Case Study 2 — The Meeting That Improved While the Job Got Worse: Rivermont Elementary School #12
All people, companies, and projects in this book are illustrative composites. Numbers are consistent within the book and are not drawn from any actual project.
Setup
The project. Rivermont Elementary School #12. $22,400,000, hard-bid design-bid-build, lump sum. Public owner — the Rivermont Unified School District. Prevailing wage. One hundred percent payment and performance bonds. A summer-to-summer schedule with a hard occupancy date the week before school starts, because a public school district cannot open late and there is no version of the story in which it does.
The contractor. Kestrel Construction Group. Different job, different team, same company as Northgate.
The people.
- Curtis Boone — Kestrel project manager. Fifteen years. Wins work by bidding it tight, buys out hard, runs a lean staff, and manages by force of personality. Genuinely skilled, genuinely likable, and wrong about the model. Everyone likes working for Curtis until month eight.
- Trevor Nyland — project engineer. One person. Owns submittals, RFIs, procurement tracking, cost coding, and now, as of this case study, the Last Planner System.
- Gus Petrakis — superintendent. Twenty-six years. Skeptical, not hostile.
- Nadia Haddad — Kestrel vice president of operations, who issued the mandate.
- Eleven subcontractor foremen, including Kilbourne Mechanical, Redfern Electric, Overton Masonry, and Dutchess Drywall.
The mandate. After the Northgate results, Nadia Haddad sent a one-page memo to every Kestrel project manager: effective immediately, all projects will implement the Last Planner System — weekly work plans, percent plan complete tracking, and reason-code reporting to the operations group monthly.
The memo said what to do. It did not say who would do it, it did not fund anything, and it did not change a single term in a single subcontract.
Curtis read it and said, out loud, to Trevor: "Fine. It's a meeting. We can do a meeting."
That sentence is the whole case study.
What Happens
Month one — everybody shows up
The first weekly work plan meeting on Rivermont #12 was, by any reasonable measure, good. Eleven of eleven foremen attended. Twenty-six commitments were made. Curtis ran it himself, and Curtis is excellent in a room. PPC came in at 62%, which is a perfectly normal first number and slightly better than Northgate's.
Corporate was pleased. Curtis reported it to Nadia's operations group with a note: "Off to a strong start."
Three things were already wrong and nobody could see them yet.
Trevor Nyland had been given the make-ready role on top of a full workload. He owned the submittal log, the RFI log, procurement tracking, and cost coding on a $22.4M job with no other project engineer. The constraint log was item number five on a list of four things he could actually do in a week. He built it. He printed it. He did not work it.
Curtis made no commitments of his own. The board had eleven trade rows and no Kestrel row. It did not occur to him that he was a trade with deliverables; it occurred to him that he was the person running the meeting.
Nothing in any subcontract had changed. Curtis had bid the job tight and bought it out hard — he had shopped three of the eleven trades in the process. The subcontractors were on fixed-price agreements with thin margins, no obligation to attend a planning meeting, and a well-earned belief that Curtis's schedule was Curtis's problem.
Month two — the first reason-code review
Curtis put the reason-code Pareto on the wall and did what came naturally to him, which was to be direct about it.
Curtis: "Nine misses. Four of them are Kilbourne. Ray, you want to tell the room what happened?"
The Kilbourne foreman explained that the block work in the corridor was not done, so he could not hang his overhead.
Curtis: "Tony, that's your third week telling me the masonry's not ready. Overton, where are we?"
The Overton foreman said his mason tenders had been pulled to another job for four days.
Curtis: "Right. So it's Overton."
It took about nine minutes and it was not cruel. It was a project manager doing what project managers do — locating responsibility. And it destroyed the system.
PPC that month: 58%. Attendance: ten of eleven.
Gus Petrakis said something to Trevor in the parking lot afterward that turned out to be the most accurate prediction anyone made on that job: "They're going to stop telling him things."
Months three through six
| Month | Foremen attending (of 11) | Commitments made / week | Completed / week | PPC | Schedule position |
|---|---|---|---|---|---|
| 1 | 11 | 26 | 16 | 62% | On schedule |
| 2 | 10 | 24 | 14 | 58% | 2 CD behind |
| 3 | 9 | 18 | 12 | 67% | 5 CD behind |
| 4 | 7 | 14 | 10 | 71% | 9 CD behind |
| 5 | 6 | 11 | 9 | 82% | 13 CD behind |
| 6 | 5 | 9 | 8 | 89% | 17 CD behind |
Read that table twice, because it is the most important table in this chapter and possibly in this part of the book.
Percent plan complete rose from 62% to 89%. By the headline metric, Rivermont #12 was one of the best-performing lean implementations in the company. Curtis reported 89% to the operations group in month six with a note that said the system was "fully mature."
Completed commitments per week fell from 16 to 8. The actual weekly output of reliable, coordinated handoffs was cut in half.
The job went from on schedule to seventeen calendar days behind.
Here is exactly how it happened, month by month, in the foremen's behavior:
- Month three. Foremen learned that a miss produced a public conversation and a hit produced nothing. The rational response is not to work harder; it is to commit to less and commit to safer things. Commitments dropped from 24 to 18 a week and got smaller. The Overton foreman started committing to work he had already essentially finished on Friday.
- Month four. Foremen who had other jobs started sending someone else — an apprentice, a lead man, once a delivery driver who happened to be on site. A person who does not control the crew cannot make a commitment, so the commitments got vaguer and safer still. Attendance: seven.
- Month five. The meeting became a status update. Nobody was screening anything, because Trevor had not touched the constraint log since week three. The look-ahead was printed every Friday and read by nobody.
