MRP won't start without clean data (and other shop-floor truths)
Everyone wants automated planning. Few want the boring work that makes it trustworthy. Notes on getting an operation ready so MRP is actually useful.
Every so often the same request lands: "we want the system to plan on its own." And almost always the operation isn't ready for it yet, not for lack of technology, but for lack of discipline in the data feeding the model.
MRP amplifies, it doesn't correct
An MRP takes your assumptions and multiplies them across thousands of lines. If the lead time is wrong, the error doesn't sit still: it propagates through the whole chain. Automated planning doesn't forgive a loose data master. It exposes it.
The boring list that actually matters
- Real lead times, not the ones "we've always used".
- Safety stock with judgment, sized by variability, not by fear.
- BOMs and routings that reflect how you produce today, not how you produced in 2019.
- Lot policies consistent with real purchasing and production capacity.
None of this is glamorous. All of it is the difference between an MRP the planning team uses and one they switch off within a week because they don't believe it.
Start with the diagnostic
Before touching parameters, I measure: inventory accuracy, supplier reliability, demand variability. That diagnostic tells you whether the problem is configuration or process, and it keeps you from automating a mess.
Automating a broken process just gives you a faster mess.
Fifteen years in operations taught me to distrust the elegant solution that ignores the plant. Supply chains get fixed on the floor, with data the floor recognizes as true.