Everyone’s career advice is addition. Learn this. Add that. Stack another skill on the pile and wait for the pile to pay you.
Almost nobody runs the other operation. And the math says the other operation pays first.
Your career is a barrel
In the 1840s, the chemist Justus von Liebig discovered that plant growth is governed not by the total nutrients in the soil, but by the scarcest one. Farmers later drew it as a barrel: staves of different lengths, water rising only as high as the shortest stave. Pour in more of everything else and the level does not move. Lengthen the short stave and the whole barrel holds more.
A century and a half later, Eliyahu Goldratt proved the same law inside factories in The Goal (1984). A production line’s throughput equals the throughput of its slowest step. Speed up any other step and you get more cost, more inventory, and exactly zero extra output. He called it the Theory of Constraints, and it quietly made manufacturers billions.
A career is a chain of steps like any other line: get noticed, win trust, do the work, get paid, get referred. Your income rises only to your shortest stave.
The multiplication nobody does
Skills do not add. They multiply — because every stage of the chain gates the next one. Watch what that does to two improvement plans with the same effort budget.
Say your delivery is excellent (0.9), your selling is decent (0.7), and your follow-up — proposals sent late, invoices later, past clients never re-contacted — is poor (0.3). Your effective output is 0.9 × 0.7 × 0.3 = 0.189. You keep about 19 cents of every dollar your talent creates.
Plan A, the popular one: polish the strength. Delivery goes from 0.9 to 0.95. New output: 0.95 × 0.7 × 0.3 = 0.1995. Up 5.5%.
Plan B, subtraction: fix the bottleneck. Follow-up goes from 0.3 to 0.6 — not great, just adequate. New output: 0.9 × 0.7 × 0.6 = 0.378. Up 100%.
One plan buys a rounding error. The other doubles the business. Addition feels productive because strengths are fun to practice. Subtraction pays because the constraint is where the money is stuck.
Three ways to remove a bottleneck
Fix it to adequate — then stop. The goal at the bottleneck is sufficiency, not excellence. Follow-up at 0.6 beats delivery at 0.97 every single quarter. Perfectionism at the constraint is just addition wearing a disguise.
Delegate it. If someone else can run your worst step at 0.7 while you run it at 0.3, hiring them is not a cost — it is a multiplication. In 2026 the delegate is often software: an AI agent that sends the follow-ups, drafts the proposals and chases the invoices, tirelessly adequate, which is all a bottleneck ever needs to be. Build the check in, though — as we showed in Your AI Tools Have a Sampling Problem, an agent’s first output is rarely its best one. The premise scales, too: over on MentorNet, a whole consultant kit is built on removing the owner as the bottleneck.
Delete it. Some bottlenecks exist only because you offer the thing that creates them. Drop the service line that generates the admin you dread. Fire the client segment that pays slowest. Close the channel that eats your mornings. Deletion is the only fix with a negative cost.
The 10-minute audit
- Write down the five steps your work passes through: attention, trust, delivery, payment, referral.
- Score each one honestly, out of 10. If you flinch, ask a client or a colleague to score it for you.
- Find the lowest number. That is the only one that matters this quarter.
- Pick its removal method: fix, delegate, or delete.
- Re-audit in 90 days. There will be a new shortest stave. There always is. That is the job now.
Where subtraction stops
One honest caveat: subtraction is a phase, not a religion. Once your worst step is adequate, it stops gating you — and the advantage flips back to stacking strengths on top of a clear chain, the strategy we broke down in The $100 Skill Stack. The two strategies alternate for a whole career: subtract until nothing strangles you, add until something does.
Your career is a barrel. It fills to the shortest stave.
Find it, then staff it: the AI Automation Opportunity Finder ($27) maps where your bottlenecks are hiding, and MAX’s Guide to AI Agents ($17) shows how to hand them to an agent.
Sources: Liebig’s law of the minimum (1840s); Goldratt & Cox, The Goal (1984) — the Theory of Constraints. The worked example is illustrative arithmetic, not survey data.
