A situation comes up more often than you'd think. You ship a release that answers twelve requests. One of your customers asked for three of them, voted on two more, and follows a sixth. How many emails should they get?
The naive answer is six, one per post. The right answer is one.
Six emails is spam, even when every one is relevant
Each message on its own is reasonable: "a request you follow has shipped". Six of them, arriving in the same minute, read as noise. The customer skims the first, ignores the rest, and learns that your emails come in bursts that can be skipped. The next time something they really care about ships, they skip that too.
One email that knows what they asked for
The better message is a single email about the release, written for that person: here's what shipped, and here are the requests of yours it answers. It's more useful than six separate notes, because they can see the whole picture at once.
It matters for your sending, too
Email has a cost, and not only money. Sending lots of messages that people don't open hurts how mailbox providers treat your mail. Over time, more of it lands in junk, including the sign-in links and invitations people actually need. Sending fewer, better emails protects every message you send.
And it has to be one, even on a retry
There's a technical half to this. Sending systems retry when something fails, and a careless retry sends the same email twice. The way to prevent it is to key each email on the person and the event, so a second attempt sees the first already went and stops.
That sounds like plumbing, and it is. But a customer who gets the same release email twice notices, and wonders what else is sloppy.
The principle
Write to people, not to posts. A person asked for things, follows things and cares about things, and every message should be built around that person rather than around your data model. One email per person per release is just that principle applied to a changelog.