Case Study 3.1 — The Same Building, Twice

Meridian Health System's Southbank Pavilion (design-bid-build) and Northgate Pavilion (CM at risk), compared honestly

Everyone in this case study is a Tier-3 illustrative composite — the health system, the contractors, the architect, and the people. The numbers are constructed to teach a mechanism, not to report a real project.

Setup

Meridian Health System has built two outpatient pavilions in the past eight years. They are not identical, but they are close enough that comparing them teaches something almost nobody gets to see: the same owner, the same architect, the same market, the same building type — two different delivery methods.

Pri Sethi was the owner's representative on both. She is the reason this comparison exists: after Southbank closed out, she wrote a fourteen-page internal memo that nobody outside Meridian's facilities committee has ever read, and she used it four years later to make the Northgate recommendation.

Southbank was, by every conventional public measure, a well-run procurement. Six bidders. A complete 100% construction document set. A low bid of $38,900,000 against Meridian's cost consultant estimate of $39,400,000 — under budget on bid day, which the board loved.

The spread told a different story. Low to high was 9.8%.

What Happened at Southbank

Verity Builders was a competent contractor. This is not a story about a bad contractor. It is a story about what a delivery method does to a competent contractor.

Month 3. Verity's mechanical subcontractor discovered that the third-floor mechanical room, as drawn, could not accommodate the specified air handlers with the clearances the manufacturer required for coil pull. The room needed to grow by roughly four feet in one dimension, which meant moving a corridor wall, which meant relocating two exam rooms.

RFI submitted. Halvorsen + Pike had earned its design fee in full eleven months earlier. The response took 24 days and required a redesign the architect performed as an additional service billed to Meridian.

Change order: $214,000. Time impact claimed: 18 calendar days. Granted: 9.

Month 7. The structural drawings showed a slab depression at the lab that the mechanical drawings did not reflect; the lab's floor drains and process waste piping had been laid out against the non-depressed condition. Neither drawing was wrong on its own. Together they were incompatible.

Verity built to the structural drawing, as it was contractually required to do under the order of precedence. The piping had to be redone.

Change order: $88,500. Verity argued the documents were defective. Halvorsen + Pike argued Verity should have caught the conflict in coordination. Meridian paid most of it and pursued nothing, because pursuing the architect meant a professional negligence claim against a firm it intended to hire again.

Months 9 through 16. Forty-one more change orders. None individually alarming. Two curtain-wall detail revisions, a fire-alarm device count that did not match the specification, a grade discrepancy at the loading dock, an elevator hoistway dimension that would not accept the selected cab, and a long tail of small clarifications that each cost between $4,000 and $30,000.

The close-out numbers:

Beyond the contract, Meridian paid an additional $318,000 in Halvorsen + Pike additional-services fees to process the redesigns, and its facilities office spent an amount nobody tracked adjudicating change orders.

Schedule: substantial completion came 71 calendar days after the contract date. Meridian granted 44 calendar days of time extension and, in a global settlement that also resolved Verity's remaining claims, waived liquidated damages on the balance. The clinic opening slipped a full quarter.

What Pri Wrote

The memo's central paragraph, which she reread before the Northgate board presentation:

"We did not overpay Verity. Verity earned a normal margin on a job it bid honestly. We paid $2,340,000 for errors in documents we furnished and warranted, and we paid it at change-order pricing — after the work was in place, with no competitive pressure, at a moment when our only alternative was to stop the job. Every one of those errors was findable in a set of drawings. None of them was findable by anybody who was not going to build the building, and nobody who was going to build the building saw the drawings until they were finished."

That sentence is the entire argument for CM at risk, and it is worth noticing that it is an argument about information timing, not about contractor quality.

Northgate, So Far

Kestrel was engaged at the end of schematic design. Over fourteen months of preconstruction it produced four estimates at successive milestones, ran three constructability reviews on the imaging suite, and issued 61 constructability comments on the drawing sets, of which Halvorsen + Pike incorporated 44.

The GMP was set at roughly 85% construction documents:

Meridian's independent cost consultant priced the same set at $46,900,000. The reconciliation meeting ran six hours. Pri challenged the qualifications list in seven places and Kestrel conceded on three.

Normalized comparison. Southbank finished at $42,180,000 across 118,000 SF — $357.46/SF — four years earlier. Escalating at roughly 3.5% per year to compare like with like gives a Northgate-era equivalent of about $410/SF. Northgate's GMP is $360/SF.

Treat that gap with suspicion. Northgate is a different building — an extra floor, an imaging suite, an ambulatory surgery component, a harder site with an occupied clinic on the property line — and those differences push cost up, not down. The comparison is directionally meaningful and it is not a controlled experiment. Anybody who presents it as one is selling something.

