biz-supply-chain

asgard-ai-platform ｜ 商業與營運 ｜ 原文 ｜ MIT

用 SCOR 模型分析供應鏈五大流程並找出最弱環節

---
name: "biz-supply-chain"
description: "用 SCOR 模型，從規劃、採購、製造、交付、退貨五個流程分析供應鏈營運。當使用者需要最佳化供應鏈效率、評估供應商績效、改善物流、或設計端到端的供應鏈策略時使用，即使他們說的是「我們的出貨很慢」、「供應鏈成本太高」或「我們一直缺貨」也一樣。"
metadata:
  category: "WP-16 商學院—營運"
  tags: ["operations", "supply-chain", "scor", "logistics"]
---

# 供應鏈分析（SCOR 模型）

## 概述

SCOR（Supply Chain Operations Reference，供應鏈運作參考）模型把供應鏈分析整理成五個核心流程：Plan（規劃）、Source（採購）、Make（製造）、Deliver（交付）、Return（退貨）。它提供一套共通語言，用來分析、對標和改善供應鏈績效，範圍從供應商的供應商，到顧客的顧客。

## 框架

```
鐵律：端到端，不是各部門各管各的

供應鏈最佳化必須考慮「整條」鏈。只最佳化採購（Source），
卻不考慮生產產能（Make）或交付能力（Deliver），
會在下游造成瓶頸。

一個流程的局部最佳，常常會在別處造成全域性的問題。
```

### SCOR 的五個流程

**1. Plan（規劃）**：需求預測、供應規劃、庫存策略
- 需求預測準確度、S&OP 流程、庫存政策
- 問題：「我們有在對的時間製造／採購對的數量嗎？」

**2. Source（採購）**：供應商選擇、採購、進料品質
- 供應商計分卡、前置時間、採購策略（單一 vs 多元）
- 問題：「我們有以對的成本和品質取得對的投入嗎？」

**3. Make（製造）**：生產、組裝、製造
- 生產排程、產能利用率、品質管制、在製品（WIP）管理
- 問題：「我們有效率地把投入轉成產出嗎？」

**4. Deliver（交付）**：訂單管理、倉儲、運輸、最後一哩
- 訂單滿足率、交付速度、物流成本、通路管理
- 問題：「我們有可靠且負擔得起地把產品送到顧客手上嗎？」

**5. Return（退貨）**：退貨處理、逆向物流、保固／維修
- 退貨率、逆向物流成本、整修、報廢
- 問題：「我們有有效率地處理退貨，並從中學習嗎？」

### 關鍵的供應鏈指標

| 流程 | 指標 | 公式／定義 |
|---------|--------|-------------------|
| Plan | 預測準確度 | 1 - \|實際 - 預測\| / 實際 |
| Plan | 庫存天數 | 庫存 / (COGS / 365) |
| Source | 供應商準時率 | 準時交貨數 / 總交貨數 |
| Source | 供應商不良率 | 不良品數 / 總收貨數 |
| Make | OEE | 可用率 × 效能 × 品質 |
| Deliver | 完美訂單率 | 準時、足量、無誤交付的訂單 |
| Deliver | 訂單到交付週期 | 從下單到顧客收到的時間 |
| Return | 退貨率 | 退貨數 / 總出貨數 |

### 分析步驟

1. **畫出目前的供應鏈**，從供應商到顧客
2. 依各 SCOR 流程**衡量**關鍵指標
3. 和產業標準做**對標**
4. **找出**最弱的流程（與基準差距最大的）
5. 先**改善**最弱的環節（和 TOC 的邏輯相同，鏈條的強度取決於最弱的一環）

## 輸出格式

```markdown
# 供應鏈分析：{公司}

## 供應鏈地圖
供應商 → [Source] → [Make] → [Deliver] → 顧客
            ↑ [Plan]（協調全部）↑
                     [Return] ←

## SCOR 績效儀表板
| 流程 | 關鍵指標 | 目前 | 基準 | 差距 |
|---------|-----------|---------|-----------|-----|
| Plan | 預測準確度 | X% | 85%+ | {差距} |
| Source | 供應商準時率 | X% | 95%+ | {差距} |
| Make | OEE | X% | 85%+ | {差距} |
| Deliver | 完美訂單率 | X% | 95%+ | {差距} |
| Return | 退貨率 | X% | <5% | {差距} |

## 最弱環節分析
{哪個流程的差距最大，以及原因}

## 改善建議
1. {流程}：{具體改善} → {預期的指標影響}
```

