Case Study 31-2 — Sixty-One Change Orders: Curtis Boone's Business Model, Priced Out

Kestrel Construction Group, Curtis Boone, Rivermont Elementary School #12, the Rivermont Unified School District, and every person and firm in this case study are illustrative composites. Change-order markup caps, notice periods, prequalification rules, and the recoverability of impact costs vary by contract and by jurisdiction and change over time. Nothing here is legal advice.

Setup

The project. Rivermont Elementary School #12 — $22,400,000, hard-bid design-bid-build lump sum for the Rivermont Unified School District. Public owner, prevailing wage, certified payroll, 100% payment and performance bonds, 610 calendar days, 78,000 square feet, two-story classroom wing, gymnasium, commercial kitchen. Architect of record: Sandoval Beck Architects; project architect Perry Sandoval.

How it was bought. Curtis's team built the job honestly at $22,865,000. In a forty-minute final review with the executive who normally approves the number on an airplane, Curtis shaved $465,000 across four lines — an unleveled masonry bid, an assistant superintendent, the winter-conditions allowance, and seven-tenths of a point of contingency — and bid $22,400,000. He won by $312,000 over Bowen & Sons. The full autopsy of that meeting is in Chapter 13.

Two consequences of that meeting matter here, and only here:

• The masonry number Curtis carried was Fenner Masonry's, which excluded cold-weather protection, cast stone sills and copings, and gymnasium scaffolding. Every one of those exclusions was a change order waiting to be written.
• The general-conditions line was shaved from $1,612,000 to $1,464,000 — from a real burn of $2,642.62/CD to a bid rate of $2,400.00/CD — and the contract's extended-general-conditions rate for approved time extensions was set from the bid. Curtis shaved his own recovery rate before he wrote his first change order.

The pricing schedule. The district's Division 01, typical of public school work in that market:

Compare that to Northgate's 15%/5%. Public agencies frequently cap change markups lower than private owners, and they frequently demand more backup for them — the margin is thinner and the paperwork is heavier at the same time. That combination is the single most under-appreciated fact about running changes on public work, and it is jurisdiction-specific. Read your Division 01 before you bid, not after you win.

The cast. Curtis Boone, Kestrel project manager. Terrence Bell, field engineer, twenty-three, second job. Delphine Aguirre, the district's Director of Facilities and Construction — experienced, careful, and starting from a position of genuine goodwill. And, from change order eleven onward, Braeburn Cost Advisors and a reviewer named Simone Adeyemi.

What happens

The register

Sixty-one change order requests over twenty-six months, sorted by source. This is what an aggressive change program actually produces — and I want you to notice that the first two rows are excellent.

Look at the top two rows before you look at anything else. Ninety-one percent recovery on owner-directed work, a hundred percent on the AHJ items, and the district's average response time on Curtis's first ten change orders was nineteen calendar days against a contractual twenty-one. Curtis is good at this. He priced clean scope fast, he knew Delphine Aguirre's board calendar, and he got paid.

Now look at row six. Seventeen change orders, $702,000 submitted, $88,000 collected. That row is Fenner Masonry's exclusion list, converted into paper. The cast stone, the scaffolding, the cold-weather protection — Curtis carried a bid that did not include them and then tried to sell them to the district as changes. Delphine's answer was the correct one and she gave it in one sentence: "The drawings show cast stone sills. Your bid was for the drawings."

And look at row eight. Four impact claims, $319,000 submitted, nothing paid. Not discounted. Nothing.

Change order eleven

Month ten. Curtis submitted a change order request titled Additional supervision, coordination, and cumulative impact arising from changes to date. The amount was $148,000, and the basis was one line: 8% of all labor performed to date, taken from a published inefficiency-factor table under a condition heading that read something like "moderate change and disruption."

There was no measured mile. There was no segregated cost code. There was no scrub — no giving back the hours Kestrel had caused itself. There was no attempt to connect the 8% to any particular change, or to show that the labor it was applied to had anything to do with the changes at all.

Delphine Aguirre called and asked for the backup.

CURTIS: That is the backup. That's the industry standard table. Everybody uses it.

DELPHINE: Curtis, I have to hand this to a board of seven elected people, three of whom will read it. I can't hand them a table from a book. What did it cost you?

CURTIS: It cost me a foreman running three areas instead of one for four months. You've seen it. You've walked it with me.

DELPHINE: I have. And I believe you. I still can't pay it.

She is not being difficult. She is describing her actual constraint, and Curtis — who is genuinely good with owners — should have heard it. Instead he read it as an owner getting tough, and the next week he submitted two more table-based impact claims.

