Case Study 14-2 — Forty Days That Were Not There: Auditing Curtis Boone's Schedule

The projects, companies, and people in this case study are illustrative composites created for teaching. The arithmetic is real; the names are not.

Setup

Project: Rivermont Elementary School #12 — $22.4M, hard-bid design-bid-build lump sum, public owner, prevailing wage, 100% payment and performance bonds. Contractor: Kestrel Construction Group. Project manager: Curtis Boone — fifteen years in the business, wins work by bidding it lean, runs a small staff, manages by force of personality, and is genuinely good at getting people to do things. He is not a villain. He is skilled, well-liked, and wrong about the model. Liquidated damages: $2,500 per calendar day. Kestrel's general conditions on this job: $2,100 per calendar day. Combined exposure: $4,600 per calendar day. The trigger: at month seven, Kestrel's surety asked for a schedule and a written recovery plan on two jobs in the portfolio. Nadia Haddad asked Ray Alvarez and Wei Chen to audit Rivermont #12 before it went to the bonding company.

Curtis was not worried.

Curtis: "Look at the interiors path. Forty days of float. We're fine. I've got more cushion on this job than I've had on anything in three years."

Ray printed the schedule and started the thirty-minute review from §14.8. It took him nineteen minutes to find the first defect and about two hours to find the rest.

What Happened

The headline number

Curtis's month-seven update, 340 activities, reported 40 work days of total float on the interior finishes chain — the chain that runs classroom MEP rough-in, in-wall rough, inspections, drywall, ceilings, flooring, casework, and final clean.

Defect 1 — A constraint on the finish milestone (−40 WD)

Ray started with the constraint count. Fourteen constrained activities out of 340 — high, but not alarming. Then he looked at which ones.

The project finish milestone, "Substantial Completion," carried a Finish On constraint set to a date 40 work days later than the contract substantial completion date.

The history was ordinary and completely human. In month three, an owner-caused utility delay had pushed the schedule out. Curtis had asked for a time extension, been told it was "under review," and — rather than submit an update showing 40 days of negative float, which would have triggered a recovery-plan requirement in the specification — moved the milestone to the date he expected the extension to produce and kept working.

The extension was never granted.

The consequence is mechanical. A CPM performs the backward pass from the finish. Move the finish 40 days later and every path in the schedule gains 40 days of computed float. The number was not a lie anybody typed. It was a lie the software calculated, faithfully, from a date somebody had entered eight months earlier and forgotten.

Ray: "Curtis, your finish milestone is constrained to March fourteenth. Your contract date is January seventeenth."

Curtis: "Right, because of the utility delay. We're entitled to that time."

Ray: "Maybe. Has anybody granted it?"

Curtis: "...It's under review."

Ray: "Then the forty days aren't float. They're a claim. Those are different things, and only one of them is in your schedule."

Real float on the interiors chain after correction: 0.

Defect 2 — Ninety-six open ends (−8 WD)

Ray ran the successor check. Ninety-six of 340 activities had no successor — 28 percent of the schedule dangling.

Most of them were harmless in the sense that they were genuinely terminal: final clean, punch by trade, O&M manual submission. But eleven of them were predecessor work that something real depended on, and because nothing was tied behind them, the calculation gave each one a late finish equal to the end of the project and enormous float. Every one of the eleven had been filtered out of every look-ahead Curtis had produced for seven months.

Three of the eleven mattered:

When Wei tied the eleven activities into their real successors and recalculated, the driving path grew by 8 work days.

Defect 3 — Soft logic modeled as parallel work (−12 WD)

The interiors sequence in Curtis's schedule ran overhead MEP rough-in and ceiling grid installation as parallel activities across each wing, with no relationship between them. On a Gantt chart it looked efficient. In the field it is impossible: you cannot hang ceiling grid in a room whose ductwork, sprinkler mains, conduit, and plumbing are not in and inspected. Grid goes up after the plenum is closed out, room by room.

The parallelism was not a deliberate deception. It was a schedule built at the wing level rather than the room level, so the interleaving simply had nowhere to live. But the effect was the same as a lie: the schedule showed a chain of work happening simultaneously that in reality happens sequentially, and every day of "float" downstream of that parallelism was fictional.

Wei rebuilt the wing as room-group activities with the real handoffs — overhead rough → inspection → close walls → grid → tile → flooring. The driving path grew another 12 work days.

Curtis: "The subs told me they'd work around each other."

Margo Deacon (on the phone, having walked the job): "Curtis, there's one lift in that wing."

