It's important to note that it's not just about the financial cost of keeping a service running. For example, I had one service that I built over several months

It's important to note that it's not just about the financial cost of keeping a service running. For example, I had one service that I built over several months. It got some attention, but ultimately fizzled out. It comfortably lived on some EC2 instances I already had running, sharing resources for the project I had shifted into. However, the support requests came in from the handful of users, and while I wasn't able to make it a financial success, it would've been more of a disservice to those folks to provide a partially functioning, unreliable service rather than give them the chance to migrate away to a competitor that could better serve their needs.