## 範例

### 正確的應用
**情境：** 台灣一個 DTC 電子品牌的 SCOR 分析
| 流程 | 指標 | 目前 | 問題 |
|---------|--------|---------|-------|
| Plan | 預測準確度 | 62% | 促銷期間的需求暴增沒被預測到 |
| Source | 供應商準時率 | 88% | 深圳的關鍵零組件供應商前置時間不穩定 |
| Make | OEE | 78% | 以電子組裝來說算合理 |
| Deliver | 完美訂單率 | 91% | 最後一哩的物流業者（黑貓）會遺失 3% 的包裹 |
| **最弱**：Plan（62%），修正預測準確度，能同時降低庫存（目前 45 天，目標 30 天）和缺貨

### 錯誤的應用
- 只分析 Deliver（物流），因為「出貨是我們被客訴最多的地方」→ 根本原因是 Plan（預測不準 → 缺貨 → 缺貨待補訂單 → 延遲交貨）。只修交付沒有幫助。違反鐵律：端到端的分析。

## 常見陷阱

- **長鞭效應**：零售端的小幅需求變動，會在上游被放大。銷售增加 10%，可能引發製造商產量增加 40%。S&OP 流程能緩解這件事。
- **單一來源風險**：一個供應商 = 零備援。2021 年的晶片短缺在全球證明了這點。要明確評估對單一來源的依賴。
- **自製或外購是 Source 的決策**：不只是成本，還要考慮智慧財產權保護、品質控制、前置時間和供應安全。
- **最後一哩通常最貴**：最後一哩的配送，可能占總物流成本的 40-50%。要評估配送模式（自有車隊 vs 第三方物流 vs 取貨點）。
- **退貨是利潤的漏洞**：很多公司把退貨當成事後才想到的事。電商 15% 的退貨率，代表 15% 的履約成本被浪費，還要加上逆向物流的成本。

## 參考資料

- SCOR 模型的詳細指標，見 `references/scor-metrics.md`
- 庫存最佳化的方法，見 `references/inventory-models.md`

---
name: "biz-supply-chain"
description: "Analyze supply chain operations using the SCOR model across Plan, Source, Make, Deliver, and Return processes. Use this skill when the user needs to optimize supply chain efficiency, evaluate supplier performance, improve logistics, or design an end-to-end supply chain strategy — even if they say 'our deliveries are slow', 'supply chain costs are too high', or 'we keep running out of stock'."
metadata:
  category: "WP-16 商學院—營運"
  tags: ["operations", "supply-chain", "scor", "logistics"]
---

# Supply Chain Analysis (SCOR Model)

## Overview

The SCOR (Supply Chain Operations Reference) model structures supply chain analysis into five core processes: Plan, Source, Make, Deliver, Return. It provides a common language for analyzing, benchmarking, and improving supply chain performance from supplier's supplier to customer's customer.

## Framework

```
IRON LAW: End-to-End, Not Silo-by-Silo

Supply chain optimization must consider the ENTIRE chain. Optimizing
procurement (Source) without considering production capacity (Make) or
delivery capability (Deliver) creates bottlenecks downstream.

A local optimum in one process often creates a global problem elsewhere.
```

### The Five SCOR Processes

**1. Plan** — Demand forecasting, supply planning, inventory strategy
- Demand forecast accuracy, S&OP process, inventory policies
- Question: "Do we make/buy the right amount at the right time?"

**2. Source** — Supplier selection, procurement, incoming quality
- Supplier scorecards, lead times, sourcing strategy (single vs multi)
- Question: "Are we getting the right inputs at the right cost and quality?"

**3. Make** — Production, assembly, manufacturing
- Production scheduling, capacity utilization, quality control, WIP management
- Question: "Are we converting inputs to outputs efficiently?"

