them is either a fight avoided or a fight you win in ninety seconds instead of nine weeks. Item 6 is the one that saved Kestrel's relationship with Meridian during CO #14: the entitlement argument was over in one meeting because the qualification said, in writing, exactly which equipment the price was based on.

⚠️ The preconstruction failure mode

Here is the pattern. Learn to see it coming.

A GMP is set on 60% construction documents, because the owner needs a number for their board and the schedule cannot wait. The contingency is 2%, because 5% "looked like padding" and the competitive pressure was real. The assumptions page is one paragraph that says the price is "based on the documents and our understanding of the project."

What happens next is not a mystery and not bad luck. The documents complete. Between 60% and 100% CD the design adds detail — that is what design does — and every added detail is either inside the 60% scope (contractor absorbs it) or outside it (change order). With no qualifications page, nobody can tell which, so every item becomes a negotiation: the contractor argues scope growth, the owner argues it was always implied. The 2% contingency is gone by the fourth change order, at which point the contractor starts pricing defensively, stops volunteering solutions, and begins documenting for a claim, while the owner starts reading every submittal as an attempt to cheapen the job. Within six months there are two projects: the one being built and the one being papered.

The full anatomy of what follows is Chapter 31, and case study 2 of this chapter — a job I ran and got wrong.

Now finish the Northgate story

Nine days of value engineering closed $3,027,200 of a $4,800,591 gap and left Kestrel $1,773,391 over Meridian's budget. Eleven months later, Kestrel signed a GMP of exactly $47,500,000. Here is where the rest of it came from, and this is the table I would put on a wall.

And here is the cost of the work by itself, reconciled:

Read those two tables together and then read this sentence slowly, because it is the most useful thing in the chapter and almost nobody says it out loud:

We closed a $4.8 million gap, and the building got more expensive.

Value engineering removed $2,521,700 of real work. Design development then added $2,755,010 back — the drawings gained weight between schematic design and permit, as drawings always do. Net, the cost of the work went up by $233,310. What actually closed the gap was retiring uncertainty: the design contingency went to zero because the design was finished, and the escalation allowance fell from $2,801,338 to $575,200 because 84% of the work was under subcontract at fixed prices before the GMP was signed. Roughly $4.84 million of the "savings" was never in the building at all. It was in the not-knowing.

💡 Aha moment. This is Theme 1 in its purest form: construction management is the management of risk, and most of what looks like cost is actually priced uncertainty. Notice too that the design contingency was very nearly exactly right — Tomás carried $2,614,001 against actual design development growth of $2,755,010, a 95% call on a number he set at 12% design completeness. That is not luck. That is what a contingency is for and what it looks like when it is sized from the bottom up against named risks.

So why bother with value engineering at all, if uncertainty was going to do the heavy lifting?

Because of this. Without those thirteen accepted items, the GMP cost of work would have been $40,000,000 + $2,521,700 = $42,521,700, and the GMP would have come out at:

($42,521,700 + $2,900,000 + $900,000 + $1,320,000) = $47,641,700 + fee at 4.0% = $1,905,668 + escalation allowance $575,200 = $50,122,568

$2.6 million over budget at GMP, with no design contingency left to retire and no time left to find anything. Meridian's options at that point would have been to shell the fourth floor, delay the project past their interim clinic lease expiration on October 1, Year 2, or go back to their board. All three are worse than nine days in a conference room.

And the other reason: you cannot promise an owner in March that uncertainty will retire in your favor. Contingency comes back only if the risks you named do not happen. Escalation comes down only if you buy the work out early and the market cooperates. Those are forecasts, and forecasts are the CM's job to make and the CM's exposure to be wrong about. The $3,027,200 in the VE log is the part you can put your name on the day you hand it over.

🔄 Check your understanding. Look at the SD-versus-GMP table again. The construction contingency went up — from $1,307,001 to $1,320,000 — while the design contingency went to zero and escalation fell by $2.2 million. Why would a contractor increase the construction contingency at the moment the design is complete and most of the work is bought?

The two ethics problems in this chapter, named

The CM who inflates the early budget. Set the schematic budget high, "find savings" through value engineering, and look like a hero when the GMP comes in under. It works, and everyone applauds. It is also a lie that costs the owner real money, because owners make irreversible decisions on early budgets: they size bond issues, set fundraising targets, cut program, defer other projects, and sometimes cancel. A high early number is not conservative; it is a thumb on a scale other people are standing on. If your estimate is high because of identified risk, name the risk and price it as contingency where everyone can see it. That is the honest version of the same caution, and it survives daylight.

The designer who under-designs to hit a number. The mirror image, equally real. An architect or engineer under budget pressure who thins a structural system, undersizes a mechanical system, or leaves a detail out of a set is not saving money — they are moving it into the RFI log and the change order log, where it costs more and somebody else gets blamed. If you see it, say so early and privately. The right sentence is not "this is under-designed"; it is "we've priced this at $X and I want to be sure that's what you intend, because if it grows at 90% CD it grows on the owner's dime."

