Some mornings I pick up the dumbbells before I’ve fully decided to. Not because I’m disciplined. Because they’re on the floor in front of the wardrobe and avoiding them requires more effort than using them. That’s the whole system. Fifteen minutes, four exercises, done before coffee goes cold.

I’ve been at 92 kilograms for a while. Target is 85. The gap is real and I’m not close to closing it yet. But I’m also not treating it like a crisis that needs a heroic intervention. I’ve done that before. It doesn’t work. What I’m doing instead is building something more boring: a small, trackable routine, daily inputs I can actually verify, and a log that tells me whether I’m actually doing it or just believing I am.

That last part is where the rule actually breaks.

The 1% rule is one of my actual operating principles. Not a framework I stumbled on last month. “Compounding beats intensity” is how I think about most things: writing, fitness, systems work. But there’s a version of this idea that floats around productivity writing and I’ve watched it fail people, including earlier versions of me, in a predictable way. The failure isn’t with the math. The failure is with the honesty.

Improvement without a record is just a feeling. That’s a note I wrote to myself a while back, and it’s turned out to be more useful than anything I’ve read about habit formation. The compounding math is real, but it only works if the base is accurate. You need to know that the inputs are actually consistent, not that they feel consistent. Feeling like you’ve been showing up and actually showing up are different things, and the gap between them is where most 1%-better plans fall apart.

This is what a log does. Not accountability, exactly. Accountability implies someone is watching. The log is evidence. It’s proof that the system produced real outputs, not just good intentions on the days you remembered to think about it.

Dave Brailsford built something like this into Team Sky’s coaching operation. He didn’t try to make his riders 50% better cyclists overnight. He found dozens of small, measurable sub-components, improved each one by 1%, and let them compound. Sleep quality, bike aerodynamics, hand-washing protocol before races. None of it was dramatic. Combined, it turned a team that had never won the Tour de France into one that started winning it regularly. What made it work wasn’t the insight that small improvements compound. It was the discipline of making everything specific enough to measure.

That’s the part that the productivity version of this idea usually skips.

You cannot measure a 1% improvement in “becoming a better leader” or “getting healthier.” You can measure whether you did the fifteen-minute session. You can measure whether you slept seven hours. You can measure body weight on a specific morning. When you strip away the abstractions and get to things you can actually record, the compounding math starts to apply. Until then you’re just reassuring yourself with arithmetic.

The other thing the compounding math gets wrong is the timeline. Philippa Lally’s research on habit formation puts the median closer to 66 days before a behavior becomes automatic, not 21. Some behaviors take longer than six months. My fifteen-minute dumbbell sessions are new. I don’t know yet whether they stick. What I do know is that counting on them to feel natural by week three is a way to set yourself up for disappointment at week four.

BJ Fogg’s work adds something useful here: the behavior needs an emotional payoff to encode. Without the reward, the pathway doesn’t strengthen. The dumbbell session works partly because it’s over fast and I feel decent afterward. That’s not nothing. Designing for the feeling after, not just the discipline during, is how you get behaviors to compound rather than requiring fresh willpower every time.

Habit stacking handles the cue side. Linking a new behavior to an existing one gives you a reliable trigger without relying on motivation. The dumbbells are there because I built the cue into the physical space. I didn’t leave it to memory.

The place where the 1% rule genuinely breaks is when people measure things that can’t be measured, or measure too many things at once. If you’re trying to compound across five domains simultaneously, you’re not compounding, you’re diluting. You get a fraction of a percent in five areas instead of 1% in one. The math works against you.

The other break point is the plateau. Progress eventually becomes invisible even when you’re still improving. The compounding curve doesn’t move in a straight line anyway, because you don’t. Sleep variance, stress, recovery lag, a week where work takes everything. The math assumes a consistent base and human biology doesn’t provide one. Most people hit the plateau, see a flat line, and quit, when what they’re actually seeing is normal variance in a system that’s still working. The only defense I know is to narrow what you’re tracking until the signal comes back. If the scale isn’t moving, track sleep and sessions instead. If the inputs are consistent, the output usually follows.

I haven’t had my first Dex session yet. The programming is still being built. The 92-to-85 project is real and currently more architecture than execution. What I know from running systems like this before is that the gap between having the framework and having the habit is larger than it looks from the outside. The log will tell me whether I’m closing it.