Case Study 2 — Zero: Curtis Boone Wins Rivermont Elementary by $184,000

Kestrel Construction Group, Curtis Boone, the Rivermont Unified School District, and every other person and company here are illustrative composites.

Setup

Rivermont Elementary School #12 is the opposite of Northgate in almost every dimension that matters.

The schedule is the pressure. A school must open when school opens. There is no version of this job where the district accepts a January turnover, and everybody bidding knew it.

Curtis Boone is good at this. He is fast, likable, decisive, and he has a genuine gift for the field. He does not document, he manages by relationship, and he wins work by being the sharpest pencil in the room. He is not a villain. He is skilled and wrong about the model, which is a much more dangerous combination.

What happens

The bid

Low by $184,000 — eight-tenths of one percent. Here is how the number was built:

That $1,754,000 is markup, not profit. It carries home-office overhead — estimating, accounting, insurance administration, the people who never set foot on the job — before a dollar of it is margin. Kestrel's anticipated fee on this contract, the profit portion, is $784,000, 3.5% of the contract sum. Confusing the two is how a contractor talks itself into believing it has room it does not have.

Kestrel's internal standard for a hard-bid public job is 2.5% of cost of work plus general conditions: 2.5% × $20,210,000 = $505,000. Curtis carried none of it. In the bid room he described this as "sharpening the number." What it actually was: he gave up $505,000 of protection to beat the field by $184,000.

Watch what follows.

Event 1 — Day 34. Unsuitable subgrade in the bus loop.

The geotechnical report had four borings in the building footprint and none in the bus loop. Under 6 inches of topsoil the crew found 3,100 CY of undocumented fill — a demolished maintenance building nobody had recorded. Over-excavation, engineered backfill, and disposal: $128,000.

Curtis had a good claim. The contract carried a standard concealed-conditions clause and the condition was genuinely unforeseeable from the documents. He also had a notice requirement: written notice to the district's construction manager within 10 days of discovery.

What he sent was a text message on day 35: "Found junk fill in the bus loop, going to be a number, will get you something." Then he got busy, and the written notice went out on day 41.

The district's counsel took the position that a text message to the construction manager was not written notice under the contract and that the claim was untimely for everything before day 41. After three months, the district settled at $41,000 to close it.

Net hit: $87,000. Not because he was wrong. Because he could not prove he had said so on time. (Notice requirements vary by contract and by jurisdiction, and some courts and boards are more forgiving of actual notice than others — but you never want to be the party arguing about it.)

Event 2 — Day 96. Steel escalation.

Curtis had a structural steel quote of $1,940,000, held firm for 30 days at bid time. He did not buy it out. He was short-staffed — one project engineer for a $22.4 million job — and buyout is the kind of work that waits when the field is loud.

He issued the subcontract on day 96. Mill pricing had moved 9.4% in the interval: $1,940,000 × 1.094 = $2,122,000.

Hit: $182,000. There is nobody to claim against. In a lump-sum contract the contractor owns escalation. That is not unfair; it is the deal, and it is why an escalation risk row belongs on the register with "buy out within 60 days of NTP" as its response — exactly what Tomás did on Northgate.

Event 3 — Day 171. The kitchen hood.

The commercial kitchen equipment schedule showed a Type I hood at 4,200 CFM. The mechanical drawings showed a makeup-air unit sized for 3,000 CFM. Curtis had bid the mechanical drawing.

His project engineer wrote an RFI. The architect answered: "Provide per equipment schedule."

That answer is a change. The documents conflicted, there is an order of precedence that decides which governs, and the difference between the two is money (Chapter 7). The contract required the contractor to submit a proposed change order within 21 days of any event giving rise to one.

Nobody logged it as a change. The RFI got filed as answered, the mechanical sub built the bigger unit, and the cost appeared eleven weeks later as an unexplained overrun on cost code 23-00.

Hit: $67,000, most of which was probably recoverable on day 172 and none of which was recoverable on day 255.

Event 4 — Day 244. The drywall subcontractor.

Curtis had a drywall and acoustical bid of $1,410,000 at bid time. After award he shopped it, and awarded at $1,238,000 to a smaller, unbonded outfit that had never run a job this size.

At 55% complete the sub stopped work owing three second-tier suppliers. Earned to date: $680,900. Remaining subcontract balance: $557,000. Cost to complete with a replacement subcontractor, mobilized on emergency terms, in the middle of an occupied schedule: $712,000.

The $172,000 Curtis "saved" by shopping the package cost him $255,300, and it cost the second-tier suppliers considerably more than that.

