Every company is, underneath the org chart and the brand, a body of logic. Inputs come in, something transforms them, outputs go out, and a few feedback loops decide whether the whole thing gets better or worse over time. When a company is stuck — slow, expensive, fragile, unable to scale — the instinct is to push harder everywhere at once. Hire more people. Add more process. Rewrite the whole platform. It rarely works, because effort spent where there is no leverage is just heat.
The more useful move is to expose the bare logic of the process and look for the point where a small, precise force produces a large, predictable change. A lever does not multiply your strength by working harder; it multiplies it by being placed correctly relative to the fulcrum. Most of what we do on an engagement, before we write a line of production code, is find that placement.
Consider a bank that wants to grow its loan book without growing headcount in proportion. The naive reading is 'automate lending.' The actual leverage was narrower and far more powerful: a single 22-section memo that bankers assembled by hand for every deal, most of it mechanical assembly of the same facts from the same systems. Automate the assembly, leave the judgment — risk rating, relationship summary — exactly where it was, and the constraint moves without anyone betting the institution on a black box.
Finding the fulcrum is a diagnostic skill, and it is mostly about distinguishing three kinds of thing. Some parts of a system are load-bearing: remove them and it collapses. Some are decorative: they look important and change nothing. And a small number are leverage points: touch them and the whole system shifts. The failure mode of most technology spending is treating decorative work as load-bearing, and never finding the leverage point at all.
The same discipline scales down to a single change. 'Apply force precisely where leverage lives' is as true of a diff as of a system: the maximum number of small, independently verifiable changes will always beat one heroic rewrite, because each one is reversible and you learn where the movement actually is. You earn the right to a big claim by making a series of small, checkable ones.
None of this is a trick you run once. It is the habit — read the system for what it is, not what the diagram says it is; find the knob that moves it; turn that one. The rest is restraint.