I’d be interested in any good statistics.
The bottom line is that email is remarkably reliable, despite being a service that is “unreliable” (as defined in data communications). Part of the effectiveness is that, though it’s unreliable, its’ built by concatenating a number uses of a reliable service (TCP).
In data communications, “reliable” means that if it succeeds, you’ll know it. If you don’t know it succeeded, it might have gotten through or it might not. Another term for this is “a confirmed service.” When any mail server talks to another mail server, it uses TCP, which is a confirmed service. If the server doesn’t get the “it worked” confirmation, it retries. The next server does the same, etc. However, if any of the servers in the chain fouls up (by getting an email but somehow losing it), the email gets lost with no notification.
The only time you get a notification is when a server tries but is unable to deliver, e.g., because the recipient is unknown, or the host in that direction isn’t responding. Of course, the notification is sent via email, and THAT could get lost.
Failures on email servers to forward email should be rare, but of course there are bugs and program crashes and system crashes etc. So, every now and then, email fails to make it.
That’s in addition to the spam filters and other ways where emails could get deleted more or less on purpose.
If a given mail server is particularly unreliable (crashes a lot), then anyone who needs that server will get less reliable service. Hopefully, though, someone in IT would notice and fix it.