Case Study 2 — Thirty-Four and Eleven: Two Superintendents, One Company

All people, companies, and projects in this book are Tier-3 illustrative composites. The numbers are internally consistent and realistic; they are not a real project, and the ramp, rework, and injury-cost percentages are stated assumptions rather than measurements.

Setup

Kestrel Construction Group self-performs concrete, rough carpentry, and general trades. That means Kestrel's craft workers are Kestrel's own W-2 employees, so their turnover is directly visible in one payroll system — which makes this the only place in a general contractor's business where you can measure the thing everybody argues about.

Two jobs. Same company, same wage scale, same benefit package, same corporate safety program, same recruiting pipeline, same metro labor market, same fiscal year.

Randy Voss is not a bad superintendent. This is the part everybody gets wrong about this comparison, so I am putting it first. He is technically excellent — he knows formwork better than Jamal does, his concrete is straight, his layout is precise, and Owen Baptiste will tell you Randy's job runs slightly under budget on material. Nadia Haddad has promoted him twice.

He also loses a third of his people every year, and until somebody ran this arithmetic, nobody at Kestrel had priced it.

What Happens

The separations

• Voss: 0.34 × 44 = 15 separations a year
• Foster: 0.11 × 44 = 5 separations a year
• Difference: 10 people a year

What those ten people cost

1. Hard replacement cost. From Chapter 20: roughly $4,600 per craft replacement, covering recruiting, screening, drug screen and background, orientation, badging, PPE, and the foreman's direct training time.

2. Production ramp. A replacement worker does not produce at full rate on day one. Assume 65% of standard output for six weeks, then 85% for six more — a loss of (6 × 40 × 0.35) + (6 × 40 × 0.15) = 120 equivalent man-hours per replacement.

3. Rework. Newer workers make more errors, and a crew with high churn loses the informal quality knowledge that lives in people who have worked together. Kestrel's cost codes track rework labor separately. Voss's job ran 2.9% of self-perform labor as rework; Foster's ran 1.5%. Labor base each: 88,000 MH × $66 = $5,808,000.

4. Safety. Injuries concentrate in a worker's first weeks on a job — a well-established pattern and one of the strongest arguments for stable crews that has nothing to do with morale. Recordable incident rates per 200,000 hours: Voss 4.1, Foster 1.6.

• Voss: 88,000 ÷ 200,000 × 4.1 = 1.80 recordables
• Foster: 88,000 ÷ 200,000 × 1.6 = 0.70 recordables
• Difference: 1.10 recordables

Using an illustrative blended cost of $28,000 direct plus $42,000 indirect = $70,000 per recordable — direct is medical and indemnity, indirect is investigation, replacement labor, retraining, lost production, and administration:

This one also carries a multi-year tail through the experience modification rate, which is not in the table because it lags and is company-wide rather than job-level. It is real and it is not small.

5. Overtime to cover short crews. Voss is chronically two to five people short because he is always replacing somebody, so he covers with premium time: 6% of hours at premium against Foster's 2%. Premium adds roughly $19/hour once burden on the premium is included.

• Difference: 4% × 88,000 MH = 3,520 premium hours × $19 = $66,880

The total

Roughly $350,000 a year, on a crew of 44 people.

Normalize it so the number is portable:

• $350,392 ÷ 88,000 MH = $3.98 per craft man-hour
• Against a burdened rate of $66.00, that is a 6.0% difference in the cost of every craft hour — on a Kestrel-employed workforce, between two Kestrel superintendents, with identical wages and identical benefits.

Six percent of labor is roughly the entire fee margin on a hard-bid job. It does not appear anywhere in any report as "leadership."

Where the Difference Actually Comes From

Nadia Haddad's first question, reasonably, was whether this was a job difference rather than a superintendent difference. Some of it is, and that is handled in the caveats below. But Margo Deacon spent two days on each site and came back with a list of practices, and the list is not subtle.

Count the cost of that column. Nine of the twelve are free. Two — the second sanitary unit and the tool replacement policy — are a few thousand dollars a year. One, the Thursday posting, costs about twenty minutes a week.

Twenty minutes a week and a few thousand dollars against $350,392.

Analysis

The mechanism is not morale. It is predictability. Look at what Foster's practices have in common: they let a person plan their own life. Thursday posting means a worker can arrange childcare. Two weeks of layoff notice means they can line up the next job instead of being surprised into a competitor's hiring hall. A stated end date on overtime means a Saturday is survivable. Named foreman candidates mean a person can see where this goes.

