using-ops-methodlisted
Install: claude install-skill gulmezeren2-byte/industrial-engineering-ai-skills
# Using the Ops Method
Operations analyses fail in predictable ways: answering a question nobody asked, computing on data nobody validated, reporting a metric nobody defined, and ending without a decision. The method below exists to make those failures impossible by construction. **Follow the sequence; do not skip steps because the task looks small.** Small tasks skip the brief; wrong answers follow.
## The sequence
| Step | Question it answers | Use | Artifact |
|------|--------------------|----- |----------|
| 1. Brief | What decision does this analysis feed? | `templates/analysis-brief.md` | Filled brief, confirmed by the requester |
| 2. Data contract | What data, what columns, what quality? | `templates/data-contract.md` | Contract + quality-gate results |
| 3. Honest metrics | What is the true number, under which definition? | route: delivery → `otif-analysis` · forecasts → `forecast-accuracy-review` · portfolio/stock → `abc-xyz-segmentation`, `safety-stock-review` | Metric ladder / scoreboard |
| 4. Drivers | Where does the problem concentrate? | `root-cause-pareto` | Ranked drivers + sub-analysis of #1 |
| 5. Stress test | Where does the standard answer break? | the stress sections inside the routed skill | Trust statement per segment |
| 6. Decision memo | What should the reader do? | `templates/decision-memo.md` | One-page memo with counter-metric |
For a recurring deliverable (weekly/monthly), wrap steps 3-6 with `weekly-ops-report` instead of a one-off memo.