Chapter 3 — Project Delivery Methods: Design-Bid-Build, Design-Build, CM at Risk, and IPD

The Hook: Three Consultants, Three Answers

Eighteen months before Kestrel Construction Group got notice to proceed on the Northgate Outpatient Pavilion — before there was a Kestrel on the job, before there was a guaranteed maximum price, before anybody had moved a shovel of dirt on that sloping 6.2-acre site — there was a Tuesday afternoon board meeting on the fourth floor of Meridian Health System's administrative building.

Priyanka Sethi had been Meridian's owner's representative for six years. She had a binder in front of her and a three-page memo she had rewritten four times. On the screen behind her was one slide with three words on it: How We Build.

Meridian's board had already approved the money. Sixty-one million dollars, total project budget: construction, design fees, furniture and equipment, the medical imaging package, permits, financing, and owner's contingency. They had approved the program — four stories, roughly 132,000 gross square feet, outpatient clinics, an imaging suite, ambulatory surgery, and a lobby that had to feel like the front door of the health system rather than the back of a hospital. They had approved the site: 6.2 acres on the urban edge, with an existing Meridian clinic on the north property line that would stay open and stay busy through every single day of construction.

What the board had not decided was how to buy it.

Pri had hired three consultants over the previous nine months, one at a time, and each had produced a thoughtful recommendation. The problem was that there were three of them.

The first, a program-management firm that did a lot of municipal work, recommended design-bid-build: finish the drawings, advertise the job, take sealed bids, award to the low responsible bidder. "You are a non-profit spending community money," the partner wrote. "Competitive bidding on complete documents is the most defensible thing you can do. If somebody asks whether you got a fair price, you hand them the bid tab."

The second, a national healthcare development firm, recommended design-build — one contract, one entity responsible for both the drawings and the building, construction starting while design was still running. "Ten months earlier occupancy," their principal said, and said it twice. "That is ten months of clinic revenue you are planning to leave on the table."

The third, a small owner's-representative consultancy Pri trusted, recommended construction manager at risk: bring a builder on during design as a paid advisor, use them to price and constructability-review the drawings as they develop, then convert them into the contractor with a guaranteed maximum price once the design can be priced honestly.

Harriet Voss, who chaired the board's facilities and finance committee and had spent thirty years in commercial real estate before she retired, listened to all of it and asked the only question that mattered.

"Priyanka. Which of these gets us the lowest number?"

And Pri, to her enormous credit, did not answer the question as asked. She said: "That is not what we're deciding. We're deciding who owns what we don't know yet. The price follows from that."

I have watched a lot of owners make this decision. Most of them ask Harriet's question. The good ones ask Pri's.

Because here is what nobody tells you in school: the delivery method is not paperwork you sort out before the real work starts. It is the largest cost decision on the project that does not involve changing the building. Same square footage, same 985 tons of steel, same 38,500 square feet of curtain wall — and the price moves by seven figures depending on how you buy it, because the risk moves with it. Somebody has to carry the unknown, and whoever carries it charges for it.

🏃 Fast Track: If you know the methods cold, go straight to §3.7 (the master comparison table), §3.8 (the threshold concept, with the Northgate arithmetic showing the same building at two prices), and §3.9 (the decision framework).

🔬 Deep Dive: The contract types inside these delivery methods — lump sum, cost-plus, GMP, unit price — get full treatment in Chapter 4, with the clause-by-clause consequences in Appendix G. What a CM actually does during preconstruction is Chapter 11.

3.1 Three Decisions That Look Like One

Clear up a confusion first, because it trips up people who have been doing this for a decade.

When an owner says "we're doing CM at risk," they are compressing three separate decisions into one phrase:

1. The delivery method — the structure of relationships. Who has a contract with whom? How many contracts does the owner sign? Who holds the design agreement? When does the builder show up?
2. The contract type — the pricing mechanism. Lump sum? Cost of the work plus a fee? A guaranteed maximum price (GMP)? Unit prices? That is Chapter 4.
3. The selection method — how you pick the firm. Low bid on complete documents? Qualifications-based selection with no price at all? Best value, where price and qualifications are scored together?

They are independent in theory and tightly coupled in practice, because certain combinations have hardened into conventions:

Keep them straight. When somebody tells you "we want a guaranteed maximum price," they have told you a contract type, not a delivery method — and you still don't know whether they intend to bid it or negotiate it, or whether the designer works for them or for you. Ask.

Why do they couple so tightly? Because all three answer the same underlying question from different angles: at the moment somebody commits to a price, how much do they know? Low-bid selection only works on complete documents. Qualifications-based selection is the only honest option when documents are incomplete, because there is no meaningful price to compare. And the contract type has to match how much is known — you can fix a lump sum against a finished drawing set, but against a 60% set you can only fix a cap with named allowances and a contingency.

3.2 Design-Bid-Build: The Sequential Default

Design-bid-build (DBB) is the traditional method — still the default for most public agencies in the United States, in many cases by statute. It is genuinely sequential: design finishes, then bidding happens, then building starts.

                        ┌──────────────┐
                        │    OWNER     │
                        └──┬────────┬──┘
              contract     │        │     contract
          ┌────────────────┘        └───────────────┐
          ▼                                         ▼
  ┌───────────────┐    NO CONTRACT           ┌──────────────┐
  │  ARCHITECT /  │◄ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─►│   GENERAL    │
  │   ENGINEER    │  (communication only,    │  CONTRACTOR  │
  └───────┬───────┘   through documents and  └──────┬───────┘
          │            the RFI process)             │
    design consultants                        subcontractors
    (structural, MEP,                          (20–40 firms)
     civil, landscape)                               │
                                                 suppliers

📊 Diagram (described). Read that as a triangle with one side missing. The owner has a contract with the architect/engineer (A/E) and a separate contract with the general contractor (GC). The A/E and the GC have no contract with each other at all. They communicate through documents — the drawings and specifications the A/E produced, and the requests for information (RFIs) and submittals the GC sends back. Every disagreement between them lands on the owner's desk.

That missing side is not a diagram quirk. It is the defining feature of the method and the source of everything good and bad about it.