Analysis: Where CM at Risk Did Not Help

This is the part that gets left out of the marketing, and it is the part Pri put in her recommendation.

1. Owner scope changes cost the same under both. Southbank's $940,000 of owner-driven changes had nothing to do with delivery method — Meridian changed its mind, and that costs money under any contract. Northgate is proving the same point. When Meridian's imaging vendor selected a different MRI unit after the GMP was set, requiring a deeper depressed slab, added structural framing, additional RF shielding, and a larger electrical feed, CM at risk provided no protection whatsoever. That is change order #14, and it will be expensive.

2. The fee is genuinely higher. Verity's competitive overhead and profit on Southbank was roughly 2.2%. Kestrel's disclosed fee is 4.0%. On Northgate's subtotal that difference is on the order of $800,000. It is real money and it buys real services, but an owner that pretends the fee premium does not exist is not being honest with its board.

3. Meridian's own cost went up. Pri spent fourteen months in preconstruction meetings. Meridian paid $148,000 for the independent cost estimate and reconciliation. The facilities committee had to learn to read a cost model. Under Southbank's design-bid-build, none of that existed — the owner's involvement started at award. CM at risk transfers work to the owner as well as risk away from it.

4. CM at risk did not prevent Kestrel's own mistakes. The anchor-bolt and embed submittal sat in Kestrel's office for 11 days before it went to Caldwell Structural, which then took its full 14-day contractual review. Ironbridge Steel missed its mill rolling slot, the next opening was five weeks out, and steel erection — on the critical path — slipped 23 calendar days, from August 4 to August 27 of Year 1. At Northgate's combined exposure of $10,650 per calendar day ($5,150 extended general conditions plus $5,500 liquidated damages), doing nothing would have cost $244,950. No delivery method protects an owner from a contractor's submittal log.

5. A subtle one: DBB produced better change visibility. Under Southbank, every deviation became a change order with a paper trail, a price, and a signature. Under Northgate's CM at risk, items that fall inside the contingency get absorbed and reported in a monthly summary. That is more efficient and it is also less visible. Meridian sees a contingency drawdown number; it does not automatically see forty individual stories. An owner that does not read the contingency log carefully will know less about what is going wrong on a CMAR job than on a hard-bid one.

Analysis: The Mechanism

Strip away the numbers and one mechanism explains most of the difference.

Under design-bid-build, the builder's knowledge arrives after the documents are fixed. Every piece of buildability information the contractor has is worth nothing until it can be converted into an RFI, and by then converting it costs money and time. The contractor's knowledge is real; the method has no channel for it.

Under CM at risk, the same knowledge arrives while the documents are still moving. Kestrel's 61 constructability comments cost Meridian nothing to receive and cost Halvorsen + Pike ordinary design effort to incorporate. Forty-four of them became drawing revisions. Under Southbank's method, an unknown fraction of those same 44 items would have become change orders at change-order pricing, in the field, with no competitive pressure and a crew standing by.

That is the whole thing. The method is a channel for information, and information is cheapest before it is needed. It is the same idea as "the project is built twice, and the first build determines the second," expressed in dollars.

Discussion Questions

1. Pri's memo says "we did not overpay Verity." Do you agree? Construct the strongest argument that Meridian did overpay — and then the strongest rebuttal.

2. Meridian chose not to pursue Halvorsen + Pike for the $2,340,000 of design errors, because it intended to hire the firm again. Is that a defensible use of a non-profit health system's money? What would have to be true for the answer to change?

3. The normalized comparison ($410/SF equivalent versus $360/SF) is presented with explicit warnings about confounders. Name three specific confounders and, for each, say which direction it biases the comparison.

4. Point 5 in the analysis argues that design-bid-build gives an owner better visibility into what is going wrong. Is that a real advantage or an accounting artifact? What would you put in a CM-at-risk contract to capture the visibility without giving up the efficiency?

5. Kestrel's 11-day internal delay on the anchor-bolt submittal caused a 23-day critical-path slip. Under which delivery method — Southbank's or Northgate's — would that mistake have been more expensive to Kestrel, and why?

Your Turn

You are Pri Sethi, three weeks before the board presentation. Write the one slide that presents this comparison honestly to a board that includes Harriet Voss, who is going to ask why the recommended method costs more on paper.

Constraints: one slide, no more than six lines of text plus at most one small table. It must contain (a) one number that favors CM at risk, (b) one number that favors design-bid-build, and (c) the sentence that reconciles them. You are not allowed to hide the fee premium.

Then write the two sentences you would say out loud after Harriet asks her question.