Lump-Sum Is a Cost Floor, Not Certainty: Why Atlantic GCs Are Walking Away From Low-Bid

A low-bid lump-sum number is not what your project will cost. It is the least it can ever cost. Travis Rudolph, the chief estimator who runs EllisDon's Atlantic numbers, says it to owners flat, the way you'd explain it across a job-site tailgate: in a lump-sum world, "that's your cost minimum — it just goes up from there" (Rudolph, EP 73). The signed contract price is the floor. Change orders, RFIs, and the gaps in an incomplete drawing set move the cost in exactly one direction from there, and the contract is built so they can.

What actually happens when the low number wins is best told by the people who've lived it — and lately they've started saying it out loud. Three of them, rivals who bid against each other for a living, said it together on the record.

The decisions that blow your budget are made before you ever see a price

Before a single contractor sees a public tender, the expensive choices are already locked. The structural system, the mechanical approach, the site logistics, the phasing — all of it gets set in a design room you weren't invited to. And the most damning receipt for what that costs owners didn't come from a procurement study. It came from a former architect.

Rick Buhr crossed from the design side to Bird Construction Atlantic, and he watched, from the inside, how design teams actually spend their meeting time. The proportions ran backwards to the money. "We'd spend 95% of the time talking about architecture and 5% talking about structure, mechanical, electrical," he said (Buhr, EP 73). Sit with that for a second, because it's the whole problem in one number. On a real institutional building, the mechanical and electrical scope is 40 to 50 percent of the construction cost. So the design team is spending 5 percent of its attention on half the budget — and then that half-baked half gets handed to the street as a fixed-price tender, with no contractor in the room who could have said that system will eat three months of schedule or that bay spacing costs you a hundred grand in steel for nothing. "You miss the opportunity for value engineering, constructability reviews, schedule reviews," Buhr added. By the time you, the owner, see a price, every one of those misses is already baked into the drawings — and the only tool left to fix them is a change order with a markup on top.

That is not a contractor being difficult. It is the model doing exactly what the model does.

Three firms that bid against each other agreed on the same thing — in public

Here is what made me take notice. EllisDon, Bird, and the other national outfits working Atlantic Canada are competitors. They bid the same jobs against each other; they poach each other's people; they lose work to each other every quarter. So when Rudolph, Buhr, and Vivek Tomar sat down at the same table and converged — without hedging — on the conclusion that hard-bid lump-sum is the wrong default for complex work, that agreement is itself the news. Rivals don't align in public on a self-serving story unless the story is just true. "Lump sum is not the best for the stakeholders, and it's not the best for the subs," Rudolph put it on a separate episode. "Anything collaborative" beats it (Rudolph, EP 48). When the firms who'd profit most from the change-order game are the ones telling you to stop playing it, that's worth more than any white paper.

Why the lowest bid is a warning sign, not a deal

The mechanism is not a character flaw in some contractors. It's arithmetic, and once you see it you can't unsee it. A contractor handed an incomplete set has exactly two rational moves: price all the risk honestly — load real contingency against every ambiguity — and lose the bid to someone who didn't; or price it thin, win the work, and recover the difference later through change orders. The rational response to being forced to carry all the risk is to either price that risk fully and lose, or underbid and recover through change orders. The structure rewards the second move every time. The contractor who priced it straight is the contractor sitting at home.

The contract that does this is named, and it's the one under most public lump-sum work in Canada: CCDC 2, the Stipulated Price Contract. Under CCDC 2 the general contractor carries maximum cost risk — material and labour escalation over the life of the job lands on the contractor, not the owner — and anything outside the original scope requires a formal change order. Each of those change orders typically carries a 15 to 20 percent overhead-and-profit markup. So scope that was in the building but absent from the drawings — an everyday occurrence, given that design-related problems hit nearly half of all Canadian projects — gets priced at a margin the GC could never have named in the original bid without losing the work. The owner pays for the gap, plus markup, at the worst possible leverage point: after the contract is signed and the trades are on site.

