Case Study 20-2 — What the Acceleration Actually Cost

Kestrel priced the Northgate steel recovery at $168,000. Nine months later, Wei Chen measured what it really cost — and the difference is the most valuable number in this book.

Setup

You know this story. It is the spine of half the arithmetic in this book, and this case study is the first place anyone counts all of it.

The failure. The anchor-bolt and embed submittal sat in Kestrel's own office for 11 days before it went to Caldwell Structural. Ruth Caldwell's office then took its full 14-day contractual review. Ironbridge Steel missed its mill rolling slot; the next opening was five weeks out. Steel erection start slipped from August 4 to August 27, Year 1 — 23 calendar days — and steel erection was on the critical path.

The exposure. Northgate's total daily cost of slipping substantial completion is $10,650 per calendar day: $5,150 of extended general conditions plus $5,500 of liquidated damages.

The decision Nadia Haddad had to make, presented to her by Wei Chen on one page:

On the arithmetic alone, accelerating saves $13,050 — a rounding error on a $47.5 million job, and inside the noise of any forecast.

The real driver was not on the page. Meridian Health System's leased interim clinic space expires October 1, Year 2. Six days late is survivable. Twenty-three days late means Meridian is paying holdover rent on a lease they cannot extend, in a market where they have no alternative, while their board asks the CEO why. Pri Sethi did not say this in a meeting. She said it to me in a parking lot.

Kestrel accelerated. Weeks 32 through 38.

What Happens

Part 1 — What the $168,000 actually bought

The acceleration package Tomás Reyes priced and Nadia approved:

Look at what is in that table: premiums, mobilizations, and equipment. Every line is something a vendor invoiced. Every line is something Kestrel could see coming.

Now look at what is not in it: a single dollar of lost productivity. Not on Ironbridge's crews, not on Kestrel's own crews, not on the trades whose work got resequenced around them.

That is not incompetence. It is the normal state of acceleration pricing in this industry, and the reason is structural: disruption does not arrive as an invoice. It arrives as hours that are slightly less productive than they should have been, spread across seven trades and nine weeks, visible only if somebody was measuring in units of man-hours per unit before the disruption started.

Kestrel was, barely. Wei Chen had been tracking productivity factors by cost code and by area since week 6, and manpower counts every Friday. That habit is the only reason this case study has numbers in it.

Part 2 — The three unpriced costs, measured

Nine months later, at the lessons-learned review, Wei presented this.

Bucket 1 — Overtime efficiency loss on Kestrel's own crews.

To keep the deck pours and edge work in front of Ironbridge's second crew, Kestrel's self-perform concrete and carpentry crews went to a 10×6 schedule for five weeks — two crews of eight, sixteen people.

Worked hours = 16 workers × 60 hr/wk × 5 wk = 4,800 MH

Applying the illustrative efficiency curve from §20.7 — week 1 at 1.00, then 0.95, 0.90, 0.87, 0.85, an average of 0.914:

Effective hours = 4,800 × 0.914 = 4,387 MH Hours bought and not converted into work = 4,800 − 4,387 = 413 MH 413 MH × $54.00 = $22,300

Wei cross-checked it against actuals rather than trusting the curve: the concrete and rough-carpentry cost codes ran a productivity factor of 0.91 during weeks 32–37 against 1.01 in the eight weeks before. Applied to the same 4,800 hours, that measured degradation implies a loss in the same neighborhood. When your model and your measurement agree, you can put the number in front of a CFO.

Bucket 2 — Trade stacking and congestion from the enclosure resequence.

"Resequence the enclosure by area" is a clean phrase on a recovery plan. In the field it means precast and curtain-wall crews working in areas where interior framing and MEP rough-in had already started, because the areas were released out of order.

