CEO tenure collapsing because boards got impatient and activist investors demand quarterly results

I'd predict we'll see this play out most visibly in industries where the structural problems are *capital-intensive and long-cycle*, and least in software or other hit-driven sectors. A manufacturing company with a bloated supply chain or a utility with aging infrastructure needs five years minimum to show real improvement. A social media company pivoting its algorithm might show gains in two quarters. The boards cycling CEOs every three years are probably fine—the real damage is happening at exactly the companies where it matters most. The other prediction: when we finally measure this properly (and someone will, because the underperformance vs. peer retention will eventually show up in a paper), the effect size will be large enough that it looks obvious in retrospect. We'll find that the five-year tenure group systematically underperforms ten-year on long-term value creation, controlling for industry and initial conditions. It'll be one of those things where a competent analyst should've caught it years ago, but nobody was looking because the incentives all pointed the other way. The apprenticeship analogy is tight—you can't actually see the damage until the cohort matures.

The apprenticeship analogy works until you ask what happens when the apprentice is actually bad at the job. Your system didn't measure whether they finished, sure—but did it measure whether the crews they joined got better or worse? Because there's a real difference between "we pull people before they learn anything" and "we pull people because this particular person isn't learning or isn't capable of the work." I spent years watching this play out in operations. You'd get a new service manager who'd come in, say all the right things about long-term planning, then spend eighteen months not making any hard cuts because "we need time to see the data." Meanwhile your peak-hour crowding gets worse, your cost per passenger goes up, and your operators are working splits that destroy their families. At some point the board asking questions isn't impatience—it's noticing that nothing's actually changed. You need someone who can both do the work and be honest about whether the previous person was doing it. The real problem isn't that boards measure quarterly. It's that they're measuring the wrong things, and they're not honest about how long structural fixes actually take. A CEO who needs five years to fix cost structure should be able to point to specific, measurable progress at year two—not just "trust me, it's working." If she can't show that, maybe she's the problem, or maybe the board hired someone who can't communicate under pressure. Either way, swapping her out doesn't fix it if the next person comes in with the same measurement dysfunction.

You're identifying something real about incentive misalignment, but I'd push back on one piece: the "nobody's measuring" part. There's actually a ton of measurement happening. The problem is we're measuring the wrong thing, and we know it. I watched this in legal services contracting. A firm would win a big immigration litigation contract, hire bilingual staff, build infrastructure. Year one looked messy—training costs, process development, error rates higher than they should be because people are still learning the actual work. Year two, suddenly efficient, margins looked good. Year three, contract got rebid, firm lost it to someone promising lower costs who'd never done the work before. The cycle repeated. Everyone involved—the procurement officer, the firm's CFO, the board—could see this was stupid. The measurement system (lowest cost per billable hour) was optimized for something other than what actually mattered (accuracy, consistency, not having to redo work). But changing it meant admitting the first year would look bad on paper, and nobody wanted to own that. With CEOs it's the same mechanism but worse stakes. The board has quarterly reporting requirements and analyst calls. A CEO who inherits operational debt and says "this'll take four years" is unpopular immediately, even if she's right. The new CEO who comes in, cuts 15% of staff, realizes gains, and leaves in year three looks like a hero to the people measuring at year three. The person who has to staff back up and deal with the institutional knowledge that walked out the door doesn't get to blame their predecessor in a way that matters. The real constraint isn't impatience. It's that the people making the measurement decisions have different time horizons and different accountability than the people living with the results.

I spent fourteen years watching this exact dynamic play out in immigration courts, and the mechanics are identical to what you're describing. A new immigration judge arrives—competent, energetic, wants to clear the backlog. Six months in, the metrics look great: cases resolved per month up 30%. The appellate reversal rate shows up a year later, after the judge who made those decisions has moved to another district. Nobody's measuring whether the legal reasoning held. The system optimized for throughput, not for decisions that survive scrutiny. The interpreter angle makes this worse. When you cycle through judges constantly, you lose the institutional knowledge of how to actually communicate with the people in the room. A good judge learns over years which interpreters are reliable, what questions generate interpreter errors, when a interpreted testimony needs to slow down. New judge arrives, doesn't have that, operates on raw speed. I've watched someone's asylum case turned on a single word that got interpreted loosely because nobody had built up the relationship with the interpreter to catch it. The judge was gone in two years. The decision stayed. Your point about measuring the wrong time horizon is the real problem. You can't see if someone's decisions were actually sound until you're far enough downstream. But boards and investor relations want something to report this quarter, so you get forced CEO turnover and forced judge reassignment and forced crew rotation. Everyone learns to optimize for what you're measuring in the window you're measuring it in. The actual work—whether structural problems got solved, whether the quality is real—that's somebody else's problem now.