That is resource logic, and it was not in the schedule either.

Defect 4 — Padding at the activity level (the reason nobody could recover)

Ray sampled 40 durations against Kestrel's historical production data and against the subcontractors' own manpower commitments. The durations ran roughly 15 percent long across the board — not one egregious activity, a uniform tax.

This defect does not change the CPM arithmetic at all. It matters for a subtler and more damaging reason: the padding and the "float" were the same contingency, counted twice.

When the owner asked Curtis in month five to absorb a 12-day scope addition "in your float," Curtis said yes, confident he had 40 days. What he actually had was 40 phantom days plus roughly 15 percent hidden in the durations. The crews knew about the padding — crews always know — and they had been consuming it at a normal, unremarkable rate since month one. By the audit, most of it was gone, and there was no record of it going, because invisible contingency cannot be reported.

The corrected number

Twenty work days of negative float. On a five-day calendar that is 20 × 7/5 = 28 calendar days.

Exposure: 28 CD × $4,600/CD = $128,800 — and that is before any acceleration cost, before any disruption claim from subcontractors whose sequences just changed, and before whatever the surety decides to do with a job that reported 40 days of comfort seven months in a row and was actually four weeks behind.

Analysis

Every defect was individually defensible. That is the point of this case and the reason it is worth studying more than an outright fraud would be. Curtis did not fabricate anything. He moved a milestone to a date he genuinely expected to be granted. He built a schedule at the level of detail he had time for. He did not tie successors onto activities that felt like endpoints. He carried durations his subs had given him without challenging them. Each decision was reasonable in isolation, made under time pressure, by somebody competent. The failure was cumulative and invisible, and the schedule reported comfort the entire time.

Constraints are the most dangerous defect because they are silent. Open ends can be counted. Lags can be listed. Padding can be sampled. But a constraint on the finish milestone changes every float number in the schedule and produces no warning of any kind — no red bar, no flag, no exception report. Nothing in the output looks wrong. This is why the constraint audit in §14.7.3 is the first check, not the last.

The 40 days were a claim wearing a schedule's clothing. Curtis had a real potential entitlement to time from the utility delay. Entitlement to time is a thing you pursue, with notice, documentation, and a time impact analysis. It is not a thing you assume and then bake into your baseline. When Curtis moved the milestone, he converted a claim he might have won into a comfort he definitely did not have, and he spent it. If the owner had granted the extension in month four, the schedule would have been right and nobody would have learned anything.

What Curtis should have done at month three, in one sentence: submit the update showing 40 days of negative float, trigger the recovery-plan requirement, give written notice of the delay, request the extension formally, and let the schedule tell the truth even though the truth was going to generate a difficult meeting. The difficult meeting in month three costs a difficult meeting. The same meeting in month seven costs $128,800 and a conversation with the surety.

One more uncomfortable observation. Curtis's project was staffed lean and his documentation was minimal by design — that is how he bids competitively and it is why his overhead is lower than Ray's. There is a real argument that his model wins more work. The rebuttal is this job: the audit cost Kestrel more than three years of the general-conditions savings that lean staffing produced, and the reputational cost with the bonding company is not a number anyone can put in a table. Documentation and schedule discipline are not overhead. They are the cheapest insurance in construction, and their premium is paid in months when nothing is going wrong.

Discussion Questions

1. Rank the four defects by how much damage each one did. Now rank them by how easy each one is to detect in a thirty-minute review. Are the rankings the same, and what does that imply about where to spend your review time?
2. Curtis's argument was that his entitlement to a 40-day extension was real. Assume he is right about the entitlement. Does that make his schedule correct? Explain the difference between a claim for time and float in a schedule.
3. The owner asked Curtis to absorb a 12-day change "in your float." Write the two-paragraph reply Curtis should have sent, assuming he had done the audit first.
4. Ninety-six of 340 activities had no successor and only eleven of them mattered. Does that mean the other 85 were harmless? What would you require before accepting an activity with no successor?
5. Kestrel's house standard is fewer than 1 percent constrained activities and zero mandatory constraints. Curtis had 14 of 340 — about 4 percent. Write the rule you would enforce on your own projects, and describe how you would verify it monthly in under ten minutes.

Your Turn

You have been handed a subcontractor's 180-activity schedule for the mechanical scope on a job you are running. You have thirty minutes before the coordination meeting.

Write the six questions you will ask, in the order you will ask them, with the number or condition you are looking for in each answer. Then write the one sentence you will say if the answer to question one is "we can only send you a PDF."