I worked for a well known cloud BigCo. We'd have "widget on the website is down because something expired" or "data propagation is delayed on X systems" type o

I worked for a well known cloud BigCo. We'd have "widget on the website is down because something expired" or "data propagation is delayed on X systems" type outages pretty frequently. These are the kind of outages that you'd think was your browser and you'd hit f5 and it'd work but we'd notice them internally because we had monitoring. We had outages that would be noticed more than a hundred people worldwide (e.g. major widget down, service degraded enough for us to get support tickets about it) every three quarters or so. So either they did have a lot of outages or their definition of outage included every instance where something was wrong but only customers doing A/B testing would notice sort of way.