**4. Deliver** — Order management, warehousing, transportation, last-mile
- Order fulfillment rate, delivery speed, logistics cost, channel management
- Question: "Are we getting products to customers reliably and affordably?"

**5. Return** — Returns processing, reverse logistics, warranty/repair
- Return rate, reverse logistics cost, refurbishment, disposal
- Question: "Are we handling returns efficiently and learning from them?"

### Key Supply Chain Metrics

| Process | Metric | Formula/Definition |
|---------|--------|-------------------|
| Plan | Forecast Accuracy | 1 - \|Actual - Forecast\| / Actual |
| Plan | Inventory Days | Inventory / (COGS / 365) |
| Source | Supplier On-Time Rate | On-time deliveries / Total deliveries |
| Source | Supplier Defect Rate | Defective units / Total received |
| Make | OEE | Availability × Performance × Quality |
| Deliver | Perfect Order Rate | Orders delivered on time, in full, without error |
| Deliver | Order-to-Delivery Cycle | Time from order to customer receipt |
| Return | Return Rate | Returns / Total shipments |

### Analysis Steps

1. **Map the current supply chain** from supplier to customer
2. **Measure** key metrics per SCOR process
3. **Benchmark** against industry standards
4. **Identify** the weakest process (highest gap to benchmark)
5. **Improve** the weakest link first (same logic as TOC — chain is as strong as weakest link)

## Output Format

```markdown
# Supply Chain Analysis: {Company}

## Supply Chain Map
Supplier → [Source] → [Make] → [Deliver] → Customer
            ↑ [Plan] (coordinates all) ↑
                     [Return] ←

## SCOR Performance Dashboard
| Process | Key Metric | Current | Benchmark | Gap |
|---------|-----------|---------|-----------|-----|
| Plan | Forecast Accuracy | X% | 85%+ | {gap} |
| Source | Supplier On-Time | X% | 95%+ | {gap} |
| Make | OEE | X% | 85%+ | {gap} |
| Deliver | Perfect Order Rate | X% | 95%+ | {gap} |
| Return | Return Rate | X% | <5% | {gap} |

## Weakest Link Analysis
{Which process has the largest gap and why}

## Improvement Recommendations
1. {Process}: {specific improvement} → {expected metric impact}
```

## Examples

### Correct Application
**Scenario:** SCOR analysis for a Taiwanese DTC electronics brand
| Process | Metric | Current | Issue |
|---------|--------|---------|-------|
| Plan | Forecast Accuracy | 62% | Demand spikes around promotions are unpredicted |
| Source | Supplier On-Time | 88% | Key component supplier in Shenzhen has inconsistent lead times |
| Make | OEE | 78% | Reasonable for electronics assembly |
| Deliver | Perfect Order Rate | 91% | Last-mile carrier (���貓) loses 3% of packages |
| **Weakest**: Plan (62%) — fixing forecast accuracy would reduce both inventory (currently 45 days, target 30) and stockouts

### Incorrect Application
- Only analyzed Deliver (logistics) because "delivery is our biggest complaint" → Root cause was Plan (bad forecast → stockouts → backorders → late deliveries). Fixing delivery alone doesn't help. Violates Iron Law: end-to-end analysis.

## Gotchas

- **Bullwhip effect**: Small demand changes at retail amplify upstream. A 10% sales increase can trigger 40% production increase at the manufacturer. S&OP process mitigates this.
- **Single-source risk**: One supplier = zero redundancy. The 2021 chip shortage proved this globally. Evaluate single-source dependencies explicitly.
- **Make-vs-Buy is a Source decision**: Not just cost — consider IP protection, quality control, lead time, and supply security.
- **Last-mile is often the costliest**: Last-mile delivery can be 40-50% of total logistics cost. Evaluate delivery model (own fleet vs 3PL vs pickup points).
- **Returns are a profit leak**: Many companies treat returns as an afterthought. A 15% return rate in e-commerce means 15% of fulfillment cost is wasted plus reverse logistics cost.

## References

- For SCOR model detailed metrics, see `references/scor-metrics.md`
- For inventory optimization methods, see `references/inventory-models.md`