Let’s say you have a machine with twenty distinct parts. The machine lasts about ten years before it breaks. If you pick one part and make that part a hundred times more durable, do you know how long the machine will last now?
About ten years.
See, if a machine has twenty parts, making just one of them better isn’t going to radically change the life of the whole machine – unless, that is, the part you’re fixing was significantly more deficient than average. If the rest of the parts in the machine all last twenty years, then yeah – you’ve got one weak link. Improving it will bring up the floor.
So how do you find that weak link? How do you isolate the part that’s so bad that it’s bringing down the lifespan of the whole system?
You’ve got to break that system. Twice.
Take the whole machine and – figuratively – drop it off the roof. See which part breaks. Then do it again. If the same part is what broke each time, there’s a very good chance it’s a weaker link than the rest. If a different part breaks each time, then it’s more likely that every part is roughly the same in terms of durability.
Stress-tests are good for any system. Don’t just randomly pick a part and improve it because you can – you might be wasting time if the rest of the system won’t last any longer as a result.