Build a Supplier Risk Register When Three of Your Data Fields Are Lying

---

Your ERP has a column called "Source." For one of your nine suppliers it says "Dual" — two suppliers, shared risk, nothing to worry about. It has said "Dual" for eight months. The backup contract that made it true expired last October and nobody renewed it, and nobody changed the field. On paper you have two freight suppliers. In reality you have one.

That single stale word is the whole problem with the job you were just handed. Your new VP of Supply Chain wants a prioritized supplier risk register for the Precision Assemblies category — which suppliers could actually hurt Ironforge if something breaks — and she wants it in two weeks. You have two exports to work from: a 12-month ERP spend extract and a supplier master dump. Nine suppliers, about forty rows of data between the two files.

Here is the trap: a risk register is only as trustworthy as the fields you feed it, and at least three fields in these two files are stale, miscoded, or flatly contradicted by a note two columns over. Feed them straight into any tool — a pivot table, a dashboard, or an AI — and you get a register that looks authoritative and points your VP at the wrong suppliers.

This lesson teaches the two-step fix: feed the AI clean, and make it report like an analyst. First you learn where the data lies. Then you write a six-dimension scoring brief that forces the AI to cross-check every field before ranking. Then you compress the register into one page your VP can act on.

---