Supplier Quality Scorecard: Score Every Supplier, Approve the Grade, Send It
Turn receipts, rejects, on-time delivery, and NCR/SCAR counts into a real supplier scorecard — defect rate in PPM, on-time-delivery %, and a weighted score and letter grade per supplier per period — that a quality engineer reviews and approves before it ever reaches a supplier or a sourcing decision.
A logged-in web app that imports your receipt/reject, delivery, and NCR/SCAR data per supplier per period, computes defect rate (PPM), on-time-delivery %, and a weighted score and letter grade for each supplier, trends it across periods, flags suppliers crossing your action threshold — and holds every scorecard in a DRAFT state until your quality engineer reviews and approves it, after which it emails the scorecards via Resend and exports a clean CSV.
Before you start
- A Supabase account (free)
- A Vercel account (free)
- A Resend account (free)
- Your receipt/reject data, on-time delivery data, and NCR/SCAR counts per supplier (CSV or Google Sheet)
- Your scoring weights and grade thresholds (even a rough draft)
- Claude Code or any AI coding agent
The problem this kills
Supplier review season rolls around and the scorecard gets built the way it always does: one engineer, three exported spreadsheets, a half-day of VLOOKUPs, and a "score" that's really a gut feel dressed up in a percentage. Defect rate is counted one way this quarter and another way next quarter. On-time delivery depends on whose calendar you used. The grade that lands in front of a supplier — the one that decides whether they stay on the AVL or get put on probation — rests on numbers nobody can fully reconstruct three months later.
That's a problem in two directions. A genuinely bad supplier argues the math and wins, because the math was never airtight. And a good supplier gets dinged by a data error nobody caught. Either way, sourcing decisions worth real money are being made on a number that can't be defended. You don't need a six-figure supplier-quality module bolted onto your ERP to fix this. You can build the scorecard yourself — defined once, computed the same way every period, and signed off by a named person before anyone sees it.
What you'll build
An internal web app your quality team logs into. You import the raw inputs per supplier per period — receipts and rejects (so the tool can compute defect rate in PPM, parts-per-million), on-time delivery data (to compute OTD %), and NCR/SCAR counts (nonconformance reports and the corrective-action requests you issued, plus how fast suppliers responded). The tool computes each metric, applies your weighting to produce a single score, and assigns a letter grade (A/B/C/D, or whatever your bands are). It trends each supplier across periods so you can see who's improving and who's sliding, and it flags any supplier that crosses your action threshold so they don't slip through. Crucially, every scorecard sits in DRAFT until your supplier-quality engineer reviews the computed numbers and approves them. Only an approved scorecard can be emailed to a supplier or exported for a sourcing review.
What's inside the Implementation Plan
The downloadable plan is a step-by-step file you paste into an AI coding agent. It opens by interviewing you about how your supplier quality program actually works — your real supplier names and codes, your period (monthly, quarterly), exactly how you define a "reject" and compute PPM, what counts as on-time delivery and against which date, how you count and weight NCRs and SCAR responsiveness, the weighting that rolls those into one score, and the grade bands and action thresholds you use. It reads a short spec back to you for a thumbs-up, then builds the scorecard around your definitions instead of a generic template — because the whole value of a scorecard is that everyone agrees on how it was computed. From there it walks the agent through the data model, the CSV/Sheet import, the metric and score computation, the trend view, the approval gate, the supplier email, and the CSV export. Every step ends with a ready-to-copy prompt.
The governance it includes (this is the point)
A scorecard that decides who keeps your business has to be defensible, so the plan builds the controls in from the start: login so only your quality and purchasing team can use it, row-level security so a site or division only ever sees its own suppliers and scores, and a complete audit trail of every import, computation, edit, and approval — who did what, and when. The centerpiece is a hard human-in-the-loop approval gate: the tool drafts every scorecard, your supplier-quality engineer reviews the numbers and the inputs behind them, and only an approved scorecard can be sent to a supplier or used in sourcing — nothing reaches a supplier on autopilot. And a duplicate guard keyed on supplier + period means re-importing this month's data can't quietly create a second, conflicting scorecard for the same supplier and period.
Who it's for
Supplier-quality engineers, purchasing and commodity managers, and quality managers who run periodic supplier reviews and are tired of rebuilding the scorecard by hand — and of defending numbers they can't fully trace. If you can describe how you define a defect, what "on time" means, and how you'd weight delivery against quality, you can build this.
You've got this — open the plan, paste the first prompt, and you'll be scoring your first suppliers from real data this afternoon.