3.2.1 How it runs

The A/E takes the design from schematic through design development to a complete, permit-ready set of construction documents — the "100% set." The owner advertises or invites a bid list. Contractors get four to six weeks to take off quantities, solicit subcontractor and supplier quotes, and assemble a number. On a set day at a set hour, sealed bids are opened; on public work they are usually read aloud. The low responsive, responsible bidder wins — responsive meaning the bid followed the rules (bid bond attached, alternates priced, signed, on time), responsible meaning the bidder is capable (licensed, bondable, experienced, not debarred). Bidding as a discipline is Chapter 15.

Then the contractor signs a lump sum contract: one price for everything shown. From that moment it owns the means, the methods, the sequencing, the productivity, the weather (usually), the subcontractor market, and its own estimating errors. If steel bids came in $400,000 over the estimate, that is the contractor's problem. If the crews outperform, that is the contractor's profit.

3.2.2 The doctrine that makes it fair

⚖️ What the contract says — the owner's implied warranty of the plans.

When an owner furnishes plans and specifications and directs a contractor to build to them, the owner is generally held to have impliedly warranted that those documents are adequate for their purpose. The contractor is entitled to rely on them. If the contractor builds exactly what is shown and the result does not work — the beam is undersized, the duct does not fit the chase, two drawings contradict each other — the contractor is generally not liable for that failure and is generally entitled to be paid for the fix.

In the United States this is usually called the Spearin doctrine, after a U.S. Supreme Court decision from the World War I era involving a relocated sewer that a contractor built to government drawings and that then failed. It has been adopted, with variations, across most American jurisdictions and has close analogues in other common-law systems.

Three practitioner points:

• It is a default, not a law of nature. Owners routinely try to modify or disclaim it — document-verification clauses, site-investigation clauses, duties to report discrepancies, "no damages for delay." How far those disclaimers get enforced varies significantly by jurisdiction and by how sweeping they are. Read your contract; when it matters, ask your attorney.
• It does not cover what you should have caught. Most contracts impose a duty to review the documents with reasonable care and report obvious errors before proceeding. A conflict you could have seen and did not report is much harder to recover on.
• It is why the owner, not the contractor, owns design errors in DBB. That is the most consequential risk allocation in the method, and it drives what comes next.

3.2.3 The characteristic failure mode: design errors become change orders

In DBB, the builder sees the drawings for the first time when they are finished. Nobody who knows how to build the building was in the room while it was being drawn. The result is predictable:

A conflict surfaces in the field — the mechanical main and the roof joist want the same eleven inches. The contractor writes an RFI. The A/E, now working on a fixed design fee that ran out three months ago with four other projects live, takes two, three, five weeks. The answer requires a deeper joist. The contractor prices a change order. The architect says coordination is a contractor responsibility. The contractor says the documents were defective and cites the implied warranty. Both are partly right. Nobody is building anything in that area while this happens.

Multiply by forty occurrences on a mid-size building and you have the DBB experience: a bid-day price that looked excellent, a final cost that did not, and a schedule that slipped in three-week increments nobody logged as a delay until they added up.

🏗️ From the field. I ran a DBB municipal library early in my career. Our bid was $310,000 under the second bidder and I was proud of that for about eleven weeks. Then we opened the ceiling coordination and found the structural and mechanical engineers had never reconciled the clerestory framing with the return-air path. Forty-one RFIs in that one area. We eventually recovered most of the money — but "eventually" was fourteen months after substantial completion, and the general conditions we burned waiting for answers were gone forever. The bid was low. The job was not cheap.

3.2.4 What DBB is genuinely good at

Do not read the above as a verdict. DBB is the right answer more often than the CM-at-risk sales pitch admits.

Price discovery is real. Competitive bidding on a complete set, with a deep bid list, tests the market honestly, and there is no substitute for it. It is defensible — for a public agency, the bid tab answers every "did you get a fair price" question, and that value does not show up in a cost model. It is simple to administer, requiring far less owner staff than CM-as-agent or IPD. And when the documents are genuinely complete and the building is genuinely conventional, it is often the lowest total cost. A warehouse, a fire station, a road project, a repeat prototype — DBB is fine and frequently best.

🔄 Check your understanding. In a DBB project, the architect and the contractor disagree about what a detail requires. Who has a contract with whom — and why does that structure force the owner to resolve it?

3.3 Design-Build: One Contract, One Throat

Design-build (DB) collapses the two contracts into one. The owner signs a single agreement with a design-builder responsible for both the design and the construction. The architect and engineers are employees of the design-builder or subconsultants to it.

              ┌──────────────┐              ┌────────────────────┐
              │    OWNER     │─ ─ ─ ─ ─ ─ ─▶│ OWNER'S CRITERIA   │
              └───────┬──────┘   (optional) │ CONSULTANT /       │
                      │ ONE contract        │ BRIDGING ARCHITECT │
                      ▼                     └────────────────────┘
            ┌──────────────────┐
            │  DESIGN-BUILDER  │
            └───┬──────────┬───┘
       ┌────────┘          └─────────┐
       ▼                             ▼
┌──────────────┐          ┌──────────────────┐
│  ARCHITECT / │          │ SUBCONTRACTORS   │
│  ENGINEERS   │          │  AND SUPPLIERS   │
│ (in-house or │          └──────────────────┘
│  subconsult.)│
└──────────────┘

3.3.1 What "single point of responsibility" actually buys

The phrase you will hear is "single point of responsibility," or less politely, "one throat to choke." It means this: when the mechanical main and the roof joist want the same eleven inches, there is nobody for the design-builder to send an RFI to. The conflict is internal. It gets solved by two people in the same organization walking down the hall, and the cost lands inside a price the owner already agreed to.

That is a structural elimination of an entire category of change order and an entire category of dispute. It is the strongest argument for the method and it is not marketing.

What it does not eliminate is the owner's own changes. If the owner decides in month nine that it wants a different flooring system, that is a change order under design-build exactly as under any other method. The design-builder owns design errors. The owner still owns design decisions.

3.3.2 Competitive versus progressive design-build

Two distinct animals wear the same name.

