Case Study 2 — Low by $1.9 Million: Where the Money Went on Rivermont Elementary School #12
All people, companies, and projects in this book are illustrative composites.
Setup
Curtis Boone is a Kestrel project manager. He is good at this job in most of the ways that show — likable, quick, trusted by superintendents, unflappable in front of an owner. He wins work. He has never lost a job in front of a bid opening in his life. He documents almost nothing, buys his subcontracts on price after the fact, runs the leanest staff in the company, and manages by force of personality.
He is not a villain. He is a project manager operating a different model of the business, and this case study is the price of that model, in dollars, quarter by quarter.
Rivermont Elementary School #12. $22,400,000. Design-bid-build, lump sum. Public owner — Rivermont Public Schools. Prevailing wage. 100% payment and performance bonds. Twenty months (610 calendar days). Liquidated damages $2,500 per calendar day. Kestrel's field extended general-conditions rate on this job: $2,100 per calendar day.
Boone carried 4.0% profit: $896,000 on a planned cost of $21,504,000.
What happened
Bid day
Six bidders. Here is the tab, read aloud in a school district conference room on a Thursday afternoon.
| Bidder | Base bid | vs. Kestrel | % |
|---|---|---|---|
| Kestrel Construction Group | $22,400,000 | — | — |
| Ambrose Builders | $24,300,000 | +$1,900,000 | +8.5% | |
| Torvald Construction | $24,610,000 | +$2,210,000 | +9.9% | |
| Pinehurst General | $24,875,000 | +$2,475,000 | +11.0% | |
| Granite Bay Constructors | $25,140,000 | +$2,740,000 | +12.2% | |
| Delacroix & Sons | $25,930,000 | +$3,530,000 | +15.8% | |
| Architect's estimate | $24,750,000* | *+$2,350,000 | +10.5% |
Read the shape. Five bidders cluster inside a 6.7% band — from $24,300,000 to $25,930,000 — and the architect's independent estimate lands in the middle of that cluster. One bidder sits 8.5% below every other bidder and 9.5% below the architect.
Five estimating departments read the same documents and arrived at roughly the same conclusion. One did not. In a market that competitive, the probability that Kestrel simply had a better cost structure than five firms simultaneously is small; the probability that something is missing from Kestrel's number is large.
Boone bought lunch for the estimating team.
Quarter by quarter
This is the record Kestrel's project accountant, Lorena Vasquez, assembled at the post-mortem. Approved change orders totaled $310,000, bringing final contract value to $22,710,000; the cost of that change-order work was roughly equal to its value, so it neither helped nor hurt the margin.
| Quarter | % complete | Cost to date | Forecast final cost | Forecast result | Margin |
|---|---|---|---|---|---|
| Q1 | 8% | $1,810,000 | $21,814,000 | +$896,000 | 3.9% | |
| Q2 | 22% | $4,930,000 | $22,180,000 | +$530,000 | 2.3% | |
| Q3 | 41% | $9,410,000 | $22,760,000 | −$50,000 | −0.2% | |
| Q4 | 60% | $14,240,000 | $23,540,000 | −$830,000 | −3.7% | |
| Q5 | 80% | $19,320,000 | $24,390,000 | −$1,680,000 | −7.4% | |
| Final | 100% | $24,980,000 | $24,980,000 | −$2,270,000 | −10.0% |
Two features of that table matter more than the final number.
The forecast never jumped. It crept — $366,000, then $580,000, then $780,000, then $850,000, then $590,000. There was no month in which anyone could point at a catastrophe. That is what a job going bad actually looks like, and it is why the cost-to-complete forecast, not the cost-to-date, is the instrument that tells you anything (Chapter 28).
Kestrel crossed zero in Q3, at 41% complete. Everything after that point was spending the company's money to finish somebody else's building.
Where the $3,166,000 went
Planned margin was $896,000. Final result was −$2,270,000. The gap is $3,166,000 of adverse cost. Most of it has a name — and the part that does not is its own finding.
| Source of loss | Amount |
|---|---|
| Masonry scope gap — bond beams, lintels, and grout carried by neither the mason nor the concrete sub | $318,000 |
| Mechanical rework — a bid-shopped low-tier sub; rework, re-inspection, punch, and re-testing | $402,000 |
| Roofing subcontractor default — replacement contractor hired at market, mid-job | $286,000 |
| Site utilities differing condition — notice given 31 days late; claim denied on the notice provision | $394,000 |
| Acceleration and premium time to protect the August school-opening date | $437,000 |
| Extended general conditions — 71 CD × $2,100/CD | $149,100 | |
| Liquidated damages assessed — 38 CD × $2,500/CD | $95,000 | |
| Unrecovered directed-change cost — no contemporaneous records | $571,700 |
| Bid errors — gypsum quantity short, temporary heat omitted | $375,000 |
| Untracked balance — productivity loss, rework, and punch labor never coded to a cause | $858,200 |
| Total adverse cost | $3,166,000 |
Before you move on, look at the second-largest line. $858,200 — 27 percent of the entire loss — has no cause attached to it. Nobody coded it, because productivity loss and rework do not arrive as invoices; they arrive as hours that went somewhere. A job that cannot say where a quarter of its losses came from cannot learn anything from them, and Kestrel bid the next school with the same historical data.