Wei's exposure count, weeks 33–38, in the affected areas:

At a 12 percent congestion factor — the low end of the "three or more trades stacked" band in §20.7, chosen deliberately to be conservative:

20,200 MH × 12% = 2,424 MH lost At a blended burdened trade rate near $62/hr = about $150,000 of craft time

Kestrel did not pay $150,000. Those are mostly subcontractors' hours, and subcontractors absorb a great deal of disruption without ever saying so. What Kestrel paid was what came back as claims: Sofia Marchetti at Cardinal Mechanical and the curtain-wall contractor submitted disruption claims totaling $84,000. Kestrel settled both at $47,500.

The reason the settlement was $47,500 and not $84,000 is worth stating plainly: Wei had contemporaneous manpower and productivity data by area, and the subcontractors mostly did not. The party with the measurements sets the terms of the conversation. That is the same lesson CO #14 taught from the other side — the price of a change is set by what you can document, not by what it cost you — and here Kestrel was on the winning side of it. Do not mistake that for a moral victory. The other $100,000 of loss was real; it was just absorbed by people who could not prove it, and it will come back in their next bid.

Bucket 3 — Rework.

The second erection crew, working Saturdays and reading from a detail one revision behind, set the deck-edge pour stop to the wrong dimension on two bays of level 3. Caught at layout for the topping pour.

Plus three days of float on the level-3 topping pour, which did not cost money because there was float to spend — that time. Note what caused it: not carelessness, but a new crew, on a Saturday, with thinner supervision and a document-control gap. All three of those conditions were manufactured by the acceleration.

Bucket 4 — Safety.

Week 34. North elevation, level 3. A frame scaffold partially modified overnight by a different trade running conduit; a plank lifted and not re-secured. Milo Serrano, a mason tender, stepped onto it at 7:20 a.m. The plank shifted and he went down onto the platform, catching himself on the top rail. No injury.

Bea Salgado stopped work on the elevation for the day.

Bea's investigation found three failures: a competent-person inspection tag two days stale, a scaffold modified by a trade that did not erect it with no re-inspection, and — the finding nobody wanted to write down — a crew running behind after the steel acceleration, under an unwritten expectation to make it up.

$9,800 is what the near-miss cost. It is not what it was worth. Weeks 34 through 36 also carried a visible spike in reported near-misses across the job. That spike is the system telling you something, and on this occasion it was telling the truth and nobody was seriously hurt. The next time is a coin flip that Kestrel did not pay for.

Part 3 — The total

Unpriced cost = $263,800 − $168,000 = $95,800 $95,800 ÷ $168,000 = 57%

The disruption cost fifty-seven percent of the priced premium, and nobody put a dollar of it in the decision.

Part 4 — Re-running the decision with the real numbers

Measured after the fact, the acceleration cost about $82,750 more than simply absorbing the delay would have.

Analysis

Here is where most people draw the wrong conclusion, so let us be careful.

The decision was right. The pricing was wrong. Those are two different findings and they do not cancel.

Why the decision was right. The $10,650-per-day number captures Kestrel's exposure — extended general conditions and liquidated damages. It does not capture Meridian's exposure, and Meridian's exposure was the October 1 clinic lease. Twenty-three days late puts the owner in a position with no good options and puts Kestrel in the position of being the reason. Nobody at Kestrel wrote a dollar figure for "Meridian never calls us again," and nobody should have tried, but it is not zero and it is not small. Kestrel is a $410 million contractor with a repeat-client business model; a health system client is worth more than $82,750 a great deal of the time. Accelerating bought a relationship at a measured price of $82,750, and stated that way, most executives would buy it again.

What is not defensible is that nobody knew that was the trade they were making.

Why the pricing was wrong, mechanically. Acceleration pricing captures what vendors invoice. Disruption is not invoiced — it is diffused. It shows up as:

• hours that are 9 percent less productive across five weeks on your own payroll;
• subcontractors quietly absorbing congestion losses they cannot prove;
• a rework event on a different cost code, weeks later, whose cause is upstream;
• an incident rate that rises without an incident.