Both failures share a root: a number made comfortable instead of true. In this business the comfortable number always comes due.

Spaced Review

Do not look back yet. Answer from memory, then check.

1. From Chapter 6 — contingency. What is the difference between sizing a contingency bottom-up and top-down, and why does this chapter care?

2. From Chapter 3 — CM at Risk. At what point in the project is the construction manager engaged under CMAR, and what would have been different on Northgate under design-bid-build?

3. Deep callback to Chapter 2 — why the budget exists at all. Where did Meridian's $47,500,000 come from, and what does that tell you about how much it can move?

Project Checkpoint: Conceptual Estimate and Value Engineering Log — Willow Street Community Center

In Chapter 10 you wrote the MEP systems narrative for Willow Street and predicted six coordination conflicts. Keep that list next to you — at least two of your VE options should touch it.

The situation. The City of Rivermont Parks & Recreation has $6,800,000 for the Willow Street Community Center: 24,000 SF, two stories, wood-framed second floor over a structural steel and CMU first floor, with a gymnasium, two multipurpose rooms, a commercial kitchen, offices, and locker rooms, on a flat 2.1-acre site with one existing 8-inch water main to relocate. Full program, quantities, and drawing descriptions are in Appendix K.

Willow Street is design-bid-build, so strictly speaking nobody has hired a contractor yet. That is the point: many public agencies retain a CM advisor during design on DBB projects for exactly this service, and even when they don't, a contractor deciding whether to chase the job does this analysis privately. Write it as the City's CM advisor.

Document 1: a UniFormat conceptual estimate. Build it exactly like the Northgate table in §11.3.3 — elements A through G, each with a quantity, a unit, a rate, and a subtotal. Then markups: general conditions, insurance and bonds, design contingency, construction contingency, escalation to the midpoint of construction, and, since this is a lump-sum bid rather than a fee-based GMP, contractor's overhead and profit. Show your burden multiplier.

Two mandatory cross-checks. Divide your total by 24,000 SF and ask whether the $/SF is believable for a community center (§11.3.1 gives $300–$480). Then estimate the same building a second way — $/SF from a comparable, adjusted for time, location, and size per §11.3.5 — and reconcile. If the two disagree by more than about 10%, find out why before you write the cover sheet.

Your estimate comes in at $7,410,000 — $308.75/SF — which is $610,000 over the City's $6,800,000.

Document 2: a value engineering log with eight priced options. Same columns as the Northgate log: ID, description, first-cost saving (burdened — show your multiplier), life-cycle impact, schedule impact, quality/aesthetic impact, risk, recommendation. Worth looking at on this building: the gymnasium roof structure (steel joists versus glulam versus panelized); the commercial kitchen's equipment and ventilation scope; the second-floor framing approach; paving and parking extent; CMU versus metal panel on the exterior; the locker-room finish and fixture package; skylights and daylighting; and mechanical zoning across a building whose gym, kitchen, and offices have violently different loads.

The requirement that makes this worth doing: mark every option clearly as REAL VE (function held or improved, life-cycle neutral or better) or SCOPE CUT (function reduced — legitimate only if the City chooses it knowingly). Be honest. Most people's first eight options are five scope cuts and three real ones, and learning that about your own instincts is the lesson. Then state which you recommend and — the part everyone skips — whether the eight actually close $610,000, and what you tell the City if they don't.

Next: Chapter 12 takes you from parametric rates to real quantities. Your Willow Street deliverable there is a quantity takeoff of sitework, concrete, masonry, and structure with waste factors — the numbers that replace the "1 LS" lines in the estimate you just built.

Chapter Summary

The one-page version.

The decision test, compressed: Does the function change? If yes, it is a scope cut — say so. If no, does life-cycle cost stay level or improve? If no, it is a transfer, not a saving. If yes, what schedule and quality risk does it carry, and is that risk written in the log next to the dollar figure? Then, and only then, is it value engineering.

The four numbers from this chapter worth memorizing: the estimate class ladder and its bands; escalate to the midpoint of construction; the burden multiplier; and the fact that on Northgate the cost of the work went up $233,310 while the price came down $4,800,591.

What's Next

You now know how to build a number when you do not have drawings. Chapter 12 is what happens when you do: quantity takeoff, unit costs, productivity, waste factors, and the equation that underlies every line in every estimate you will ever build — unit cost = quantity × productivity × rate. It also carries the threshold concept that reframes everything in Part III: an estimate is not a prediction, it is a priced bundle of assumptions and risks, and every one of them belongs to somebody. After that, Chapter 13 assembles the whole bid, and Chapter 14 finally puts time next to the money where it belongs.