Competitive design-build. The owner issues a request for proposals with performance criteria (sometimes a partial "bridging" design), and multiple teams each develop a preliminary design and a price. Selection is on best value, scoring technical approach and price together. Common on state department of transportation and federal work. The hidden cost is real: three or four teams each spend six figures on preliminary design knowing most will lose. Some owners pay a stipend to unsuccessful proposers. Teams price lost-proposal cost into everything they bid, so the market pays for it eventually.

Progressive design-build. The owner selects on qualifications alone, then works with that team through design open-book, negotiating a price at a defined milestone. Most progressive DB contracts include an "off-ramp": if the price is unacceptable, the owner pays for design to date, takes the documents, and goes to bid. It has grown fast, particularly in water and wastewater, because it gets the collaboration benefit without paying for four competing preliminary designs — structurally it behaves much like CM at risk with the design contract on the other side of the line.

3.3.3 Bridging documents and the criteria problem

Bridging documents are a partial design — typically 20–40% complete — that the owner develops with its own architect (the "criteria consultant") before soliciting design-builders. The design-builder completes the design from there.

Bridging is a compromise. It gives the owner more control over aesthetics, layout, and quality than a pure performance specification does. It costs some of the design-builder's ability to innovate and, importantly, it muddies the single point of responsibility — because now there is a design the owner furnished, and arguments about whether a problem originated in the owner's bridging documents or the design-builder's completion of them are exactly the arguments design-build was supposed to prevent.

⚖️ What the contract says — how design-build shifts the implied warranty.

Under design-build the owner is generally not warranting the adequacy of a full design, because it did not furnish one. That is the shift. But it is not as clean as owners assume:

• The owner does still generally warrant what it furnished — the performance criteria, the bridging documents, the geotechnical data, the survey. A defect in the owner's criteria is the owner's problem.
• The design-builder's obligation for the design is usually a professional standard of care — the care and skill ordinarily exercised by design professionals on similar work — not a warranty that the result will be fit for the owner's unstated purpose. Those are very different standards. A warranty means "it will work." A standard of care means "we were not negligent."
• Insurance follows that distinction, and this is where owners get hurt. Professional liability insurance covers negligence. It generally does not cover a contractual warranty or guarantee of a design outcome. An owner who negotiates a hard "fitness for a particular purpose" warranty into a design-build agreement may have created an obligation the design-builder cannot insure — which means if the design-builder's balance sheet is thin, the owner has a promise and no money behind it.

Bring that one to your attorney and your broker together, not separately. The question is: is this obligation insured, and by which policy?

3.3.4 Selection and fee structures

Selection is qualifications-driven in nearly every form of design-build, because on day one there is no design to price. Typical evaluation factors: past performance on comparable work, the specific team proposed (named individuals, not firm resumes), technical approach, schedule commitment, and management plan — then price, weighted anywhere from 20% to 50%.

3.3.5 The characteristic failure mode: you get exactly what you asked for

Design-build's failure mode is the mirror image of DBB's. In DBB the owner over-specifies and owns the consequences. In DB the owner under-specifies and owns those instead.

If your criteria say "concrete slab on grade suitable for warehouse operations" and do not name a floor flatness tolerance, do not name the material handling equipment, and do not require coordination with the racking vendor — you will get a slab that meets ordinary warehouse practice. Which is a completely reasonable interpretation. Which will not take a wire-guided turret truck in a 38-foot narrow aisle. And the argument you are about to have is lost before it starts, because the design-builder will hold up your criteria document and say, correctly, "this is what you asked for." That exact dispute is worked with numbers in case study 2.

The discipline design-build demands of an owner is front-loaded. Owners who are good at it write criteria in measurable terms: air changes per hour, foot-candles at the work surface, floor flatness values, sound transmission class ratings, energy use intensity targets, structural live loads, spare capacity in the electrical service. Owners who are bad at it write criteria full of the word "adequate."

🧩 Productive struggle. Take three minutes before reading on. An owner wants a building fast and does not want to manage a design team. Design-build gives them both. So why doesn't every owner use design-build for everything? Write down three real reasons — not "tradition," and not "public procurement law," which is real but too easy. Look for what the owner gives up.

3.4 CM at Risk: The Advisor Who Becomes the Risk-Taker

Construction manager at risk (CMAR, sometimes CM/GC) is the method Meridian chose for Northgate. It is the most common delivery method for complex institutional buildings in the United States, and it is the one this book's spine project runs on, so learn it best.

                        ┌──────────────┐
                        │    OWNER     │
                        └──┬────────┬──┘
              contract     │        │     contract
          ┌────────────────┘        └───────────────┐
          ▼                                         ▼
  ┌───────────────┐   PRECON COLLABORATION   ┌──────────────┐
  │  ARCHITECT /  │◄════════════════════════▶│  CM AT RISK  │
  │   ENGINEER    │   (no contract — but a   │ (holds the   │
  └───────┬───────┘    real, daily working   │     GMP)     │
          │            relationship from     └──────┬───────┘
    design consultants  schematic design on)        │
                                              subcontractors

The contractual picture looks like DBB. The behavioral picture is completely different, because the builder arrives eighteen months earlier.

3.4.1 The two phases, and the hinge between them

Phase 1 — Preconstruction services. The CM is engaged during design, usually at the end of schematic design or early in design development, for a negotiated fee — on a project of Northgate's size, typically a fraction of a percent of construction value. For it the CM provides progressively detailed estimates at each design milestone, so the design team learns what its choices cost while it can still change them; constructability review, which is a builder reading the drawings and saying "you cannot get a crane there," "that panel exceeds the trucking limit," "that detail needs three trades in an eight-inch cavity in a specific order"; value engineering as priced alternatives rather than indiscriminate cheapening; schedule, phasing, and logistics planning; long-lead identification; and market intelligence about which subcontractors are hungry and where escalation is coming from. All of it is Chapter 11.

The hinge — GMP conversion. At an agreed milestone the CM prepares a guaranteed maximum price: an open-book estimate of the cost of the work, plus general conditions, insurance and bonds, a construction contingency, a fee, and an escalation allowance — capped. Above the cap, the CM pays; below it, savings are usually split. This is where the CM stops being an advisor and becomes a party at risk, and it is the most dangerous moment in the method.

Phase 2 — Construction. The CM builds it, holding all the subcontracts exactly as a general contractor would — except the books are open, the contingency is a disclosed and administered line, and the owner sees actual cost.

