Bottlenecks Govern Flow

You improved something measurably. Why did nothing get faster?

The idea

Improving non-bottlenecks rarely improves total throughput.

In the real world

More leads do not help when onboarding is full.

Going deeper

Throughput is set at the constraint, so improvements anywhere else accumulate as queues rather than output. This is why a 50% increase in leads with onboarding at capacity produces a longer waiting list and no additional revenue.

The corollary is uncomfortable: capacity above the bottleneck is not useful and effort spent creating it is largely wasted. Finding the constraint is usually straightforward — it is where work piles up in front and starves behind — and the discipline is to keep improving that one place until it moves somewhere else, then follow it.

Where it stops applying

Bottlenecks move, sometimes quickly, and optimising hard for the current one can create the next. Systems with variable demand may also have different constraints at different times.

Why it matters

Only one part of a system sets throughput, and effort spent elsewhere produces local gains and no total change.

Try this today

Find the step where work waits longest.

Test yourself

Lead volume rises 50%. Revenue does not move. Onboarding is at full capacity. What did the extra leads produce?

Show the answer

A longer queue. Throughput is set at the constraint, so improvements upstream accumulate as waiting rather than output. Until onboarding capacity rises, more leads add cost and delay without adding revenue.

Learn this in the feed Answering from memory, then again days later, is what makes it stick.

More in Product & Systems Thinking

Queues Reveal Capacity Local Optimisation Hurts Scope Grows Unless Bounded Where to Push a System Solve a Real Job Design for Failure

All Product & Systems Thinking lessons