Construction Procurement Methods Compared

What does "procurement method" actually mean?

Most arguments about procurement are really three separate decisions being discussed as one. Separating them makes the comparison tractable, because an owner can mix them in almost any combination.

Delivery method — who holds a contract with whom. Design-bid-build, design-build, CMAR and IPD are contractual structures. They decide whether the designer and the builder answer to the owner separately, to each other, or to a shared agreement. Everything about risk allocation follows from this.

Selection method — how the owner picks. Low bid, best value, or qualifications-based. Public owners are often constrained here by statute in ways private owners are not. Selection is independent of delivery: a design-build project can be awarded on low price, and a design-bid-build project can be awarded on best value.

Payment type — how money moves. Lump sum, cost-plus, cost-plus with a guaranteed maximum price, or unit price. This is the decision people are describing when they say "contract type," and it can be layered onto most delivery methods. A GMP is not a delivery method; it is a payment ceiling that happens to be common under CMAR.

Design-bid-build: the owner owns the documents

The owner holds two separate contracts — one with the designer, one with the builder — and the design is substantially complete before contractors see it. Bidders price the same documents against the same deadline, which is exactly why owners and public agencies keep using it.

Risk splits cleanly. Under the Spearin doctrine, which US courts have generally applied to owner-furnished plans and specifications, an owner is often held to impliedly warrant that they are adequate for the work. How far that warranty extends, and whether contract language can disclaim it, varies by jurisdiction and by contract — confirm it with counsel for your state and form. Errors, gaps and conflicts in the documents are typically the owner's exposure, and the contractor's usual route to recovery is a change order — subject to the contract's notice requirements, disclaimers and any duty to report obvious discrepancies. Everything downstream of the documents — means, methods, production rates and the price itself — belongs to the contractor.

It suits an owner with a settled program, no urgent schedule, a statutory obligation to competitively bid, and a preference for a defensible paper trail over discretion in selection.

The inconvenient part: low bid does not select the best contractor, and the low number is often the one carrying the largest omission or the thinnest risk allowance. Owners who then fight change orders line by line are fighting a consequence of their own selection method. For the estimating side, design-bid-build is the highest-pressure format there is — one number, complete documents, a fixed deadline, and no second look.

Design-build: one contract, and the design risk moves

The owner signs one contract covering both design and construction. Scope is defined by performance criteria, a program narrative, or bridging documents prepared by an owner's consultant, rather than by a finished set of drawings.

The design risk moves with it. For the portions the design-builder designs, the owner is no longer warranting adequacy — the builder is responsible for delivering a result that performs. Two qualifications matter. Owner-furnished criteria and bridging documents still carry the owner's implied warranty, and the seam between what the owner prescribed and what you developed is the most litigated question in the method, so mark that boundary explicitly in your proposal. And an obligation to deliver a result that performs is a warranty, which professional liability policies written to a negligence standard commonly exclude; confirm what your designer's policy actually covers before signing performance language. Single-point responsibility is the real product design-build sells, and it removes most of the designer-versus-builder finger-pointing that design-bid-build generates.

The estimating consequence is severe and often underestimated: you are pricing a scope you have not designed yet. The number is only as good as the clarification and assumption log attached to it, which functions as the scope definition until drawings exist. Design contingency has to be tracked separately from construction contingency, because it is consumed as the design develops rather than held against field conditions. Escalation is a third line and not a contingency at all: any price committed before construction has to be escalated from the date of your cost data to the midpoint of construction, at a stated rate, shown on its own so the owner can see it and argue with it. Carrying escalation inside contingency is the most common way an early number goes wrong without anyone noticing until buyout.

It suits schedule-driven owners, repeatable building types, and owners with enough in-house technical capacity to write meaningful performance criteria. The honest limitation is that cost savings are not guaranteed: outcomes depend heavily on the quality of the owner's criteria and the competitive field. The owner trades price transparency and design control for speed and single-point responsibility. And if the program is unsettled, every owner change becomes a contract amendment negotiated with the only party that can price it.

CMAR: pricing a building that is not designed yet

Under construction manager at risk, the owner keeps a separate contract with the designer and brings a construction manager on during design under a preconstruction services agreement. The CM advises on cost, constructability and phasing, then converts to an at-risk contractor holding a guaranteed maximum price. Because the designer is still under the owner's contract, design risk largely stays with the owner — this is the main structural difference from design-build.