3.4.2 The Northgate GMP, line by line

At 132,000 gross square feet that is $360 per square foot. Contract time 565 calendar days, notice to proceed March 3 of Year 1, contract substantial completion September 18 of Year 2, liquidated damages $5,500 per calendar day. Meridian's total project budget, all in, is $61,000,000.

Four things make this different from a lump sum. The contingency is visible — in a hard bid the contractor's risk money is buried and nobody but the estimator knows how much it is; here it is $1,320,000 on its own line with a monthly report on what has been drawn and why. Escalation is a named line, not a gamble. The fee is disclosed at 4.0%, so nobody argues later about the contractor's margin. And the savings split matters: on Northgate it is 75% owner / 25% Kestrel on unused GMP contingency, which aligns incentives better than any speech about partnership.

💰 Money check — how the savings split behaves. Suppose at final accounting $400,000 of the $1,320,000 contingency is unspent. (An illustration, not a forecast — you do not know this number until the last subcontract closes out.)

• Owner's share: 75% × $400,000 = $300,000 returned to Meridian
• Kestrel's share: 25% × $400,000 = $100,000 added to Kestrel's profit

What it means for the job: Kestrel's project manager has a personal, dollar-denominated reason not to spend contingency casually, and a reason to solve problems cheaply rather than write change orders. That $100,000 is roughly 5.5% on top of the $1,804,800 fee. Incentive design is not a soft topic; it is why the contingency gets managed instead of consumed.

3.4.3 The dual role, and the tension inside it

Here is the uncomfortable part the brochures leave out. During preconstruction the CM advises the owner on cost; after the GMP, the CM is exposed to that same cost. So during preconstruction the CM has a quiet incentive to build a comfortable number — richer allowances, healthy contingency, conservative productivity — because it will live inside that number for two years.

Owners manage this three ways, and good ones use all three:

• An independent estimate. The owner's own cost consultant prices the same drawings in parallel; where the two differ by more than a few percent, somebody explains why. On Northgate, Meridian's consultant came in at $46.9M against Kestrel's $47.5M, and the reconciliation meeting took six hours and was worth every minute.
• Open-book buyout with competitive subcontractor bidding. The GMP estimates the cost of the work; buyout sets it. If the GMP carried $4,100,000 for mechanical and the trade buys out at $3,860,000, the $240,000 does not vanish into Kestrel's pocket — it goes to contingency or back to the owner per the contract.
• A savings split worth having. 75/25 gives the owner most of the upside while leaving the CM enough to care.

🔍 Why this works. The mechanism is not trust. CMAR replaces price competition — which you cannot have when the design is incomplete — with cost transparency plus trade-level competition, which you can have at any design stage. The competitive pressure has not been removed from the system; it has moved from one competition among general contractors on bid day to twenty-five competitions among subcontractors during buyout, held under the owner's eye. That is why open-book matters so much. Close the books and CMAR becomes cost-plus with a cap and a handshake, which is a much worse deal.

3.4.4 The characteristic failure mode: a GMP set too early

CMAR goes wrong almost always the same way: the guaranteed maximum price gets set on drawings that are not complete enough to price.

An owner under schedule pressure wants a number for a board. The design is at 50%. The CM does not want to lose the job by saying "I cannot price this yet." So a GMP gets set with a page and a half of assumptions nobody on the owner's side reads carefully, and the number becomes real. Then design development continues, and every additional line the architect draws is either inside what the CM assumed or it is a change — and the fight about which runs for two years.

On Northgate, Kestrel set the GMP at roughly 85% construction documents with fourteen named allowances and a four-page qualifications list that Pri Sethi read line by line and challenged in seven places. That six-hour meeting is why the job's change-order history is manageable. Chapter 4 takes the GMP apart clause by clause.

🔄 Check your understanding. A CM at risk sets a GMP of $30,000,000 including a $900,000 construction contingency, with a 70/30 owner/CM savings split. At closeout, actual cost of the work plus general conditions plus insurance came to $28,600,000, and the fee was $1,150,000. Was there contingency left, and who got what?

3.5 CM as Agent, Multiple Prime, and the Rest of the Family

3.5.1 CM as agent (CMa / CM advisor)

CM as agent — CMa, "agency CM," or "CM advisor" — is construction management where the CM never takes construction price risk at all. The CM is a professional service provider, paid a fee, acting as the owner's agent.

                    ┌──────────────────────────┐
                    │          OWNER           │
                    └─┬────┬────┬────┬────┬───┬┘
       ┌──────────────┘    │    │    │    │   └────────────┐
       ▼                   ▼    ▼    ▼    ▼                ▼
  ┌─────────┐      ┌────────────────────────────┐   ┌─────────────┐
  │ ARCH /  │      │  TRADE CONTRACT #1 … #N    │   │ CM AS AGENT │
  │  ENG    │      │  concrete · steel · MEP ·  │   │ fee only;   │
  └─────────┘      │  drywall · roofing · …     │   │ no price    │
                   └────────────────────────────┘   │ risk        │
                                                    └─────────────┘

The owner holds every trade contract directly — eight of them or thirty. The CM manages, coordinates, schedules, and administers, but does not hold the money and does not guarantee the price.

This is the multiple-prime model, and in some jurisdictions it is required for certain public work: statutes in a handful of states have historically mandated separate prime contracts for general construction, plumbing, HVAC, and electrical. Those requirements have loosened in many places and vary considerably — working public in an unfamiliar state, find out rather than assume.

It fits owners with real internal construction staff (large universities, health systems with facilities departments, industrial owners with plant engineering), phased renovation in occupied buildings where packages release over years, developers who want direct relationships with trades they reuse, and jurisdictions that mandate it.

What the owner signs up for is every scope gap between trade contracts. Under a general contractor, when the drywall contractor and the framer each thought the other was installing blocking, that is the GC's problem — it signed for the whole building. Under multiple prime, that gap is the owner's. Add up forty small gaps and the "savings" from cutting out the GC's fee is gone, and then some.

⚠️ Safety alert — the multiple-prime coordination gap. Under a general contractor or CM-at-risk model, one entity has site-wide control of safety: it sets the rules, runs orientation, controls access, and has contractual authority over every trade on site. Under multiple prime, that authority is fragmented by design.