And there's a timing trap underneath all of it that owners rarely hear about. Sub prices don't hold. Rob Clinch, who runs Avant Garde Construction and Management in Moncton, has watched the quote window collapse in real time. "My price is good for 36 hours," he said, quoting a typical sub (Clinch, EP 63). A tender package carried for two weeks of competitive bidding can close on trade prices that are already fiction. The number in the sealed envelope is stale before it's opened.

None of this is a secret in the trade. Ontario Construction News laid it out plainly: some contractors bid below cost on purpose on public lump-sum work, expecting to recover through change orders — "the lowest bid isn't the lowest cost; it's a warning sign." The policy people have caught up. Canada's Office of the Procurement Ombud, in its 2025 best-value study, concluded that lowest-price procurement fails to capture total cost of ownership and routinely produces outcomes that work against the owner's own objectives. And the cleanest number in the whole debate comes from a peer-reviewed study of 118 design-bid-build projects in the International Journal of Construction Education and Research: best-value contractors produced 37 percent less cost growth during construction than lowest-bid contractors, and owners paid less than one percent more at award to get that. One percent more up front. Thirty-seven percent less bleed after. The low bid is frequently the most expensive way to buy a building.

A general contractor's real job is managing risk — so why does low-bid punish doing it well?

If you want the cleanest frame for all of this, it comes from Ian Boyd, who co-founded Iron Maple Constructors after years at Bird and Rideau. Strip away the contract forms and the procurement jargon, he says, and the job is one thing: "What we are as risk managers — and if you do that well, I think you'll be successful" (Boyd, EP 23). That single reframe is the most useful idea in this whole essay, and it reorganizes the entire lump-sum question. A bid is not a price competition. It is a risk-allocation decision wearing a price tag.

Run the low-bid model through that lens and the perversity is obvious. When you bid thin on an incomplete set, you are not just accepting a smaller margin — you are accepting risk you cannot price, because you don't have the information to price it. The contractors who price all the risk lose. The ones who win are the ones who took the risk cheapest, which is usually the same as the ones who under-priced it. The system selects against the discipline that defines good contracting. Boyd was blunt about where that leaves you: "Bidding lump-sum work where you're totally commodity-driven eats the low price every time — not a very fun place to live" (Boyd, EP 23). So Iron Maple made a choice at founding. They went after construction management and design-build, not hard-bid tenders — "we knew we were gonna trade a lot in our relationships and sort of more private type of work." An eighteen-month-old firm can't beat a fifty-year scorecard in a price-only race. But it can win relationship-driven CM work on the strength of who its people are. The contract-model choice was a strategy and an honest read of where their edge actually sat.

That same risk discipline is what keeps Boyd cautious at the far end of the spectrum, too. He's watched P3s long enough to respect the downside: "When you lose money on a P3, you don't lose ten thousand or twenty thousand — those numbers are such a big multiple" (Boyd, EP 23). The research backs the wariness — a study of 28 Ontario P3 projects found they ran 16 percent more than conventional delivery, once private borrowing and transaction costs were counted. The P3 pitch is risk transfer, but you have to be able to price and carry the risk you're transferring. For most firms, most of the time, the answer Boyd lands on is a portfolio, not a religion: "A good healthy business would have a nice mix — CM delivery models, higher risk profiles with higher margin, DB." No single model wins. The right mix depends on your capacity, your team, and an honest accounting of what risk you can actually hold.

Four provinces, four markets — and a contractor who can only work one is stuck

Here is where the risk-manager's portfolio thinking runs straight into a hard regional fact: Atlantic Canada is not one construction market. Lorin Robar, who runs Atlantic for Pomerleau, gave the sharpest map anyone's put on the record: "Newfoundland is in a DB-P3 kind of market… New Brunswick is lump sum… Nova Scotia here is CM" (Robar, EP 14). A firm that can only operate in one delivery mode can't cover the region. The way you stay working across four provinces is the way Pomerleau does it — running design-build P3 long-term-care homes in Gander and Grand Falls-Windsor while sitting in a CM role on the Cape Breton Regional Hospital expansion as a joint venture with Lindsay Construction.