The reasons people give for leaving construction are consistent across every survey anybody has run: unpredictable schedules and layoffs, travel, the physical toll, no visible path forward, and treatment. Foster's twelve practices address four of those five directly, and they address them at the level where they actually operate — the crew — rather than at the level where companies usually try to address them, which is a recruiting campaign.

The feedback loop is the expensive part, and it is why this compounds. Voss is short-handed → he runs overtime to cover → sustained overtime erodes productivity and increases fatigue → the job falls further behind → more overtime, less notice, more pressure → more people leave → he is shorter-handed. Each turn of that loop is individually rational and the loop as a whole is a machine for converting money into exhaustion. Foster is not outside the loop because of temperament. He is outside it because Thursday posting and two-week layoff notice break the link between "the job is hard" and "I cannot plan my life," which is the link that actually makes people quit.

Four honest caveats

I would rather you argue with this arithmetic than repeat it.

1. Some of the gap is job mix, not management. Alder Point had two extended weather delays and a four-week gap between structure and interiors that forced a genuine layoff. A reasonable estimate is that three to five of Voss's fifteen separations were structural rather than behavioral. That reduces the difference — it does not eliminate it, and note that how a forced layoff is handled is still a practice, not a fact.
2. The rework and recordable rates are illustrative. Kestrel does track rework as a cost code, which is more than most contractors do; the 2.9%/1.5% split is realistic and it is not a measurement from a real study.
3. Causation runs both ways on rework and safety. Does churn cause rework, or does poor supervision cause both? Almost certainly both, and the honest position is that they are two symptoms of one management difference rather than a clean chain. That argues for the practice list, not against it.
4. The overtime line partly double-counts the productivity effect. Sustained overtime erodes efficiency, and some of that erosion is already sitting inside the rework and ramp numbers. Treat $350,392 as the top of a range whose bottom is somewhere near $250,000.

Even at the bottom of that range, on a crew of 44, in a company where nobody had ever priced it.

What Kestrel did

Nadia Haddad did not fire Randy Voss, which would have been the stupid move — he is technically excellent and the problem is teachable. She did three things:

1. Added craft turnover by superintendent to the monthly operations report, which had previously reported it only company-wide. What gets measured at the level where it is produced is the entire trick.
2. Made the Thursday posting and the two-week layoff-notice practice a company standard, because both are free and neither requires anybody to change their personality.
3. Sent Voss to spend three days with Foster at 6:00 a.m. Not to a leadership seminar. To a job site, in the dark, watching a man walk the work with his foremen before the crews arrive.

Voss's turnover the following year was 19%. Still high. It is a two-year fix, and it is worth about $200,000 a year while it is happening.

Discussion Questions

1. Six percent of every craft man-hour is roughly the whole fee margin on a hard-bid job. Why does this cost appear in no standard construction cost report, and where would you put it if you were designing one? Name the cost codes and the report it belongs on.

2. Nine of Foster's twelve practices cost nothing. Given that, explain why they are not universal. What does a superintendent actually give up to do them, and why is that price invisible on a spreadsheet?

3. Caveat 3 says causation may run both ways between turnover, rework, and injuries. Design a way to test this using data a general contractor already has. What would you look at, and what result would change your mind?

4. Randy Voss is technically better than Jamal Foster at formwork and layout, and Kestrel promoted him twice. Construct the argument that Kestrel's promotion criteria are the actual root cause here, and say what you would change about them.

5. Voss went from 34% to 19% in one year. Is that a success? Compute the value of the improvement, then argue whether the remaining gap is worth another year of effort, or whether the marginal money is better spent somewhere else on his job.

Your Turn

Build the same comparison for a workforce you can actually observe — your company's craft, a single trade on your job, or your own project engineering staff.

1. Get the separation count for a twelve-month window and the average headcount. Turnover = separations ÷ average headcount.
2. Price one replacement honestly. Recruiting, screening, orientation, PPE, badging, and the supervisor's training hours. Keep the production ramp separate so you do not inflate it.
3. Estimate the ramp — your own assumption about weeks and percentages, stated out loud so somebody can argue with it.
4. Pull rework and recordables if you have them by crew or by job. If you do not have them at that level, that is itself the finding.
5. Then build the practice table. Twelve rows. What does the better-retaining supervisor do on a Thursday afternoon that the other one does not?

Bring the practice table to your superintendent, not the dollar total. The dollar total wins the argument with your CFO. The practice table is the only part that changes anything on Monday.