The GMP is set before documents are complete, so the gap is carried explicitly: a dated basis-of-design document list, a clarifications and exclusions log, allowances for undesigned scope, and a stated CM contingency. Books are typically open to audit, with unspent contingency and buyout savings split on a formula agreed up front.

CMAR typically carries one of the heaviest estimating loads, because the same building is priced repeatedly as the design develops — at concept, schematic design, design development and construction documents — and on public work the CM's number is often reconciled against an independent estimate prepared for the designer. Producing the estimates is the smaller half of the job — reconciling them is the deliverable. An estimate that moved between milestones and cannot be explained line by line, with causes, costs the CM credibility faster than the variance itself does.

The limitation worth stating: a GMP set early buys certainty with contingency the owner funds whether or not it is used, and savings-sharing returns only part of it. A CM converting to at-risk also has a straightforward incentive to price conservatively at the moment of conversion, which is precisely why the milestone estimate history matters to the owner.

IPD: the cost target comes before the design

Integrated project delivery puts the owner, designer and builder — and sometimes key trade partners — on a single multiparty contract with a shared risk-reward pool. Profit is placed at risk against collective project outcomes, and the signatories agree to limited mutual waivers of liability, with carve-outs negotiated in the agreement.

The estimating model inverts. Under target value design, an allowable cost is set first and the design is steered to hit it, so cost modeling is continuous and built into design decisions rather than performed periodically to validate them. Estimators sit with design clusters instead of receiving their output.

The limitation is availability. IPD generally suits a sophisticated, repeat owner. It can sit awkwardly with public procurement statutes, conventional construction lending, and surety and insurance products built around single-prime liability, though the rules differ by jurisdiction and some states have expanded the delivery methods public owners may use — check your own statute. Projects described as IPD are not always structured as multiparty agreements; many are CMAR with collaborative practices attached — which is a legitimate and far more obtainable middle ground, and worth asking about directly when an owner uses the term.

JOC and unit-price contracts: rates first, scope later

Both of these invert the usual sequence. Rates are agreed before anyone knows the specific work, and pricing an individual job afterward is mostly arithmetic. That makes them efficient for owners with recurring work, and it concentrates all of the contractor's estimating risk into a single decision made at award.

Job order contracting works from an owner-adopted unit price book covering construction tasks with pre-set prices. The contractor competitively bids a coefficient — an adjustment factor — applied to those book prices, and individual work orders are priced by selecting the applicable line items and extending them. It is common in public works, K-12 and higher education facilities, and federal agencies. The estimating skill is not price development; it is scope identification and completeness. Scope you fail to identify before the work order is executed is difficult to recover; adding it usually means revising the work order or pricing it as a non-prepriced item, both of which require owner agreement.

The JOC limitation is structural. The coefficient is bid with the least information you will ever have about the actual work, and it fixes your margin for the performance period it covers. Read how the contract handles renewal before you treat that as permanent: many programs re-bid or index the coefficient at each option year and adopt an updated price book annually, and many require separate coefficients for normal hours, after hours and secure or occupied areas. Where none of that applies, you carry the drift for the full term. Unit price books are updated on a fixed cycle rather than continuously, so they can lag a fast-moving market: in a volatile materials period the contractor absorbs the difference between book pricing and what the supplier quotes this week.

Unit-price contracts work differently. The owner's engineer publishes estimated quantities, the contractor bids a rate per measured unit, and payment follows remeasurement of what was actually installed. Quantity risk therefore sits with the owner, which is why the method dominates earthwork, utilities and paving, where subsurface quantities genuinely cannot be known at bid.

The estimating trap is distribution. Time-related costs land differently depending on how the bid schedule is built. Public unit-price schedules often include separate pay items for mobilization — commonly capped at a percentage of contract value stated in the specification, and paid against milestones — plus traffic control and field facilities, so only what those items will not carry has to be spread across production units. Where the schedule offers no such items, all of it rides on the unit rates, which creates a standing temptation to unbalance the bid — loading early-paying items or items you believe will overrun. Owners screen for it, and many contracts include a variation-in-quantity or changed-quantities clause that allows a unit price to be adjusted once the actual quantity passes a stated threshold — read the specific clause, because both the threshold and who may invoke it vary by form. Your exposure on this work is production rates and material escalation together. The quantity is not yours to control and the rate is fixed, so a job running across several seasons carries fuel, liquid asphalt, aggregate, cement and pipe escalation at a rate you bid in an earlier market. Check whether the contract includes fuel and bituminous price-adjustment clauses before you assume the risk is covered — where it does not, that escalation is entirely yours.

