The quarter is the wrong unit of progress

If my 2022 training block had been a startup, and I had been its board, I would have shut it down in month four.

The middle of that year was the flattest stretch of my athletic life. Month after month of long, slow, deliberately unimpressive work. The paces were not dropping. The long rides were not getting longer. Anyone reviewing my numbers on a ninety-day window would have seen a project that had stalled: high cost, heavy time commitment, no visible progress, and a stubborn founder insisting the plan was working. Every instinct trained by quarterly reviews says you cut that project.

In October I crossed the line at Ironman Barcelona in 9:55:14. The race was not won in race week, and it was not won in the sharp final build that looks so good on a training log. It was won in the flat months, the ones a quarterly review would have flagged as failure.

I have spent a lot of time since then thinking about why the quarter, the unit almost every company is managed by, is so bad at seeing the work that matters most.

The visibility lag

Endurance training runs on a delay. The adaptation you feel today was caused by work you did months ago. The aerobic base, the tendon strength, the metabolic efficiency, all of it accumulates silently and cashes out long after the sessions that built it. Inside any ninety-day window, the relationship between effort and visible result is close to zero. Across a twelve-month window, it is close to everything.

This is not a motivational observation, it is a measurement problem. If you sample a compounding process at a resolution shorter than its compounding period, what you see is noise, and you will make decisions on the noise. The flat stretch in my log was not stagnation. It was loading. But nothing in a ninety-day chart can tell you the difference.

Companies have the same delay, and we mostly pretend they do not. The work that compounds in a company, the data pipeline, the eval sets, the infrastructure that makes the tenth integration cheaper than the first, the slow accumulation of customer trust, the hire you took three extra months to get right, none of it demos well at ninety days. All of it decides where the company is in three years.

The quarterly optics tax

Here is what board cadence actually trains founders to do. Every ninety days there is a meeting, and the meeting needs a story, and the story needs visible motion: features shipped, logos signed, a chart that went up. So the roadmap quietly reorganizes itself around what will be visible by the next meeting. Nobody decides this. It is just what happens when a compounding process reports to a sampling window shorter than its compounding period.

The invisible work loses that internal argument every single quarter, because it never has a slide. And the cost is not zero, it is cumulative. I have watched portfolio companies where the honest engineering answer was "we need a quarter of unglamorous foundation work," and the quarterly story pressure converted it into three more releases nobody remembers. The foundation work still had to happen. It happened a year later, at triple the price, under worse conditions.

Athletes know this tax by name. Racing every month feels productive and photographs well, and it is the most reliable way I know to arrive at the race that matters with nothing underneath you.

Slow to build, fast to lose

The second thing a long block teaches you is the asymmetry. Fitness takes months to build and weeks to lose. Miss a fortnight and you feel it; miss six weeks and you are rebuilding, not resuming. The compounding runs slowly upward and quickly downward.

Everything durable in a company has the same asymmetry. Engineering discipline, data quality, the habit of writing things down, a customer's trust: years to accumulate, one bad quarter of neglect to crack. Which means the popular rhythm of heroic sprint followed by recovery drift is not neutral, it is a slow leak. Each pause costs more than the equivalent continuation would have. The athlete who trains eleven months at a moderate, boring, sustainable load beats the athlete who alternates spectacular quarters with broken ones, every time, and the same is true of teams.

This is why consistency is not a personality trait, it is a strategy. It is the only strategy that respects the asymmetry.

Training age

Coaches have a concept that I think about constantly as an investor: training age. Not how old you are, but how many consecutive years of uninterrupted work your body carries. It is the single best predictor of an athlete's ceiling, better than any recent block, because the deepest adaptations only arrive after years of stacked, unbroken seasons. I started this sport in 2018. The 2022 result was not built in 2022.

Companies have a training age too, and it is one of the first things I try to read now. Not how old the company is, but what it has been doing continuously, without interruption, for more than a year. A team that has run the same weekly release rhythm for three years, kept the same eval discipline through two pivots, held onto its earliest customers through every repositioning: that team has a ceiling the pitch deck cannot show you. A company that has restarted its strategy every two quarters is, whatever its founding date, brand new.

Telling flat-and-loading from flat-and-dead

The obvious objection: sometimes flat really is dead, and patience is just denial with better branding. True. A twelve-month horizon is not a license to stop measuring, it is a reason to measure different things.

In training, the surface numbers can be flat while the inputs tell you everything. Is the same session costing less? Is recovery coming faster? Is the morning resting heart rate trending the right way? If the inputs are improving underneath a flat surface, you are loading. If the inputs are eroding underneath a flat surface, you are stuck, and no amount of horizon extends your way out of it.

Companies have the same second layer. Revenue can be flat while cost per unit falls, cohort retention strengthens, and the sales cycle shortens: loading. Revenue can be flat while all three erode: dead, whatever the founder's conviction says. The quarter cannot tell these two situations apart, because on the surface they are identical. The inputs can always tell them apart. So the discipline is not "wait longer." The discipline is: pick the two or three input metrics that lead your outcome by six to nine months, and judge the flat stretch on those.

What I changed

Three things, concretely.

I judge companies on twelve-month deltas, and I say so out loud at the first meeting, because founders calibrate to what their investors sample. In every review I ask one question that has no quarterly answer: what are you doing right now that will only show up in nine months? A founder with a crisp answer is building something. A founder with no answer is being managed by their own board deck.

I push every team I work with to protect one line of invisible work per quarter, explicitly, with a name on it, so it stops losing the argument by default.

And I hold my own building to the same standard. The agent stack I run today is the compounding result of unglamorous months where the honest status update was "flat, still loading." I would not have survived my own quarterly review either.

Nobody claps in month five. That is the whole test. The quarter tells you what happened. The year tells you what you built. Choose the unit you manage by carefully, because it silently chooses which work survives.