- Month six. Five foremen, nine commitments, 89% PPC. Curtis was genuinely pleased. Gus Petrakis had gone back to running the job the way he had run jobs for twenty-six years, which is by walking the building at 6 a.m. and shouting at people, and which was — given what the meeting had become — the correct decision.
The moment it became visible
Month six, the district's construction representative asked, in the owner-architect-contractor meeting, why the gymnasium was seventeen days behind when the contractor's own production reports showed 89% plan reliability.
Curtis did not have an answer. There is not one. The two numbers are measuring different things, and one of them had stopped measuring anything at all.
Analysis
The four conditions, and how each one failed
From §27.9 of the chapter: four conditions have to be true. On Rivermont #12, all four were false from the first week.
| # | Condition | What actually happened |
|---|---|---|
| 1 | Somebody has real, protected time to remove constraints. | Trevor Nyland was handed make-ready as a fifth priority on a four-priority week. Nobody was taken off anything. The constraint log existed as a document and never as an activity. Total hours per week actually spent removing constraints: effectively zero after week three. |
| 2 | The general contractor makes and keeps its own commitments, publicly, on the same board. | There was no Kestrel row on the board. Over six months Curtis verbally committed to roughly fourteen things — RFI pushes, access, a hoist, a temporary power drop — and delivered four. None of it was recorded and none of it was scored. Foremen were being asked to be measured by a party that was not. |
| 3 | A miss is safe to declare. | Month two, nine minutes, in front of ten people. That was the end of it. Everything from month three onward is the predictable consequence: the participants optimized the metric because optimizing the metric was the only safe move available to them. |
| 4 | The commercial terms let the parties share the benefit, or the GC accepts it is investing. | Fixed-price subcontracts, three of them bid-shopped, thin margins, no contractual obligation to participate, no incentive of any kind. The subcontractors were asked to expose their float to a general contractor who had already demonstrated, during buyout, how he treats information they give him. |
The mechanism, stated plainly
PPC is a ratio, and the denominator is controlled by the people being measured. Anywhere you find that arrangement — in construction, in sales, in safety statistics, in software — you will find the metric drifting away from the thing it was supposed to represent, and you will find that the drift looks like improvement.
This is why the chapter insists on tracking commitments made alongside PPC. On Rivermont #12, that single extra column would have made the failure visible in month three instead of month six:
RIVERMONT #12 — WHAT ONE EXTRA COLUMN WOULD HAVE SHOWN
Month: 1 2 3 4 5 6
PPC: 62% 58% 67% 71% 82% 89% ← what Curtis reported
▲ ▲
└──────── "improving" ────────┘
Commitments made: 26 24 18 14 11 9 ← what nobody tracked
Completed: 16 14 12 10 9 8 ← the real output
▲ ▲
└──────── halved ─────────────┘
Schedule: 0 -2 -5 -9 -13 -17 CD
What Curtis got right, which matters
It is important not to turn Curtis Boone into a cartoon. He is not lazy and he is not stupid. He did what the memo said. He held the meeting every single week for six months, which is more than a lot of project managers manage. He was direct about performance, which is usually a virtue. He reported his numbers accurately — the 89% was real arithmetic on real data.
He was defeated by a mandate that specified an activity instead of a system. Nadia Haddad's memo said "implement the Last Planner System." It did not say "staff twenty hours a week of constraint removal," it did not say "put your own commitments on the board," it did not say "here is what you may not do with the data," and it did not change a word in the subcontract template. A project manager who read that memo literally and executed it competently would produce exactly Rivermont #12.
The corporate failure is the bigger failure. Kestrel took a result from one project — a project with a dedicated make-ready resource, a general contractor scoring itself, and a project manager who deliberately declined to weaponize the data — and rolled out the artifacts of that result without any of the conditions that produced it.
What it cost
Seventeen calendar days on a school with a hard occupancy date. Kestrel recovered eleven of them over the summer with premium time and a second drywall crew, at roughly $96,000 of unrecovered acceleration cost, and turned the building over four days late with a temporary certificate of occupancy and a district that has a long memory about which contractors it recommends.
And the second cost, which is larger and never appears anywhere: eleven subcontractor foremen now believe the Last Planner System is a management fad that is used to blame them. They will carry that onto the next four jobs they work, including yours. Every failed implementation raises the cost of the next real one.
Discussion Questions
- Rewrite Nadia Haddad's one-page mandate. What five things must it specify to prevent Rivermont #12 — and what would each of them cost Kestrel across a portfolio of projects?
- In month two, Curtis spent nine minutes locating responsibility for four misses. He is a project manager; locating responsibility is his job. Script what he should have said instead, in the same room, with the same nine misses on the wall, without pretending the misses did not happen.
- Trevor Nyland was given make-ready as a fifth priority. If you are Trevor and you cannot get more hours, what do you do? Is there an honest version of "I cannot do this" that a project engineer can actually say to a project manager?
- Curtis had bid-shopped three of the eleven subcontractors during buyout. Trace the line from that decision to the month-four attendance figure. Is it a fair line to draw?
- The 89% PPC in month six was arithmetically correct and completely misleading. Name two other construction metrics with the same structural weakness — where the party being measured controls an input to the calculation — and say what second number you would track alongside each.
Your Turn
You are taking over Rivermont #12 in month seven. Seventeen days behind, five foremen attending, 89% PPC on nine commitments, a constraint log nobody has touched since week three, and eleven subcontractors who have concluded this system is a blame machine.
Write the first thing you say in your first weekly work plan meeting — 200 words or less, out loud, to the five people who bothered to show up. You may not promise them anything you cannot deliver, you may not criticize Curtis, and you may not use the word "lean."
Then list, in order, the first three actions you take in that week and say who does each one, in what hours, taken from what. If you cannot name where the hours come from, you have written a memo, not a plan.