Case Study 11.1 — Nine Days in a Conference Room: The Northgate $4.8 Million Gap

Kestrel Construction Group, Meridian Health System, and the Northgate Outpatient Pavilion are illustrative composites assembled from real projects. The numbers are internally consistent and the mechanics are real; the people are not.

Setup

The project. Northgate Outpatient Pavilion — 4 stories, 132,000 gross square feet, structural steel frame, outpatient clinics, an imaging suite, ambulatory surgery, a café and lobby, on a sloping 6.2-acre urban-edge site with an active Meridian clinic immediately across the north property line. Delivery: CM at Risk with a Guaranteed Maximum Price (GMP).

The constraint. Meridian's board approved a construction budget of $47,500,000 in February. The number came out of a pro forma that assumed the pavilion opens before Meridian's interim clinic lease expires on October 1, Year 2. That date is not negotiable, and neither is the board's number without going back to the board.

The trigger. Sixteen months before Notice to Proceed, Kestrel's chief estimator Tomás Reyes delivered a Class 4 elemental estimate of the schematic design set: $52,300,591, or $396.22/SF.

The gap: $4,800,591.

The room. Ray Alvarez (Kestrel PM), Tomás Reyes (Kestrel chief estimator), Margo Deacon (general superintendent), Grace Lindqvist (VDC manager), Priyanka "Pri" Sethi (Meridian owner's rep), Dale Whitcomb (project architect, Halvorsen + Pike), Ruth Caldwell (Caldwell Structural), Sofia Marchetti (Cardinal Mechanical), Devlin Achebe (Halcyon Electric). Two half-day workshops, nine calendar days apart.

What happens

Day 1, first ten minutes

Tomás put up one number before he put up anything else.

Tomás: "The burden multiplier on this estimate is 1.2005. Design contingency, construction contingency, escalation, and fee all ride on the cost of work. That means every dollar of work we remove takes a dollar twenty off the bottom line. So we are not looking for four point eight million dollars. We are looking for four million dollars of actual work — $3,998,920, if you want it exact — and the markups will hand us the rest. Every number in the log I'm about to show you is already burdened. Don't do that math twice."

Pri: "And if we don't find four million?"

Ray: "Then we come back to you in nine days with what we found, and you decide what to do about the rest. I'm not going to close this gap with a number I can't stand behind."

That exchange is the whole case study. Everything after it is arithmetic.

The workshop method

Grace put the coordinated schematic model on the screen. For each major element, the group did function analysis before anybody proposed anything — the element's job stated as a verb and a noun, deliberately abstract, so that alternatives could exist.

Curtain wall → exclude weather, admit daylight, resist wind Perimeter moment frames → resist lateral load Eight rooftop air-handling units → condition air, control zones, provide redundancy Depressed slab at imaging → support and isolate equipment at the required elevation

The moment somebody said "the curtain wall," Tomás would stop them: "That's a product, not a function. What does it do?" It sounds pedantic. It generated three of the four largest savings in the log.

The log

Eighteen options. Thirteen accepted, five rejected. All savings burdened.

Result: revised schematic estimate $49,273,391 ($373.28/SF). Still $1,773,391 over budget.

The three rejections that mattered

VE-06 — floor-to-floor reduction, $384,200. Rejected in four minutes.

Devlin: "Eight inches across four floors. Less skin, less steel, less stair. That's the biggest single number on the board."

Sofia: "Grace, put the third-floor clinic wing up. Section at grid F." (pause) "There's your 26-inch main with insulation, sprinkler main under it, med gas, cable tray, and the exam-room ceiling is fixed by Meridian's own standards. Take eight inches and I'm reshaping and rerouting that main around structure at every bay. You'll save $384,000 on the shell and spend more than that on ductwork, and then you'll spend it again on RFIs in month fourteen."

Ray: "Log it as rejected with Sofia's reason attached. Nobody re-proposes it in six months."

VE-12 — cheaper corridor flooring, $228,100. Rejected by Meridian.

Pri: "What's the wear layer on the substitute?"

Tomás: "Ten years, with seams. The specified product is heat-welded, twenty-year."

Pri: "Our infection-prevention committee wrote the seam requirement. And I'd be re-flooring occupied clinic corridors in year eleven. That's not $228,000 of savings, that's $228,000 borrowed at a rate I can't calculate."

Analysis: Function did not stay constant. The life-cycle column was the entire argument, and the owner made it herself — which is exactly what happens when the owner is in the room instead of receiving a log by email.

VE-14 — cut the contingencies, $2,038,900. Rejected after the hardest conversation of the nine days.

This one arrived from Meridian's finance office, not from the workshop. It closes 42% of the gap with a pen stroke and changes nothing about the building.

Ray: "That doesn't change the cost of the project. It changes how much of it we've admitted to. The design contingency is our estimate of what the drawings add between now and permit. I think that's about $2.6 million on this job. If we don't fund it, we'll be in this room again in nine months with fewer options and a worse mood."

Pri: "And if you're wrong and the drawings only add a million?"

