The flip side of this is that many of us know the excuses big email services make about how bad a problem spam is and how our content might be spam and we just don't understand... when the content is the kind of reasonable and necessary communications that 'tzs mentioned further up in this thread.

I'm getting awfully bored of bounce messages from apparently credible organisations whose members have signed up, telling me that the legally required and personalised information we have just sent to our customer appears to be unsolicited bulk email. There has been a small wave of these recently, which leads me to think that someone screwed up a scoring system that multiple organisations are using to vet incoming traffic.

In any case, that's not a minor mistake, that's an automatic recognition system that is hopelessly inaccurate. It could even be dangerous, for example if a customer isn't getting notifications about payments and forgets they haven't cancelled a subscription, or if some action is required of them if they don't want to accept some change in terms for a service they use.