This matters legally as well as practically. Under OSHA's multi-employer worksite enforcement approach on construction sites (29 CFR 1926), duties can attach to an employer based on its role — the creating, exposing, correcting, or controlling employer. An owner or agency CM that in practice directs sequence, controls access, and tells trades when and where to work can find itself in the controlling-employer position with the exposure that carries, even though it wrote a contract saying safety was each trade contractor's responsibility.

If you run a multiple-prime job, do not leave site-wide safety to the general conditions of thirty separate contracts. Write one site safety plan, make participation a condition of every trade contract, run one orientation, and put one named competent person in charge of the site. Risk allocation generally is Chapter 6; safety as a system is Part IV.

3.5.2 Public-private partnerships and design-build-operate-maintain

Public-private partnerships (P3) and design-build-operate-maintain (DBOM) extend the bundle down the asset's life. A P3 concessionaire may design, build, finance, operate, and maintain a facility for 25 to 35 years, recovering its investment through tolls, availability payments from the agency, or user fees.

Two honest observations. The genuine value is life-cycle alignment: if the entity that builds the roof also maintains it for thirty years, it will not install the cheapest roof — an incentive no construction contract can replicate. The genuine cost is complexity and financing: legal, financial-advisory, and lender due-diligence costs only make sense above a substantial project size, commonly hundreds of millions, and private capital costs more than municipal debt, so the agency pays a premium for risk transfer and speed. Whether that trade is worth it is genuinely contested among serious people, and I will not pretend it has a settled answer.

You will most often meet P3 from inside: the contractor on a P3 is doing design-build, on a schedule dictated by financing milestones, with lender's-engineer oversight on top of the owner's.

3.5.3 Job order contracting

Job order contracting (JOC) serves the opposite end of the spectrum — high-volume, small, repetitive work for a public owner. The agency competitively procures a contractor for a multi-year term against a unit price book, a published catalog of construction tasks with unit prices. Contractors bid a single coefficient (an adjustment factor, e.g. 1.08 or 0.94) applied to every line in that book. When work comes up, the parties scope it, price it out of the book times the coefficient, and issue a task order — no individual bid cycle.

It is genuinely useful for roof replacements, classroom renovations, ADA upgrades, and deferred-maintenance backlogs, and a poor fit for anything with design complexity or novel scope. It requires disciplined oversight, because the integrity of the pricing depends entirely on scoping tasks honestly against the book — an area where both sides can misbehave and where audits regularly find that they have.

3.6 Integrated Project Delivery: The One Everybody Admires and Few Attempt

Integrated project delivery (IPD) is not CMAR with better manners. It is a structurally different arrangement.

        ┌────────────────────────────────────────────────┐
        │           ONE MULTI-PARTY AGREEMENT            │
        │   Owner  ·  Architect  ·  Constructor          │
        │   (+ key trade contractors joining the same    │
        │      agreement — often MEP and structure)      │
        ├────────────────────────────────────────────────┤
        │  • Shared risk / reward pool                   │
        │  • Target value design against a target cost   │
        │  • Mutual waiver of claims among members       │
        │    (with defined carve-outs)                   │
        │  • Co-located "big room" team                  │
        │  • Joint decision-making body                  │
        │  • Profit at risk before cost is at risk       │
        └────────────────────────────────────────────────┘

One multi-party agreement. Owner, architect, and constructor sign the same contract, not three bilateral ones; key trade partners join it. This eliminates the missing side of the DBB triangle entirely.

A shared risk/reward pool. Members' profit comes out of their individual contracts into a common pool; direct costs are reimbursed. Beat the target cost and the pool grows. Exceed it and the pool shrinks — members lose profit before anybody loses cost recovery. That ordering is deliberate: a problem in the mechanical scope costs the architect money too, which is the point.

Target value design. Rather than design and then estimate, the team sets a target cost first — from what the owner can afford and what the business case supports — then designs to it. Cost becomes a design input, not a design result.

Mutual waiver of liability. Members generally waive claims against each other except for defined carve-outs (willful misconduct, gross negligence, warranty obligations, insured claims). This is what makes "no blame" more than a poster in the trailer: if you cannot sue your teammate, your only remaining strategy is to solve the problem.

Co-location. The team works in one room — not a weekly meeting; a room, most days, for months. The transaction cost of asking a question drops to near zero, and that single change does more for coordination than any software.

The honest assessment. IPD produces excellent outcomes when it works, and it is rare for reasons that are not stubbornness. It requires a repeat, sophisticated owner who can staff a big room and decide at speed, and it asks a board to accept shared accountability instead of "we bid it and took the low number." Public procurement law often forbids it outright — waiving claims and sharing profit pools does not fit most competitive-bidding statutes. Insurance is complicated, because standard professional and general liability policies were not written for a mutual waiver among project participants. And it fails badly with an immature team: a shared pool without trust just means everyone watches it drain and starts protecting themselves.

Learn IPD's techniques even if you never sign a multi-party agreement, because the good ones travel. Target value design, big-room co-location, pull planning, transparent cost models, and early trade involvement all work fine inside a CM-at-risk contract. Most "IPD-ish" projects you meet are exactly that — CMAR or progressive design-build with IPD behaviors bolted on. That is not cheating; it is most of the benefit at a fraction of the institutional cost.

3.7 The Master Comparison

This table is the reference artifact of this chapter. Read down a column to understand a method; read across a row to understand a risk.

            0        6       12       18       24       30       36   (months)
            |--------|--------|--------|--------|--------|--------|

DBB         DDDDDDDDDDDDDDDDDDDDDD bbb aa CCCCCCCCCCCCCCCCCCCCCCCCCCCCC
                                                                     ▲ SC

CMAR        DDDDDDDDDDDDDDDDDDDDDDDDDDDD
              ss        gg
                    CCCCCCCCCCCCCCCCCCCCCCCCCCCC
                                               ▲ SC

DB          DDDD cc rrrr DDDDDDDDDDDDDDDDDDDD
                        CCCCCCCCCCCCCCCCCCCCCCCC
                                              ▲ SC

  D = design       c = owner writes criteria     r = RFP and selection
  s = CM selected  g = GMP negotiated            b = bid period
  a = award and bonds     C = construction       ▲ SC = substantial completion

📊 Diagram (described). Three horizontal bands. DBB: a long unbroken design bar, then bidding, then award, then construction — nothing overlaps, and substantial completion lands furthest right. CMAR: the same design bar, with the CM selected early inside it and the GMP negotiated about three-quarters through, and a construction bar that begins before design ends and runs shorter. DB: a short criteria period, a selection period, then design and construction heavily in parallel, finishing earliest. The horizontal distance between those three ▲ marks — not the price — is usually what decides the argument in the boardroom.

