This website uses cookies

Read our Privacy policy and Terms of use for more information.

A plant controller in the Midwest opened Business Central to close out March and found a purchase order sitting in approval status since November. The vendor had shipped anyway, on the strength of a decade-long relationship and a promise that payment terms would sort themselves out later. No one in the company had flagged the PO. No one owned the flag. The approval workflow existed in the system, the way most controls exist in most ERPs: as a configuration, waiting for a human to notice it needed attention. The gap between configured and enforced ran four months long, and that gap is where a chunk of working capital had gone to sleep.

The Stall Nobody Owns

Approval stalls rarely come from a broken process. They come from a process built for the normal case, with no owner assigned to the exception. An approver goes on leave and forwards nothing. A threshold changes and the old routing rule stays. The PO sits in a queue that only surfaces if someone opens the right report and reads it line by line, which someone did, eventually, four months later.

This is decision latency in its plainest form: the distance between the moment a decision becomes visible inside the system and the moment a person acts on it. The ERP recorded the PO correctly from day one. The record was never the problem. The silence between record and action was.

Manufacturers and distributors tend to discover this pattern during month-end close, during an audit, or during a supplier escalation, which is the worst possible time to discover it. By then the cash is already committed, the vendor relationship is already strained, and the fix is a scramble instead of a routine.

Where the Four Months Actually Go

In composite scenarios built from the pattern we see across mid-market Business Central shops, the timeline breaks into three stretches, and each one is invisible unless someone is looking for it.

The first stretch is detection. The PO crosses its approval deadline and the system knows it, in the sense that the data exists in a table. Nothing surfaces that fact to a person. The second stretch is ownership. Even if someone stumbles onto the aged PO, there is no named individual whose job includes clearing it, so it gets forwarded, discussed, and set aside again. The third stretch is the absence of a deadline with teeth. A report that says "overdue" carries no weight unless a person is accountable for closing it inside a bounded window, and most BC environments have no such window built in.

Stack those three gaps and four months stops being surprising. It becomes the expected outcome of a system that records everything and enforces very little.

Governed Execution, Not Automation

The fix is not a smarter dashboard. Dashboards report; they do not close gaps by themselves, and a fourth screen nobody checks just adds a fourth thing to ignore. The fix is a sequence: Signal, Route, Approve, Execute, Audit. A risk gets detected inside the ERP, it routes to a named owner where that owner already works, a human approves the action, the system executes it back into BC, and the whole chain leaves a record.

This is the architecture behind OpsGrid, our decision infrastructure product for Dynamics 365 Business Central. It watches for exactly this kind of stall, among other signals like dead stock and supplier reliability drift, and sends the flagged PO to a named owner inside Microsoft Teams with the context needed to decide. Nothing writes back to BC without that person's explicit approval. That constraint is architectural, not a setting someone could switch off under deadline pressure. The company keeps the judgment call. The system keeps the clock honest and the trail complete.

The Usable Idea

Every unresolved decision inside your ERP has an age. Most companies can tell you the age of their inventory and the age of their receivables. Few can tell you the age of a decision sitting in an approval queue right now, and that number, once you can see it, tends to be the most expensive one in the building.

If you want to know where those stalls are hiding in your own environment, the diagnostic takes about three minutes and scores your setup without asking for an email first: intelliconnectq.com/diagnostic?src=nl