None of these has a vendor, an invoice number, or a person whose job it is to report them. A cost that has no owner does not get priced.

The three rules Kestrel wrote down afterward

1. Price a disruption allowance on every acceleration package, and label it. Not as contingency — as a named line: "disruption and productivity loss allowance." On Northgate it would have been about 57 percent of the premium. Do not memorize 57 percent. It depends on how much overtime, how many trades, how congested the areas, how long it runs. Build your own factor from your own measured jobs, and until you have one, use a range and say it is a range.

2. Present acceleration as three numbers, never one. Premium cost · disruption cost · residual exposure. A recovery plan that shows only the premium is a recovery plan that will be approved on false arithmetic, and the person who approved it will remember that you presented it.

3. Measure before you disrupt. Wei's productivity factors by cost code and area, tracked from week 6, are the entire reason there is a $47,500 settlement instead of an $84,000 one, the reason Bucket 1 could be cross-checked against actuals, and the reason this analysis exists at all. You cannot measure disruption without a baseline, and you cannot create a baseline retroactively. Contrast Curtis Boone on Rivermont Elementary School #12: same industry, same company, no cost-code discipline, no manpower tracking — and when his job ends in a claim, he has a story and no measured mile. Chapter 33 is where that difference becomes money.

One more thing, and it is the important one

Go back to Bea's third finding: a crew running behind after the steel acceleration, under an unwritten "make it up" pressure.

That pressure was not in the recovery plan. Nobody wrote it, nobody said it out loud, and if you had asked me at the time whether I was pressuring the masons, I would have said no and meant it. It came down the tiers anyway — from an owner's date, through a prime contract, through a flow-down clause, through a subcontractor's foreman, and out onto a scaffold at 7:20 in the morning.

Every dollar in the table above is a productivity number. The line under the table is a safety number, and they were produced by the same decision. That is theme 4, and it is why acceleration is a safety decision that happens to be denominated in dollars. Chapter 24 and Chapter 29 take the two halves of it.

Discussion Questions

1. Kestrel settled $84,000 of subcontractor disruption claims for $47,500 because Kestrel had contemporaneous data and the subcontractors did not. Is that a good outcome? Argue both sides, and then say what you would actually do as the general contractor's PM — including whether "they couldn't prove it" is a sufficient reason to pay less than what you believe the loss was.

2. The unpriced cost was 57 percent of the priced premium on this job. Name three project characteristics that would push that ratio higher and three that would push it lower, and explain the mechanism for each.

3. Rebuild Wei's one-page decision memo the way it should have been presented to Nadia in week 31 — three options, three cost components each, plus the non-monetary factors stated in words rather than buried. Keep it to one page. Then ask yourself honestly: would Nadia have decided differently, and does that change whether the memo was worth writing?

4. Bucket 2 estimated $150,000 of craft time lost across all parties and $47,500 of cost to Kestrel. Where did the other $100,000 go, and what happens to it over the following two years?

5. The rework event was caused by a new crew working Saturday from a superseded detail. Design the specific control that prevents it — not "better document control," but the actual mechanism, who executes it, and when. What does it cost, and would you have paid for it in week 31?

Your Turn

You are the project manager on a $31 million job. You are 14 working days behind on a critical-path activity with roughly 9,000 remaining man-hours of work in it. Your daily exposure is $7,400 per calendar day of extended general conditions and liquidated damages combined.

Your superintendent proposes a 10-hour, six-day schedule for eight weeks on the affected trades.

Produce a one-page recovery memo with three costed options. Include, for each: the premium cost, an estimated disruption cost with your reasoning and your factor stated as a range, the residual days not recovered, and the total. Then add a short section headed "What this memo cannot price" — and put the honest things in it.

Two constraints. One: you must show your overtime premium computed on the base wage with statutory loading, not naively at 1.5 times the burdened rate. Two: at least one of your three options must not involve overtime at all.