64% is the savings when comparing the cost of running the instances before and after scheduling. This directly correlates to the hours we dictated the machine

64% is the savings when comparing the cost of running the instances before and after scheduling. This directly correlates to the hours we dictated the machine to run (instances "on" about 36% of the time). This is also a very rudimentary example without dynamic scaling. So getting down to comparative costs, I wrote the script in about 90 mins while watching a TV program. Coming up with the numbers to compare before and after was 10 minutes at most. While neither of these incurred no loss of regular schedule time during the day (done at home on my time), you could account my hours at home by my some hourly rate. Even at very generous consulting rates, the ROI for these 100 minutes was easily realized in the first week that the script was in place. We were able to save $183 per week with this solution and that only touched 6 of our instances. We still have 1-2 orders of magnitude of instances to apply this to. This was identified as a quick and easy change and had the ROI been something months ahead, we would've addressed the issue in myriad other ways. These machines were already provably not being used during specific hours of the day, so uptime wasn't so much surrendered as the instances are set to be running when needed stopped when unused.