The running total

And then the part that got people's attention

Rivermont Elementary showed a profit every month through month eight. Curtis forecast cost-to-date against billings, not cost-to-complete against remaining scope. As long as he was billing ahead of cost — and on a front-loaded schedule of values he was — the report looked fine.

In month ten, when the drywall replacement and the accumulated unlogged items were finally posted, the forecast moved $640,000 in a single reporting period.

Owen Baptiste, Kestrel's CFO, does not raise his voice either. What he said was: "A job doesn't lose six hundred and forty thousand dollars in a month. It loses it over ten months and tells you in one."

At the surety's annual renewal, the underwriter asked about Rivermont three separate times and asked to see Kestrel's go/no-go process in writing. Capacity held at $150 million aggregate and $60 million single project. Nadia Haddad's comment afterward was short: "We get one more of those and we don't."

Analysis

Start with the uncomfortable part, because dodging it teaches nothing. If Curtis had carried the standard $505,000 contingency, his bid would have been $21,151,000 × 1.085 ≈ $22,949,000 — third of four. Palmetto Builders would have built the school.

So the lesson cannot be "always carry the standard percentage." On a hard-bid public job the standard percentage is often the difference between winning and not winning, and a contractor who never wins is not a safer contractor. It is a former contractor.

The honest lessons are these four.

1. Contingency is not the only lever, and it is the last one you should reach for. Every risk that hurt Curtis had a cheaper response than money. Event 1 was a transfer — the district owned it, and all Curtis had to do was give notice correctly, which costs nothing. Event 2 was a mitigation — buy the steel out inside 60 days, which costs nothing but attention. Event 3 was a mitigation — log the change when the RFI came back, which costs one email. Event 4 was a transfer — bond the package or prequalify the sub, at roughly 1–2% of $1,238,000, or about $12,000 to $25,000. Four events, $591,300 of damage, and the total cost of the responses that would have blunted them is under $30,000 plus discipline. Contingency is what you carry for the risks you cannot respond to. It is not a substitute for responding.

2. If you choose to bid thin, you must then run the job as though the contingency is zero — because it is. That means notice discipline that is faster than everyone else's, buyout inside sixty days, a change logged the day it appears, and a cost-to-complete forecast every month. Curtis's actual mistake was not the zero. It was the zero plus running the job with the habits of a man who had a cushion. He behaved exactly as he would have on a job carrying $505,000, and there was nothing underneath him.

3. The go/no-go decision is where this was really made. A $22.4 million hard bid, a 610-day schedule with an immovable opening date, a geotechnical report with no coverage outside the footprint, and a district that had already compressed the timeline — that is a risk profile that does not fit a standard 2.5% contingency. The right answer was either a bigger number that loses, a qualified bid, or a no-bid. That decision belonged in the pursuit meeting, not in the bid room at 10:40 a.m. on bid day (Chapter 15).

4. And the enterprise view. Kestrel survived Rivermont comfortably — $811,300 is painful, not fatal, against roughly $10,000,000 of annual net at a 2.5% margin. What it actually cost was harder to see: an underwriter's attention at renewal, a project manager's confidence, and the credibility of the monthly cost report across the whole company. Somebody in every project review for the next two years said "yes, but is this a Rivermont?" That is not a small tax.

Discussion questions

1. Curtis's bid was low by 0.8%. What would you have needed to know at bid time to argue for a bid $549,000 higher, and who at Kestrel would you have had to convince?
2. Event 1 was a valid claim destroyed by a notice failure. Write the written notice you would have sent on day 35. What must it contain to preserve the claim, and what should it deliberately not contain?
3. Bid shopping the drywall package saved $172,000 and cost $255,300. Is your objection to bid shopping ethical, economic, or both? Where exactly is the line between shopping a bid and legitimately negotiating scope after award?
4. The forecast moved $640,000 in one month. Describe the specific reporting practice that made that possible, and the specific practice that would have surfaced it in month four instead.
5. Compare Curtis's incentives to Ray's. Northgate's 75/25 savings split rewards Ray for underspending contingency; Rivermont's lump sum gives Curtis 100% of savings and 100% of overrun. Which incentive structure produces better behavior, and under what conditions does each one fail?

Your turn

Rebuild Curtis's bid with a defensible risk position. Using the four events as your starting register, add at least four more rows you would have identified at bid time from the facts given. For each row, assign a probability and a cost impact and compute an expected value.

Then decide three things and write one sentence justifying each: (a) the contingency you would carry, bottom-up; (b) which two rows you would transfer rather than fund, and what the transfer costs; and (c) the bid number that results. If your number is above $22,584,000, say plainly whether you would have submitted it anyway — and why that is a defensible professional answer.