When a product comes off spec, it is rarely because someone made a mistake. It is more often because each team made a perfectly reasonable decision based on the information available to them. And the information available to them was incomplete.
The formulator works from ingredient specifications that represent a statistical average across many deliveries. That is a reasonable approach when real-time lot data is not available. But ingredients vary. Moisture levels shift between deliveries. Protein concentrations fluctuate. When the actual batch on the line differs from the assumed value in the formulation database, that gap does not announce itself. It shows up quietly; in moisture readings at the end of the run, in protein figures that come back slightly low, in a right-first-time failure that triggers a review.
The operator, meanwhile, is working with considerable skill and experience. They know their line intimately. They are reading it continuously, adjusting process parameters in real time based on what they see and what they have learned over years. What they do not have is visibility into the upstream assumptions: what moisture the formulator was expecting, what protein level the recipe was built around, where the tolerances sit. That context would make their adjustments more targeted. Without it, even the most experienced operator is compensating without the full picture.
Quality control brings analytical rigour and a clear standard to measure against. Their role is to confirm the product before it ships. And they do that well. The structural limitation is timing: quality data becomes available after the batch has run. That means findings can prevent the next problem, but not the current one.