Only one bottleneck matters
posted on February 16, 2025In any given process, there is a single bottleneck, and that bottleneck determines the throughput of the entire process. In a world of limited time and money, it is this bottleneck alone that must be focused on and eliminated in order to improve this process.1
This is fundamentally how incremental improvements are made, and is the engine of progress which takes an invention from 1 to 100. It's easy to forget this and focus on all the many things that can be seen and improved (often more easily than the bottleneck), but this is an error2.
To be sure, stepwise changes (e.g. new inventions) live outside of this rule, and incrementalism is the pathological outcome of focusing on incremental improvement over paradigm shifts. It's worthy to ask "is this process worth improving, or should we end it and consider starting a new process"? One could argue that this is but another instantiation of the single-bottleneck rule, in which a process is in fact part of a greater process, and so on.
-
For a more in-depth treatment, I highly recommend reading The Goal by Eliyahu Goldratt. A friend characterized it as "one of the worst novels, but best business book ever written". I often apply the lessons from this book in my work and life, which is the best recommendation I can give. A version of The Goal in an IT/tech context is The Phoenix Project, which I'd recommend as well. ↩
-
When you can parallelize this work, it often makes sense to work on what you presume will be the bottleneck after the current bottleneck is eliminated. There are times where the bottleneck is truly unassailable, and this can be a temporary or permanent state. This is rare. ↩