How the methods compare on risk

Where the exposure sits once the contract is signed.

When does the work actually get priced?

This is the difference that changes how an estimating department is run, and it is usually left out of procurement comparisons entirely. What matters is how complete the documents are at the moment you commit, and how many times you commit.

Design-bid-build is one estimate against near-complete documents, compressed into a bid window. Technique is settled — detailed quantity takeoff, priced and extended — and everything rides on catching what the other bidders missed without missing something yourself.

Design-build and CMAR are a sequence of estimates against documents that change underneath them. The first number may be parametric or assembly-based because there is nothing to take off. The last one is built mainly on bought-out trade packages — subcontractor bids received, leveled and scope-checked against the documents — with the CM's own detailed takeoff covering scope not yet bid and acting as the yardstick for what comes in. A GMP resting on internal takeoff alone, with no market pricing behind the major trades, is a number you will be defending every month until closeout. These are different estimating techniques producing comparable outputs, not one estimate progressively refined, and the handoff between them is where variance is introduced.

That is why reconciliation dominates the effort on multi-milestone work. "The design developed" is not an explanation of a cost movement. A defensible reconciliation shows the variance at whatever level the documents actually support, with causes separated into scope added, design detail resolved, quantity corrected, escalation, and market movement. At concept and schematic design that means UniFormat elements — substructure, shell, interiors, services — because there is no trade detail to reconcile. MasterFormat divisions become the right frame once design development pricing is built trade by trade, and the crosswalk between the two frames is itself something you should be able to show. Owners who have been through one GMP process ask for exactly this, and estimators who cannot produce it lose the argument on every subsequent milestone.

JOC and unit-price work spread pricing across many low-drama events, but the decision that determines profitability — the coefficient or the unit rate — was made at award. Pricing a work order afterward is bookkeeping compared to that.

The staffing consequence is real. Hard-bid pipelines spike hard around bid dates and go quiet between them. CMAR and design-build pipelines are steady, long, and demand continuity from the same people. Those are different capacity models, and a contractor who works across both methods is chronically either idle or underwater.

What to settle before bidding under an unfamiliar method

A short list of the things that most often go wrong the first time a contractor works outside its usual structure.

• Name the document set your price is based on, by date and revision, on the face of the estimate.
• Separate design contingency from construction contingency, and state which one your number includes.
• Write the assumption, clarification and exclusion log before the number, not after — under design-build and CMAR it is the scope definition.
• Confirm whether selection is low bid or best value; a best-value proposal that competes only on price ignores whatever share of the evaluation the solicitation assigns to non-price criteria.
• On unit-price work, read the measurement method and the variation-in-quantity clause before setting rates — how a unit is measured changes what it costs.
• On JOC, price the coefficient against the work you actually expect to be issued rather than the catalog average, and re-test it against the market before extending the term.
• Confirm who pays for preconstruction services and whether that fee is separate from, or credited against, the construction contract.

Where outsourced estimating fits, and where it does not

Procurement method changes the answer to whether external estimating support is worth anything, and the answer is not uniformly yes.

Hard-bid design-bid-build: the clearest fit. Complete documents, self-contained scope, and demand that spikes around bid dates. Everything an external estimator needs is in the plan set, and the work is genuinely portable. This is where outsourced takeoff and estimating earns its place, and it is the strongest case for it.

CMAR and design-build milestones: workable, with one condition. The volume is real and recurring. But the value in this work is reconciliation, and reconciliation depends on whoever holds the assumption history from the previous milestone. A firm that rotates estimators between your SD and DD estimates will generate variance it cannot explain. If you outsource this, insist on continuity of the same estimator, or keep reconciliation in house and outsource the takeoff underneath it.

Early design-build conceptual pricing: usually keep it. At that stage the number is mostly judgment about design intent you have not published yet, and commercial calls about what you are willing to carry. Nobody outside your business can make those calls, and no plan set communicates them.

JOC: the line-item work outsources, the coefficient does not. Matching field scope to catalog line items completely and correctly is systematic, document-driven work that transfers well. Bidding the coefficient is a decision about your own cost structure, backlog and market view, and it stays with you.

And the version that costs us work. If you have a full-time estimator, one delivery method, and a steady pipeline that matches their capacity, outsourcing probably adds coordination overhead for no gain. External estimating support earns its keep at peaks, on trades or methods you do not normally touch, and on volume you would otherwise have to no-bid. Outside those cases, it is worth being honest that you do not need it.