Leverage
Output per unit of input. Change the multiplier, not the effort behind it.
The Model
A lever converts a small force into a large one by trading distance for force. Archimedes proved the ratio and then made the boast everyone remembers about moving the earth given somewhere to stand.
Everything since called leverage is that same trade. The same hour goes in, a larger result comes out, because something else is carrying the load.
Four forms, and the distinction between them decides your strategy.
- Labour — other people do the work.
- Capital — money does the work.
- Code — a program does the work, repeatedly, at no marginal cost.
- Media — a piece of content does the work while you are asleep, for as long as it stays findable.
The first two require permission. Somebody has to agree to work for you, or agree to fund you.
The last two do not, and they replicate at zero cost. That asymmetry is the entire story of the last two decades of business formation.
The fulcrum matters as much as the force. A lever placed under the wrong thing moves the wrong thing faster.
Why operators get this wrong
Operators apply the multiplier before they have validated what they are multiplying.
The clearest case runs weekly in every founder-led company. Outbound gets automated before a single message has been proven by hand — so instead of forty poorly targeted emails a week, four thousand go out.
The domain reputation is damaged, the list is burned, and the one thing the manual version provided, which was reading replies and rewriting on the evidence, is gone. The multiplier worked perfectly. That was the problem.
The second failure is confusing motion with leverage. Coordinating eight contractors on low-value work is labour leverage aimed at the trivial many: you now carry the management overhead and receive no multiplier, because the underlying task never justified one.
The third is the most common among competent people. They accumulate leverage they never pull.
The list exists and is not mailed. The process is documented and not delegated. The price has not been raised in three years.
Available leverage that is not applied is indistinguishable from not having it.
Applied
Body
In training, the lever is the movement that produces the most adaptation per unit of recovery — because recovery, not time and not motivation, is the scarce input.
That is a short list: squat, hinge, press, pull, carry. Five patterns loaded properly and progressed over years beat twenty exercises rotated monthly, and it is not close.
The mechanism is not mystical. Recovery capacity is finite and roughly fixed. Spending it on movements that recruit large amounts of tissue returns more adaptation per unit than spending it on movements that recruit small amounts.
The second lever sits outside the gym and is larger than anything inside it. Sleep improves every other input simultaneously — training quality, appetite regulation, glucose handling, mood, decision-making at four in the afternoon.
It costs nothing except the decision to stop working at half past ten. Which is why it is the most under-pulled lever available to any operator reading this.
Business
Price is the highest-return lever in most small businesses and the least used, because it is the one that feels dangerous.
The arithmetic is not close. A ten percent price rise with no volume loss flows almost entirely to profit. A ten percent volume rise brings its own delivery cost, its own support burden and its own hiring conversation.
At a thirty percent net margin, ten points of price is a third more profit. Nothing else on the operating plan does that in an afternoon.
The second lever is distribution over production. One partner who already holds the attention of the people you want is worth more than the year of content you would need to assemble that audience yourself.
The question to ask is not how do I build an audience. It is who already has one, and what do they need that I can make.
The third is documentation, and it is the one that converts labour into something closer to code. A process written down once can be executed by somebody else forever. A process that lives in your head requires you, permanently, and quietly caps the size of the company at the size of your week.
AI Leverage
AI is code leverage without the engineering prerequisite. That has not previously existed — the multiplier used to require capital or years of training, and now it requires neither.
But it multiplies judgment, and judgment is the one input it does not supply. Point it at a good decision process and it compresses weeks into hours. Point it at a bad one and it produces the wrong answer at industrial volume, formatted beautifully.
The highest-return use is not writing. It is collapsing the gap between a question and an answer on decisions you defer for lack of time: the pricing sensitivity you keep meaning to model, the six scenarios for the hire, the forty-page supplier contract you skim to clause four.
Those decisions were postponed because analysis was expensive. It is no longer expensive, and most operators have not updated the list of things they consider worth analysing.
The lowest-return use is producing more of whatever nobody was reading at the old volume.
The Drill
Open last week's calendar. Every block gets a number.
One: done once, benefits once. A support reply, a status call, an invoice.
Ten: produces something reusable. A written process, a template, a piece that stays findable, a recorded training.
One hundred: changes what everything else is worth. A price change, a hire, a positioning decision, a partnership, firing a client who was consuming the capacity.
Most weeks come back almost entirely ones, and the honest reaction to that is uncomfortable rather than motivating. Sit with it for a minute before you move on.
Then take one hundred-level item and place it at nine on Tuesday morning. One. Defend it the way you would defend a call with your largest customer, because it is worth more.
Repeat the audit in four weeks and check whether the shape of the week changed. If it did not, the block was not defended and the calendar is telling you the truth about your priorities.
Stoic parallel
Two Stoics sit at opposite ends of every measure of external leverage, and they reached the same conclusion.
Epictetus was born a slave. He had no labour, no capital, no property, no audience and, for part of his life, no legal standing. His teaching was that the only lever always available is the one applied to your own assent — what you agree to believe about what has happened to you.
Marcus Aurelius commanded the largest concentration of external leverage available to any human being alive. Armies, treasuries, the machinery of an empire.
He read Epictetus constantly and wrote about the same lever. Not because the armies were useless, but because he had noticed that external leverage multiplies whatever is already inside the operator, including the errors.
That is the warning worth keeping. A multiplier applied to confusion produces more confusion, at scale, with better formatting. Fix what is being multiplied first.
One model per week.
Applied to training, business, and AI leverage. No fluff.
Related models
- Pareto PrincipleResults are not spread evenly across inputs. Measure the distribution before you decide where the effort goes.
- CompoundingReturns applied to returns. Boring for long enough that most people quit before the curve arrives.
- Circle of CompetenceThe size of the circle does not matter. Knowing exactly where its edge runs is the whole discipline.
Origin: Archimedes for the mechanics of the lever; the four-form taxonomy of labour, capital, code and media is Naval Ravikant's framing