Now put that beside the bid tab. Kestrel was $1,900,000 below the next bidder and lost $2,270,000. The loss was $370,000 larger than the gap.
That relationship is the point of this case study. The $1.9 million gap was not the mistake. It was the symptom — the visible, measurable, free-of-charge early warning that Kestrel's number was missing something. Two of the nine loss lines ($375,000 of bid errors and $318,000 of masonry scope gap, $693,000 together) were literally in the bid on the day it was opened, and could have been found in an afternoon of scope review.
The remaining $2,473,000 came from how the job was run. But it came from how the job was run on a budget that had already been set too low — which is exactly how a bid error becomes a management crisis. When there is no room in the number, every recovery decision gets made under pressure: you buy the cheapest mechanical sub, you push the crews, you skip the T&M tickets because there is no time, and you accelerate at premium rates to protect a date you can no longer protect any other way.
The claim
Boone submitted a claim for $1,840,000 — the differing site condition, the directed changes, and delay damages. He recovered $265,000.
Not because he was wrong. Because he could not prove it. The utility claim died on a 31-day-late notice. The directed-change claim died because there were no daily time-and-material tickets signed by anyone from the district. Entitlement, causation, and damages are three separate proofs and you must win all three (Chapter 33); Boone could argue entitlement all day and never got past damages.
Analysis: the three go/no-go signals in Kestrel's own files
Run Rivermont Elementary #12 backward through the model in §15.2 and it scores 268 out of 500 — 53.6%. Decline, with a knockout.
Signal one: the designer and the documents, scored 2. The bid set was at roughly 92% construction documents, with eleven items marked to be resolved by addendum. Three addenda issued in the final ninety-six hours before bids, including one that changed the roof assembly. Kestrel's own preconstruction file scored that architect a 2 out of 5 on document quality from two prior projects. Nobody opened the file.
Signal two: staff availability, scored 1 — a knockout. The go/no-go form listed the superintendent as "TBD." Boone said he would find somebody. The job ran with a superintendent hired nine days before notice to proceed who had never built a school. The masonry scope gap, the mechanical rework, and the missing T&M tickets all trace to a field organization that never had the bandwidth to catch anything.
Signal three: the bid spread itself, examined in zero minutes. Five bidders inside 6.7%, Kestrel 8.5% below all of them, and no post-bid scope review before signing. A four-hour division-by-division review against the second bidder's likely scope would very probably have surfaced the masonry gap and the temporary-heat omission — $693,000 of it — while withdrawal was still legally available.
A fourth, for completeness: the client file, scored 2. Rivermont Public Schools had averaged 24-day RFI turnaround on two prior Kestrel projects and had denied every time-extension request on both. That is a client who will not give you time back, on a job with an immovable August opening. It was in the file. Nobody pulled it.
Every one of these was known, free, and available before Kestrel spent an estimating hour.
Discussion questions
- Kestrel crossed from positive to negative margin in Q3, at 41% complete. What specific report, produced monthly, would have made that crossing visible in Q2 — and what could a project manager actually have done in Q2 that was no longer possible in Q4?
- Two loss lines totaling $693,000 existed in the bid on the day it was opened. Given that bid withdrawal generally requires a clerical rather than a judgment error, could Kestrel have withdrawn? What would you have needed to demonstrate, and to whom, and by when?
- Boone shopped the mechanical package and saved about $190,000 at buyout. It cost $402,000 in rework and re-inspection. Build the argument against bid shopping that you would actually use with a project manager who does not care about ethics — one made entirely of money.
- The claim recovered $265,000 of $1,840,000. Pick one of the three failed elements — the 31-day-late notice, the missing T&M tickets, or the delay damages — and describe the specific, low-cost habit that would have preserved it.
- Boone is described as skilled and likable, not incompetent. What does that tell you about how loss jobs actually happen, and about what a go/no-go model is really protecting a company from?
Your turn
Score Rivermont Elementary School #12 on the eleven-criterion model in §15.2 yourself, using only the facts available before bid day — the 92% set, the designer's history, the "TBD" superintendent, six bidders, an immovable August opening, $2,400/CD liquidated damages, and the client's RFI record. Do not use anything you learned from the outcome.
Then write the one-page no-bid memo Boone should have written: your score, the band, the two knockouts or near-knockouts, and three sentences to the VP explaining what Kestrel gains by declining. Finally — and this is the harder half — write the two-sentence version of the opposite memo, arguing this is a legitimate strategic pursuit, with the capped budget and the named sponsor. Decide which memo you would actually send, and write one line saying why.