3.8 🚪 The Same Building, Two Prices

🚪 Threshold concept. Delivery method and contract type are one decision, and that decision is the price. You are not choosing a procurement process; you are choosing who absorbs the unknown.

Once you have this, you cannot go back to seeing the job the way you did before. Here is the arithmetic that installs it.

3.8.1 The counterfactual

Tomás Reyes, Kestrel's chief estimator, keeps a file he does not show to owners. When Kestrel wins a negotiated CM-at-risk job, he runs a second estimate: what would we have bid this at, cold, hard-bid lump sum, against a finished set?

He ran it on Northgate. His number was $45,850,000 — $1,650,000 below the $47,500,000 GMP.

Sit with that. On bid day, design-bid-build would have looked $1.65 million cheaper. Any board would have taken it. Harriet Voss would have taken it.

3.8.2 Where the $1,650,000 actually sits

Read the right-hand column. That is the whole lesson.

Nothing in that ladder changed the building. Same 985 tons of structural steel, same 38,500 square feet of curtain wall, same 33,000 square feet of slab on grade, same 412,000 pounds of ductwork. Every line is a risk changing hands or a margin changing size. Of the $1,650,000 gap, roughly $856,550 is real, permanent savings — the fee. The rest is risk repriced or risk deferred into the change-order and claims process.

3.8.3 Now run it to completion

Bid day is not the finish line. Here is what Pri put in front of the board — expected outlay, not contract price.

Difference: $885,250 in Meridian's favor — $6.71 per gross square foot on 132,000 SF, about 1.9% of the GMP.

And then the number that actually decided it.

💰 Money check — the schedule, which is the real money. Under DBB, Meridian would have had to finish 100% construction documents before advertising. That is roughly 122 additional calendar days of sequential design that, under CMAR, ran concurrently with early construction. Add a 35-day bid period and 21 days for award and bonding. Then add the 45 extra calendar days of construction duration.

• 122 CD + 35 CD + 21 CD + 45 CD = 223 calendar days ≈ 7.3 months later

Northgate's contract substantial completion is September 18 of Year 2. Meridian's leased interim clinic space expires October 1 of Year 2 — thirteen days of margin. Push substantial completion 223 days and you land near the end of April in Year 3.

Meridian's interim clinic lease runs $118,000 per month. Extending it 7.3 months: 7.3 × $118,000 = $861,400 — before a dollar of deferred clinic revenue, and before the operational cost of running a split practice for another two seasons.

Add the $885,250 modeled construction advantage to the $861,400 lease exposure and CM at risk is worth roughly $1.75 million to Meridian — against a GMP that looked $1.65 million more expensive on bid day.

What it means for the job: this is Theme 2 in its purest form. The schedule and the budget are the same conversation. An owner who evaluates delivery methods on contract price alone is reading one column of a two-column ledger.

3.8.4 Before and after

🪞 Learning check-in. Stop here. Genuinely stop — take four minutes with a pen.

1. Which method felt "obviously best" while you were reading it? Almost everyone has one. Write it down. Now write the specific project characteristics that would make it the wrong choice. If you can't produce three, you have a preference rather than an understanding — go back to the "characteristic failure mode" row in §3.7.

2. Where did you resist? Most readers resist one idea in this chapter. For some it is that the low bid is not the low price. For others it is that a contractor paid a disclosed percentage fee could possibly be cheaper than one competing on price. Name what you resisted and write one sentence on why — is the evidence thin, or does it conflict with something you already believed?

3. The retention test. Without looking: name the four things a CM-at-risk GMP is built out of above the direct cost of the work. Fewer than three means you read it rather than retrieved it. Cover the table in §3.4.2 and write it from memory. That single act of failed retrieval followed by checking will hold better than four more readings.

4. The habit to start now. From here forward, every time you meet a project — in this book, at work, in a news story about a stadium — ask one question before any other: who owns the unknown here? If you cannot answer it, you have found the thing that is going to hurt somebody.

3.9 How Owners Actually Choose

3.9.1 The decision tree, in prose

Start here: are you legally free to choose? A public agency's procurement statute may decide this outright. Many jurisdictions have authorized design-build and CM at risk for public work over recent decades, often with conditions — size thresholds, prequalification, a written justification, sometimes state board approval. Others still default hard to competitive sealed bidding. This varies by state, by agency type, and by year. There is no point scoring a method you cannot legally use.

If you are free: is the design complete and confidently coordinated? If yes, the project is conventional, and you are not in a hurry — design-bid-build. Take the competitive price. It is defensible and often genuinely lowest, and nobody should talk you out of it because it seems old-fashioned.

If not: do you know what you want in measurable terms? If you can specify performance in numbers, your users' requirements are stable, and you would rather have speed and a single accountable party than day-to-day design control — design-build. If requirements are still moving because your operators are still figuring out their process, do not choose it. You will pay for every discovery at a negotiated price.

If requirements are still moving: how much price certainty do you need, and when? Fixed board- or bond-approved budget, a hard number needed before design finishes, constructability input on complex systems — CM at risk. Set the GMP as late as the schedule allows and no earlier than 60% documents.

Real internal construction staff, a desire for direct trade relationships, or a statute requiring multiple primes — CM as agent. Budget for the scope gaps.

A repeat owner with a continuous capital program, a mature team, staff you can put in a room, and legal freedom to sign a multi-party agreement — IPD. Missing any one of those five? Take the IPD techniques into a CM-at-risk contract instead.

3.9.2 The scoring worksheet

Prose trees handle one factor at a time; real decisions have seven pulling different directions. Here is the weighted worksheet Pri filled out, with Meridian's weights. Weights (1–5) reflect what matters to this owner on this project. Scores (1–5) reflect how well each method fits that factor.