Delphine engaged Braeburn Cost Advisors the following month.

What the consultant did to the job

That is what happened, and the mechanism is worth stating plainly: an owner who has been handed one unsupported number starts checking every number. Simone Adeyemi read all sixty-one. She was not hostile — her reports were careful, and on two items she recommended the district pay more than it had offered. But every submission now took forty-six days instead of nineteen, every submission needed a second round of backup Curtis's lean staff was not built to produce, and Kestrel was financing all of it.

The district paid Braeburn roughly $84,000 over the remaining sixteen months. It was the best $84,000 the district spent.

The one that got away quietly

In month seven, Perry Sandoval answered an RFI about a conflict between the mechanical drawings and the equipment schedule with four words: "Provide per the equipment schedule." Tallgrass Mechanical had bid the drawing. The schedule was the more expensive scope.

That answer created a change on the day it was written. Nobody logged it. Terrence Bell filed the response, Tallgrass began installing to the schedule because that is what the answer said, and eleven weeks later it surfaced as a $67,000 overrun on cost code 23-00 in a monthly cost report.

By then the notice clock had run four times over, and the answer was in a document Curtis had accepted without comment. Tallgrass ate part of it, Kestrel ate the rest, and nobody ever submitted a change order at all.

The change was created on the day of the answer. The claim died on the day nobody logged it. RFI responses, meeting minutes, and submittals returned "approved as noted" are the three places changes hide, and a project without a log finds all three of them eleven weeks late (Chapter 25).

The gymnasium roof

The best claim on the job was the one that was denied.

In week 9, Curtis's crew opened the gymnasium roof coordination and found the structural and mechanical drawings had never been reconciled — the main supply duct and a roof joist wanted the same space. Curtis wrote an RFI, properly and early. The answer took five weeks and required deeper joists in the gym bay. He priced it at $214,000 and 21 calendar days.

The architect denied it: coordination is a contractor responsibility under the general conditions, and the contractor has a duty to review the documents and report discrepancies. Curtis filed a claim, and eleven months later it settled for $96,000 and zero days — $118,000 eaten (Chapter 3).

Then came the part that cost more than the $118,000. Curtis stopped writing early RFIs. He started building to his own interpretation and arguing afterward. His reasoning was not stupid — an RFI had cost him five weeks and gotten him denied anyway. That single behavioral change is the parent of most of row five in the register: fourteen design-error change orders with late notice or no backup, thirty percent recovery.

Analysis: did the strategy make money?

What the executed change orders actually contained

$841,000 sounds like a successful change program. Here is what it is made of.

Kestrel's margin on $841,000 of change orders was $45,800. Everything else was somebody's cost moving through the contract. That is the first thing a change-order strategy gets wrong: it confuses revenue with profit on a mechanism the contract has deliberately capped at ten and five percent.

And the 22 compensable days, at the shaved rate: 22 CD × $2,400/CD = $52,800. At the honest as-built burn of $2,642.62/CD they would have been worth $58,146. Curtis gave away $5,346 of legitimate entitlement on bid day and collected it in change orders twenty months later.

Where the $2,623,000 went

Sit with the fourth row. Curtis submitted $1.3 million of change orders that simply evaporated — no executed change order, no denial he pursued, no claim, no entry in any final accounting. They were written, argued about, filed, and forgotten. Every one of them consumed his time, Terrence's time, Simone Adeyemi's time, and the district's patience.

The honest bottom line

And then the $1,299,600. Not all of it was real cost — a good deal of it was priced at what Curtis hoped the district would bear, which is exactly the line §31.11 draws. But a consultant retained after the job put roughly $430,000 of it as cost Kestrel genuinely incurred on scope the district genuinely owned, and simply never proved. Add even that, and the change-order program is a $624,750 hole.

A strategy built on recovering margin through change orders returned negative six hundred thousand dollars.

Two honest caveats, because this is an accounting and not a sermon. First, this is not additive with Chapter 13's finding that the four bid-day shaves cost $998,000 and turned a $784,000 fee into a $214,000 loss. They are two views of the same money. The masonry exclusions appear in both — once as a scope gap on bid day, once as seventeen rejected change orders twenty months later. The shave and the change program are the same decision seen from opposite ends of the job. Second, the nine unresolved proposals and the schedule entitlement go on to become a $1,375,800 claim, and what that is actually worth is Chapter 33's subject, not this one.

