This thread hits home. What finally reduced my panic wasn’t better alerts — it was changing *how* I look at failures.
Two things helped me:
1. Instead of just “workflow failed,” I started tagging every failure by *cause* — was it the client’s expired credential, an API change on their side, or my own flow logic? Once I could see “this one’s on them, this one’s on me” at a glance, the finger-pointing conversations basically stopped. The client can’t argue when it’s clearly their expired Google login.
2. I gave each client a simple status view — just “running / needs your attention” — so they stop emailing me to ask if things are working. Weirdly, showing them a plain status page made them trust me *more*, not less.
Curious if anyone else separates failures by “whose fault” like this, or if you just lump them all together?