Why optimising everything makes you slower

It feels counterintuitive that working on more parts of your funnel could make you grow slower, so it is worth slowing down on the mechanism, because once you see it you cannot unsee it.

The sum of local optima is not the global optimum

Imagine you spend a fortnight making your landing page convert ten per cent better. Genuine improvement, real work, measurable win. But if the constraint on your business right now is that qualified leads sit for three hours before anyone responds, those extra converted visitors simply join a longer queue at the stage that is actually broken. You produced more input for a step that was already the limit. Net new customers: zero. Worse than zero, in fact, because you have now loaded the constraint further and possibly degraded it.

This is the throughput-accounting truth that trips up almost everyone: optimising each part of a system individually produces a globally under-optimised whole. Throughput accounting under the Theory of Constraints measures the system by what comes out the end, not by how busy each part looks, and only the capacity of the bottleneck changes total output. A funnel where every single stage is 'being improved' grows more slowly than one where only the bottleneck is worked, because the first funnel is burning effort on links that cannot move the result while the constraint sits there untouched.

Busyness is the tell

There is a sharp diagnostic here. A company where everyone is busy a hundred per cent of the time is terribly inefficient, not impressively productive. Full utilisation everywhere means effort is being poured into non-constraints, which by definition cannot lift output. If your week is a blur of activity across every channel and the growth number is flat, that is not a sign you need to work harder. It is the symptom of working everywhere instead of working the constraint.

For a solo operator this is liberating

Reframe it and the whole thing becomes a relief rather than a burden. One bottleneck means you get to ignore most of your backlog this cycle, guilt-free. The fifty experiments you could run collapse into the one that matters. You stop apologising to yourself for not getting to everything, because getting to everything was never the goal and was actively slowing you down. The skill is not doing more. It is having the nerve to leave nine good ideas on the shelf while you break the one stage that is holding the whole machine back. This is the same discipline that makes channel selection work: one constraint, one channel, one lever, then move on.