That hospital JV carries a principle worth marking: when a national firm enters a new geography for big institutional work, the local partner's sub-trade relationships and market knowledge aren't a nice-to-have — they're a risk-management instrument. The JV structure buys both the balance sheet and the local read. Robar is also one of the few who'll name the cost of just chasing this work: a P3 pursuit runs "a year-long process start to finish, easy" (Robar, EP 14). Estimating, design coordination, equity partners, financial advisors — all near-full commitment for twelve months before you know if you've won. Firms that run pursuit programs without pricing that in are subsidizing the process out of margin from other jobs.

Nova Scotia, notably, has been moving structurally off pure low-bid. The Service Nova Scotia Procurement Manual now codifies a median-pricing method — the bid closest to the determined median wins, not the lowest, deliberately to avoid awarding to outliers. And the province's Construction Contract Guidelines set quality-based scoring gates a bid must clear before its price is even opened. The structure means the government literally cannot award on price alone in applicable solicitations. That shift didn't come from nowhere — it came after the province's Auditor General reported $1.6 billion in over-budget appropriations in 2022-23 and $1.38 billion in 2023-24, spending the AG called "not accountable or transparent," including a Halifax street project that landed almost 200 percent over what was originally advised for what looked like a reduced scope. Those numbers are what eventually move procurement policy. They're also the bill for the orthodoxy.

The evaluation culture has to change before the contract can — here's what that looks like on paper

Picking a better delivery model is the easy part. The harder part — the part that actually traps owners — is that low-bid is a habit of evaluation, not just a contract form, and the habit outlives the form. You can put "construction management" on the cover of an RFP and still run the selection like a price auction underneath. The roundtable named exactly this failure: under the old CM solicitations, the panel "would score everything technically almost identically, so it always came down to fee" (Rudolph et al., EP 73). If every qualified bidder clears the technical bar with near-identical marks, the rubric is theatre and you've quietly rebuilt low-bid inside a collaborative wrapper.

So what does a genuinely shifted evaluation look like? Concretely, on the page: it stops scoring "experience" with a checkbox every serious firm passes, and starts scoring the things that actually de-risk the job — the named individuals assigned to your project and their availability, a written constructability read on your actual concept, the firm's open-book approach to subcontractor pricing, and how they propose to run the target-cost and contingency mechanics. Nova Scotia's two-step gate is the structural version of this: a bid that scores below the non-price threshold is disqualified regardless of price, so technical merit isn't a tiebreaker after the money — it's a floor the money never gets to override. As the roundtable put it, "now it's not just about the price on the evaluations — they've got all these other metrics" (Rudolph et al., EP 73). The weighting is the policy. An owner who wants collaborative outcomes has to write a rubric where fee is one input among several that actually move the score, not the silent tiebreaker every other line resolves into.

What collaborative delivery actually demands — from you, the owner, too

That rubric only pays off if the work behind it changes, and collaborative delivery asks more of everyone in the room, the owner included. Rudolph described EllisDon's position with no fuzz: "We're always sitting at the table during early stages… sitting down with the owner and the design team" (Rudolph, EP 48). That seat — in the room before the design is finished — is where the value is actually made. The estimators run historical comparables against early sketches, flag constructability problems before they harden into change orders, and stress-test the budget before anyone's spent real money on drawings. Rudolph's pitch for it is unembarrassed: "We'll save you far more than you'll ever spend on us — that's my sales pitch" (Rudolph et al., EP 73). One avoided design error on a fifty-million-dollar building covers the CM fee many times over.