Ray: "Then the contingency comes back to you and you're a million dollars happier. That's the deal. It comes back if the risks don't happen. It doesn't come back if we never funded it."

Meridian rejected VE-14. Nine months later, design development added $2,755,010 to the cost of work — 95% of what Tomás had carried.

Eleven months later: the one that came back

VE-04 replaced a full-bay depressed slab and isolation pit at the imaging suite with a localized thickened slab and a steel-framed isolation platform, sized to the vendor's generic template. Burdened saving: $267,700. Accepted with a written risk note, because constructability finding CR-03 had already flagged that Meridian's imaging vendor had not been selected.

After the GMP was set, the vendor selected a different MRI unit. It required a deeper depressed slab, added structural framing, additional RF shielding, and a larger electrical feed. Pri gave a verbal go-ahead on a Thursday; Kestrel's assistant superintendent, under schedule pressure, let the concrete crew build it Monday — no written directive, no agreed price, no time-impact analysis, no time-and-material tickets for four days.

That is CO #14:

Roughly $71,300 of the $186,400 was demolishing and rebuilding a slab Kestrel had just poured — work the original full-bay depressed slab would have absorbed without demolition.

Net on VE-04: $267,700 saved, $71,300 returned, $196,400 net. Still a win. And the log never got updated to say so.

Analysis

What worked.

Function analysis before options. VE-01, VE-02, VE-08, and VE-18 all came from asking what an element does rather than what it is. Three of the four largest accepted savings came out of a question, not a catalog.

The owner in the room for all of it. Pri rejected VE-12 herself, on grounds nobody at Kestrel could have argued as well. A VE log the owner receives by email is a document you argue about for the rest of the job.

Burdened numbers, computed once. Removing the 1.2005 conversion from everyone else's head removed an entire category of meeting confusion and stopped people from double-counting savings.

Rejections logged with reasons. Nobody re-proposed the floor-to-floor reduction. Sofia's reason was in writing with her name on it.

Modified acceptances instead of binary ones. VE-03 and VE-10 both survived in modified form because someone asked "how much of this can we keep?" instead of "yes or no?" Together, the modifications gave back $79,600 and bought two real operational protections.

What failed, and why.

Kestrel was inconsistent about open-decision risk on the same day. VE-13 (spare conduit) was rejected because the imaging vendor was unselected. VE-04 was accepted with the same open risk. The difference in reasoning was never articulated — VE-13's cost of being wrong was bounded and small, VE-04's was not, and nobody said that out loud. A VE risk register, revisited at each design milestone, would have forced the question. It also would have caught VE-04's risk when the vendor was finally selected, which was before the concrete was poured.

The log became a historical document. Once signed, nobody revisited it. VE-04's $71,300 giveback never appeared anywhere, so Kestrel's estimating database still records a $267,700 saving. The next project will inherit an optimistic number.

The honest accounting. Nine days of VE closed $3,027,200 — 63% of the gap. The other 37% closed because the design contingency retired to zero when the documents completed and the escalation allowance fell from $2,801,338 to $575,200 once 84% of the work was bought at fixed prices. Between SD and GMP the cost of the work went up $233,310. The building got more expensive and the price came down, because most of what closed the gap was never in the building — it was in the not-knowing.

But without those thirteen items, the GMP cost of work would have been $42,521,700 and the GMP would have landed at $50,122,568 — $2.6 million over, with no design contingency left to retire and no time to find anything. The VE was the difference between a job that happened and a fourth floor that didn't.

Discussion questions

1. Ray told Pri he would not close the gap with a number he could not stand behind, knowing that left Kestrel $1,773,391 short at the end of the workshop. What would have happened over the following nine months if he had instead accepted VE-14 and declared the gap closed?

2. VE-13 was rejected and VE-04 accepted on the same afternoon, both carrying the risk of an unselected imaging vendor. Articulate the principle that distinguishes them. Then apply your principle to a third case: a $340,000 saving that depends on an owner decision due in four weeks.

3. VE-16 (the café) is labeled a program decision rather than value engineering, even though it was accepted and counted in the $3,027,200. Why does the label matter if the dollars are identical?

4. Meridian's finance office proposed VE-14 in good faith. Construct the strongest possible argument for cutting the contingencies — the version a competent CFO would actually make — and then say precisely where it fails.

5. Kestrel's estimating database still shows VE-04 as a $267,700 saving. Design the specific control that would have caught the $71,300 giveback, and say who owns it and when it runs.

Your turn

Take three of the five rejected items (VE-06, VE-12, VE-13, VE-14, VE-15) and write, for each, the two-sentence rejection note that goes in the log — the version a stranger reads eight months later when they are tempted to re-propose it. Each note must name the function that would have changed or the risk that would have transferred, and it must attach a person's name.

Then answer the question Ray had to answer: at the end of the workshop Kestrel is $1,773,391 over Meridian's budget with no more VE to find. Write the one-page memo to Nadia Haddad, Kestrel's VP of Operations, recommending whether Kestrel should commit to a GMP of $47,500,000 eleven months from now — and say what has to be true for that commitment to be defensible.