Check the arithmetic yourself: DBB = (5×1)+(4×1)+(3×4)+(5×3)+(4×1)+(2×5)+(3×4) = 5+4+12+15+4+10+12 = 62. CMAR = 20+20+12+25+20+10+9 = 116.

3.9.3 Three warnings about scoring models

Use that worksheet, and distrust it.

Three points is not a margin. CMAR beat DB 116 to 113 — well inside the noise of anybody's subjective 1–5 scoring. A scoring model does not make the decision; it organizes the argument so you can see which factors are driving it and test whether you believe them.

The weights carry all the meaning. Change schedule urgency from 5 to 2 and DBB climbs while DB falls. The honest work is defending the weights, not doing the arithmetic. Make the owner say out loud why schedule is a 5.

A model cannot see a tiebreaker that is not one of its rows. Which is exactly what happened at Meridian.

🔄 Check your understanding. An owner scores design-build highest and chooses CM at risk anyway. Give two legitimate, non-arbitrary reasons that could justify overriding the score.

3.10 What Each Method Means for You — and How Meridian Decided

3.10.1 Your job changes with the method

You will spend your career inside somebody else's delivery-method decision. Here is what each one does to your working life.

Notice the pattern down the leverage column. In every method, your leverage exists earliest and decays fastest. In DBB it is the pre-bid question and the first thirty days. In CMAR it is preconstruction. In DB it is criteria review before signature. By the time the problem is visible in the field, in all five methods, you are negotiating from weakness with money already spent. That is Theme 3 in contractual terms: the project is built twice, and the first build — on paper — determines the second.

3.10.2 Why Meridian chose CM at risk

Pri went back to the board in November and recommended CM at risk. Four reasons, in her order:

A phased opening. Meridian needed ground-floor clinics and the imaging suite first, with ambulatory surgery and upper-floor clinics following. Phased occupancy requires the builder in the room during design — where the phasing lines go, how temporary separations run, how life-safety systems stay compliant in a partially occupied building, how the certificate of occupancy sequences. Under DBB the phasing plan is drawn by an architect and priced by a contractor who has never seen it, and the gap between those two documents is a change order.

An occupied adjacent clinic. The existing clinic on the north property line stays open every day of construction. Patients, deliveries, and staff share a boundary with a 6.2-acre site — including 44,000 cubic yards of mass excavation and 32,000 cubic yards of net export hauled past a working medical practice. That is a construction problem, and it had to be solved while the site plan was being drawn.

A fixed budget. Meridian had $61,000,000 and would not get more. CM at risk gives a real cost model at every design milestone, so the design team learns what its decisions cost while it can still change them. Under DBB you find out on bid day, and if bids come in high your options are redesign (months) or cut scope under duress.

Constructability input on the imaging suite. Imaging is the hardest room in an outpatient building: structural loading, shielding, vibration criteria, magnetic clearances, equipment access paths, and a slab that has to be right the first time. Kestrel ran it through three constructability reviews before it was drawn at 100%.

And what Meridian gave up. Pri wrote this section herself, and I have quoted it to owners ever since. Meridian gave up market price testing — there is no bid tab, and the defense (independent estimate, open-book buyout, savings split) is a constructed defense rather than a market one, requiring staff to run it. Meridian paid a higher fee: 4.0% negotiated exceeds a hard bidder's competitive overhead and profit by roughly $856,550, per §3.8.2. And Pri took on fourteen months of preconstruction meetings and learned to read cost models she had never read before.

CM at risk also did not solve the change-order problem — it shrank it. When Meridian's imaging 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, that was a scope change and it cost money exactly as it would under any method. Change order #14 is coming, it is going to be ugly, and the delivery method will not save anybody. (Chapter 6, and dissected in Part VI.)

3.10.3 The contrast: Rivermont Elementary School #12

Across town, Curtis Boone was running Kestrel's other job: Rivermont Elementary School #12. $22,400,000, public owner, hard-bid design-bid-build lump sum, prevailing wage, 100% payment and performance bonds.

An 11.1% spread from low to high on a fully designed building is a signal, and the signal is: the documents are not clear, and six bidders read them six different ways. A tight spread means everybody priced the same building. A wide spread means somebody is about to be very surprised, and it is usually the low bidder.

In week 9, Curtis's crew opened the gymnasium roof coordination and found that the structural and mechanical drawings had never been reconciled: the main supply duct and the roof joist wanted the same space. Curtis wrote an RFI. The architect routed it to the structural engineer, who had to coordinate with the mechanical engineer — neither of whom had a contract with Curtis, both of whom were working on a design fee fully earned eleven months earlier. The answer came back in week 14: five weeks, and it required deeper joists in the gym bay. Curtis priced it at $214,000 plus 21 calendar days. The architect denied it — coordination is a contractor responsibility under the general conditions, and the contractor had a duty to review the documents and report discrepancies. Curtis filed a claim. Eleven months later it settled for $96,000 and zero days of time extension.

Now watch what that did to the behavior, because that is the real lesson. After the denial, Curtis stopped writing early RFIs. He started building to his own interpretation and arguing afterward. His reasoning was not stupid: an RFI had cost him five weeks and gotten him denied anyway, so why volunteer? The delivery method had taught him — correctly, from inside his own experience — that asking questions was expensive and being right was not rewarded.

That is not a character flaw. That is an incentive structure working exactly as designed.

🏗️ From the field. Margo Deacon, thirty-one years a superintendent, put it to me once in the Northgate trailer while we were arguing about something else entirely.

"You want to know the difference between these two jobs? On this one, when Dale Whitcomb draws something I can't build, I call him and we fix it in an afternoon, and it costs nobody anything because the concrete isn't poured yet. On Curtis's job, when the architect draws something he can't build, the only move he's got is a piece of paper that takes five weeks and gets denied. Same two guys. Same problem. Different contract."

She is describing risk allocation, and she has never used the phrase in her life.

📋 Try it. Three owners walk in on the same afternoon. Recommend a delivery method for each. Give a two-sentence justification and name one specific risk you are accepting. Then check yourself.

