Every intake system we have been asked to look at measures the same thing: what it processed. Messages read, records filed, accuracy against a labeled set. All of it real, all of it useful, and all of it blind to the one failure that actually costs money.
The record that never arrived generates no event. There is no message to score, no extraction to check, no row to reconcile. It is invisible to a system whose entire model of the world is built from things that showed up. It becomes visible on filing day, when someone notices a gap and starts making phone calls.
The fix is not clever, but it does have to be designed in rather than added later. The system has to hold the expectation, not only the receipt. It has to know which assignments owe a record for which period, independently of whether anything has arrived for them. Once that exists, a missing record is a state the system can be in, and states can be surfaced.
In practice this means the queue reads differently. Instead of a list of things that came in, it is a short list of decisions: this period is empty and filing day is in six days — chase it, or record why it is legitimately empty. The second option matters as much as the first. An empty period with a documented reason is a closed item; an empty period with no reason is an open risk, and they should not look the same.
The general point is that an operational system is defined as much by the events it can represent as by the ones it processes well. If absence cannot be represented, absence cannot be managed, and it will keep being discovered by a person on a deadline.