What it cost that has no invoice

The district. Rivermont Unified passed a new bond program and, as many public owners do, adopted contractor prequalification for projects above a stated threshold. The scoring criteria included change-order and claims history. Kestrel did not clear the threshold for Rivermont Middle School #4 ($34.6M) or the districtwide mechanical modernization package ($11.9M) — $46.5 million of public work it could not bid on. Be honest about what that number is: Kestrel would not have won both and might have won neither. The loss is not the fee. The loss is the option, and options are exactly what a mid-size contractor lives on.

The subcontractors. Tallgrass Mechanical's next proposal to Kestrel on public work carried a two-page exclusion list and a separate line for "change-order administration." Fenner Masonry did not bid. Subcontractors learn the game you teach them, and they price it back to you (Chapter 16).

The design team. After the gym roof, Perry Sandoval's office started routing every Kestrel RFI through its own counsel's review. Response times on that job went from ten working days to nineteen calendar days — which then became the causation problem at the center of Curtis's own delay claim (Chapter 33). He built the bottleneck he later claimed against.

Why Curtis is not a villain

Because he is right about several things, and one of them is important.

He did get 90.7% recovery on owner-directed work in an average of nineteen days, which is better than most project managers ever achieve. He was owed money — the consultant's $430,000 says so, and Delphine Aguirre would have paid a meaningful share of it against real records. He was not padding out of greed; he was pricing what he sincerely believed the job had cost him, using the only tool available to a man with two labor cost codes and no measured mile: a percentage.

That is the whole mechanism. Curtis did not choose to be aggressive. He chose to be lean, and aggression is what lean looks like when the bill comes due. A project manager with no field engineer writing tickets, no segregated codes, no contemporaneous schedule updates, and no measured mile has exactly one way to describe a real loss, and it is a number with a percent sign after it. Owners have learned to reject that number on sight — and they learned it from people like Curtis, which is why the honest impact claims that come after his get rejected too.

Ray's version of this, offered without much comfort in it: the difference between Curtis's change program and mine is not integrity. It is that I spent eleven thousand dollars a year on a field engineer whose job is to write things down, and he spent it on his fee.

Discussion Questions

1. Curtis's first ten change orders recovered 71.1% in an average of nineteen days. His last fifty-one recovered 26.7% in forty-six days. Identify the single submission that separates those two columns and explain the mechanism by which one document changed the price of every subsequent one. Then name the equivalent document on a job you have worked on.

2. The $148,000 impact claim at change order eleven was, by Curtis's account, describing a real loss. Rewrite it. Using only §31.5.5, specify exactly what records Curtis would have needed to be keeping from month one for a version of that claim to survive Simone Adeyemi's review — and estimate what those records would have cost him.

3. Kestrel's margin on $841,000 of executed change orders was $45,800 — about 5.4%. Curtis's job fee was 3.5%. Construct the strongest possible argument that a change-order-driven strategy is therefore rational, then dismantle it using the disposition table.

4. Delphine Aguirre paid roughly $84,000 for a consultant to review sixty-one change orders. Compute her return on that decision from the district's side. Then argue the other position: that hiring Braeburn cost the district money it cannot see, and say where that cost shows up.

5. Curtis stopped writing early RFIs after the gym roof denial. That was a rational response to his own experience and it was catastrophic. What structural change — in the contract, in Kestrel's process, or in the delivery method — would have made the rational response and the correct response the same response?

Your Turn

You have inherited Rivermont Elementary School #12 at month eleven. Change order eleven has been submitted and Delphine Aguirre has just told you she is engaging a consultant. Nothing has been logged since month four. There are two labor cost codes on the entire job. You have Terrence Bell and no one else.

Produce, in two hours:

1. A triage of the register. Sort every open item into three piles — provable today, provable if I start keeping records now, and gone. State your criterion for each pile in one sentence.
2. The withdrawal letter. Pick the impact claim you are going to withdraw, and write the paragraph you send Delphine. It must withdraw the item without conceding entitlement on anything else, and it must say why in a way that makes your next submission more believable, not less.
3. The four records you start tomorrow, named specifically, with who creates each one and how long it takes them per day. You have one field engineer; if your answer needs two people it is not an answer.
4. A one-page memo to Nadia Haddad stating the pending-change exposure, the financing cost of carrying it, and what you are asking her to do about the two items above $100,000.
5. The measured mile you can still build. Find one — the classroom wing was built before the change flurry and levels the same details as the wing built during it. Say which period you would use as the unimpacted baseline, what data you need to reconstruct it, and the one sentence you would say out loud in the meeting about how much weaker a reconstructed baseline is than a contemporaneous one.