Profile A — Municipal fire station. City of Rivermont. 14,500 SF, three apparatus bays, living quarters, training tower. Funded by a state grant requiring substantial completion within 26 months of award or the money is returned. Design is at 15%. The fire chief has strong opinions about apparatus bay layout and is two years from retirement.

Profile B — Biotech tenant fit-out. 32,000 SF of lab and office inside an existing shell. The tenant has not selected its major analytical equipment package — mass spectrometers, imaging, cold storage — and will not for another five months. The vibration, power, exhaust, and gas requirements of that equipment will drive the entire MEP design. Move-in is contractually required in 15 months.

Profile C — 400-unit apartment complex. A repeat developer who has built eleven similar garden-style complexes with the same architect and the same prototype in nine years. Knows its cost per unit within about 3%. Private financing with a hard interest carry of roughly $340,000 per month once drawn.

Spaced Review

Cover each answer and retrieve from memory first. Getting it wrong and then checking beats reading it again.

From Chapter 1 — the project lifecycle. Name the phases from an owner's first idea through occupancy. Then the harder part: at which phase does the construction manager enter under each of the five delivery methods?

From Chapter 2 — contractual tiering and thin margins. Sketch the tiers from owner down to the person installing the work, and recall roughly what net margin a general contractor earns on a hard-bid commercial building. Then connect it: why does a 2% net margin make the delivery-method decision existential for a contractor?

Deep callback. Chapter 1 claimed that construction management is the disciplined management of risk under uncertainty. Write one sentence connecting that to what you read today. If your sentence contains the word "who," you have it.

Project Checkpoint: The Willow Street Delivery-Method Comparison Memo

In Chapter 2 you built a market and project-type analysis for the Willow Street Community Center — who the players are and what each of them wants. Pull it back out.

The project. City of Rivermont Parks & Recreation. $6,800,000, 24,000 SF, two stories: wood-framed second floor over a structural steel and CMU first floor. Gymnasium, two multipurpose rooms, commercial kitchen, offices, locker rooms. Design-bid-build, lump sum. 425 calendar days. Liquidated damages $1,200 per calendar day, 5% retention, 100% payment and performance bonds, prevailing wage. Site: 2.1 flat acres with one existing 8-inch water main to relocate. Full package: Appendix K.

Deliverable: a one-page memo to the City. One page. If it runs to two, you have not finished thinking. Address it to the Parks & Recreation director, in four sections.

1. Why design-bid-build was chosen — stated fairly. Do not straw-man it. A parks department spending public money has real reasons: a probable statutory requirement for competitive sealed bidding, a defensible public record, an entirely conventional building type, limited staff to administer anything more involved, and a design that can genuinely be finished before bidding. Write the strongest honest version.

2. What it costs the City in risk. Under DBB the City owns design errors and omissions through the implied warranty of the documents it furnishes. Model the exposure: 4% of $6,800,000 is $272,000 — money the City has not budgeted unless it carried an owner's contingency. Then name three specific coordination risks on this building: the bearing and transition detail where the wood-framed second floor meets the steel and CMU first floor; the commercial kitchen's hood, gas, grease-waste, and make-up-air coordination; and the 8-inch water main relocation, where a differing site condition is the City's risk under most standard general conditions. Say who owns each and why.

3. What it costs the City in schedule. Construction cannot start until documents are complete, bids are taken, and award is made. Estimate that front end honestly — a 4-week bid period plus 3 weeks to award and bond is a reasonable 49 calendar days after the documents are done. State what that means against 425 calendar days of contract time. If the City wants to open before a specific recreation season, put a date on it.

4. What you would have recommended instead, and what it would have cost them. Take a position. The defensible answer is probably CM at risk if the City's statute permits it, for one reason: the kitchen and the wood-over-steel-and-CMU transition are precisely the details that benefit from constructability review before they are drawn at 100%. Write the other side too — a negotiated CM fee on a $6.8M job is real money against a small budget, and the City may not have staff to sit in preconstruction for eight months.

Close with one sentence naming the single risk you would most want the City to price and own deliberately rather than discover.

File this as document 3 in your project notebook. In Chapter 4 you build the contract-type risk map for Willow Street — clause by clause, which provisions move which risk onto you, with dollar exposure attached. This memo is its foundation: you have just decided who owns the unknown. Next you find out which words in the contract say so.

Chapter Summary

The one-line version: you are not choosing a procurement process; you are choosing who absorbs the unknown — and whoever absorbs it charges for it.

The decision, in seven questions, in order:

1. Does procurement law leave me a choice? (If no, stop.)
2. Is the design complete and coordinated? (If yes and it's conventional: DBB.)
3. Can I specify what I want in measurable numbers, with stable user requirements? (If yes: DB is live.)
4. Do I need a hard price before design is done? (If yes: CMAR is live.)
5. Do I have real internal staff and want direct trade contracts? (If yes: CMa is live.)
6. Am I a repeat owner with a mature team, staff for a big room, and legal freedom for a multi-party agreement? (All five, or take the techniques instead of the contract.)
7. Which risks am I accepting, and have I priced them?

Four things to carry out of this chapter:

• Delivery method, contract type, and selection method are three decisions. Never let someone tell you one and assume you know the other two.
• The bid-day price and the final price are different numbers, and the delivery method lives in the gap. On Northgate the hard-bid number was $1,650,000 lower and the modeled outcome $885,250 higher — before the $861,400 of interim clinic lease the seven-month schedule difference would have cost.
• Your leverage is greatest before anything is built, in every method. In DBB it is the pre-bid question and the first thirty days; in CMAR it is preconstruction; in DB it is criteria review before signature. After that you negotiate with money already spent.
• When you cannot answer "who owns this risk," you have found the thing that will hurt you. That question is the whole job, and it does not stop being the whole job when you get promoted.

What's Next

Chapter 4 goes inside the box you just chose: lump sum, cost-plus, GMP, unit price, and the hybrids — what each guarantees, what each does not, and the specific clauses that move risk across the table while everyone is shaking hands. It also takes apart the sentence that trips up more owners than any other in this business: a guaranteed maximum price guarantees the contractor's exposure, not the owner's cost. Then Chapter 5 gives you the legal framework — liens, bonds, insurance, and disputes — standing behind every promise in all of those contracts.