I like the distinction between visible failures and unnoticed changes. That explains why drift is so difficult to manage.

I also like your point about a single source of truth. In my experience, many teams naturally document code, but configuration often evolves through the admin interface without the same discipline. Once those changes stop being visible, keeping sites aligned becomes much harder.

I like your observation that the hardest problems become operational rather than technical as the number of sites grows.

I've found something similar on a smaller scale. It's rarely the updates themselves that cause issues. More often it's the small differences that accumulate over time. Most of the time it isn't a big change. It's a handful of small decisions made over weeks or months that slowly push sites in different directions.

Those little differences are usually harmless on their own, but across dozens of sites they become difficult to track and reason about.

Out of curiosity, have you found configuration drift or unexpected admin changes to be a bigger source of inconsistency than the automation itself?