The federal Atlantic Science Enterprise Centre in Moncton made that early involvement contractually real: the architecture contract and the CM contract were awarded so design and cost estimation developed in parallel, with the constructor checking the budget against the drawings as they came up rather than after they were done. The alternative — finish the design, then tender — hands the owner a number they can't adjust, backed by documents that are often incomplete. And the collaborative model only works if everyone's incentives point the same way. Under the CCDC 30 integrated-project-delivery structure, every party's profit goes into a shared pool that rides on the project hitting its targets — "it's a profit pool on top of the hard costs… everyone's going to win or everyone's going to lose" (Rudolph et al., EP 73). Clinch's version of the same idea is built on transparency as the price of entry — and on the owner sharing the load. "Do you want to take the risk?" he asks them. "Because I'm not doing it alone" (Clinch, EP 63). Collaborative delivery is not a way for the owner to off-load risk and look away. It's a way to manage it together, in the open.

The honest caveat: collaborative is not a silver bullet

I won't sell you the clean version, because the trade won't let me. Collaborative delivery has its own failure mode, and Atlantic Canada has a fresh, well-documented one. Halifax Water's Burnside Operations Centre was procured as an integrated project delivery job with a Bird-Chandos JV, and its budget jumped from $52 million in April 2023 to $89 million that November — a 71 percent leap — before being trimmed to about $87 million. The Nova Scotia Utility and Review Board's consultant called the budget "excessive" and the contingencies inflated, and flagged that the IPD contract language obligated the owner to pay the cost of work even when it blew past the target — which, in the consultant's words, "defeats the purpose of the IPD approach." That's the receipt, with the number and the criticism attached, and it deserves to stand without spin.

What Burnside proves is not that collaborative delivery doesn't work. It's that no contract form is self-enforcing. IPD's aligned incentives only hold if the target-cost discipline and the governance behind them are real. Drop the discipline and you can overrun inside the most collaborative contract in the book just as surely as inside a hard bid. The roundtable said it themselves, and it's the line to keep: "There's no silver bullet here in this industry" (Rudolph et al., EP 73). The point was never that CM or IPD or design-build is magic. The point is that low-bid lump-sum has a structural failure baked into its incentives, and the collaborative models at least give a disciplined owner the tools to manage cost instead of discovering it.

Three questions an owner should answer before choosing a model

So how do you actually decide? Skip the contract acronyms for a moment and answer three questions honestly, because they sort the problem faster than any procurement manual.

First: How complete is my design? If it's genuinely finished — every system resolved, every interface drawn — a stipulated price can work, because there's little left to turn into a change order. If it's not, a fixed price just sets the floor and hands the contractor the upside on every gap.

Second: What is the time value of knowing my budget early? CM and IPD buy you a cost trajectory you can steer at 33, 66, and 99 percent design, before anything's cast in concrete. If your financing, your board, or your tenants need budget certainty early and the ability to adjust, that early knowledge is worth real money — pay for it.

Third: Am I prepared to be at the table? Collaborative delivery is not a way to off-load risk and look away. As Clinch puts it, "you just can't muscle your way through it anymore" (Clinch, EP 63). If you want the upside of a transparent, open-book relationship, you have to staff it and show up. If you can't, a fixed price and an arm's-length contract may genuinely fit you better — and that's a legitimate answer, not a failure.

Which brings it back to where Boyd started, because the three questions are just the risk-manager's discipline applied to your own side of the table. The general contractor is a risk manager; so is the owner, whether they've accepted the job or not. The low-bid envelope feels like certainty precisely because it postpones the risk conversation instead of having it. Then the first change order arrives and the conversation happens anyway, at the owner's worst leverage, with a markup attached.

So put the cost-floor sentence back in front of every owner who still believes the sealed envelope is the answer, the way Rudolph hands it to a client across the table. In a lump-sum world, the bid is your cost minimum — it just goes up from there. The signed number is not what the building will cost. It's the least it could ever cost, on the most optimistic day, before the first RFI. Atlantic Canada's most experienced general contractors — rivals who bid against each other for a living — have figured that out, said it on the record together, and built their companies to get out of the trap. The owners who hear them will spend their certainty buying knowledge early. The ones who don't will buy it late, by the change order, at full markup.

Building something complex in Atlantic Canada and weighing how to procure it? Come on the show and put your playbook on the record.