My newsletter job never trusts the platform's word that it worked

Every Tuesday morning my newsletter goes out without me doing anything, and the part of that I trust the most is the one piece of it that does not just believe what it is told.

When an approved issue is ready, a scheduled job creates the email campaign and tells MailerLite, the service that actually sends it, to fire it at 9am Eastern the next Tuesday. MailerLite answers with an “OK, scheduled” response. Most jobs would stop there and move on. This one does not.

Instead it asks for the same campaign back and checks, on its own, whether the time MailerLite actually recorded matches the time it asked for. A busy service can accept a request and still schedule it wrong, or not at all, and an “OK” response only means the request was received, not that the result is correct. If the times do not match, the job does not shrug and hope. It cancels the campaign itself, deletes it, and fails the run loudly, on purpose, rather than letting a newsletter sit scheduled for the wrong moment with nobody watching for it.

There is one more check layered under that one: before creating anything, the job asks MailerLite what is already waiting to go out, and refuses to schedule a second campaign into a slot something else already holds. Two campaigns racing for the same Tuesday morning is a worse failure than one newsletter going out a day late, so the job is built to catch the first and tolerate the second.

None of this runs because I am watching on Tuesday morning. It runs because the job was built to not take anyone’s word for it, including the service it depends on, and to say so clearly the moment what actually happened stops matching what it asked for.