Check the Platform, Not the Dashboard

Every publishing tool has a dashboard that turns green when a post goes out. It is reassuring, and it is often only telling you that the tool did its part. Whether the post actually exists on the platform, on the right account, visible to the audience, is a separate question.

Where silent failures hide

Posts fail in ways a dashboard does not always catch. An access token quietly expires. An upload is accepted and then rejected during processing. A post goes to the wrong account because two accounts share a similar name. A collaboration invitation is never accepted, so the post only reaches half the audience it was meant to.

In each case the tool reports success, because from its point of view the request was sent. The failure only exists on the platform.

So we check the platform

Every evening an automated check goes through everything our systems say they published that day and confirms each post on the platform itself: that it exists, on the correct account, publicly visible. Anything missing, stuck or still waiting on approval is listed in one message the team sees before the end of the day.

It takes no human time on a normal day and catches problems while they are still hours old, not days.

Apply the same rule to alerts

The same principle applies to warnings. An alert that something failed, raised from a database record alone, is sometimes wrong the other way: the post is fine and the record is stale. Before anyone acts on an alert, it should be confirmed on the platform too. Otherwise the team learns to ignore alerts, which is worse than having none.

A dashboard is a claim. The platform is the evidence.