By putting enough time and money aside to find what should be saved, and to fix it. Here is my example, I have recently finished a project where while under de

By putting enough time and money aside to find what should be saved, and to fix it. Here is my example, I have recently finished a project where while under development we were running instances at around 1.5x then we needed. We knew they were oversized, but not by how much because we were more focused on getting features done then making it efficient. When we were "done" we took a sprint to figure out what the "right" size needed to be, by measuring it. This means over the 6 months we "wasted" $2-3k, but you know what would have cost 25-30k? If we were running underpowered machines, that uncovered an intermittent bug that added a month to our development schedule. Remember the goal is to be frugal not stupid. So before you try and do ANY cost cutting activity you need to go figure out what the impact of it might be, make sure you have some way to measure that impact, and then only do it if the cost of implementing is less then the savings.