Scaling Up and Scaling Out Design Thinking: Cost Saving and Risk Taking in AI-Driven Hardware and Software Development

Abstract

In the era of rapid artificial intelligence evolution, hardware and software product development must simultaneously balance Scaling up (vertical expansion) and Scaling out (horizontal expansion) strategies to achieve optimal balance between Cost Saving and Risk Taking. This paper combines Six Sigma design methodologies (DMAIC, DMEDI, DFMEA) with AI + Big Data support infrastructure to propose a dynamic system architecture that encompasses business strategy, value chain, customer relationships, and key process activities as the core foundation for product design and decision-making.

Keywords: Scaling strategies, Six Sigma, design thinking, cost optimization, risk management, artificial intelligence, big data analytics

1. Introduction

The current AI infrastructure market is experiencing unprecedented growth, highlighting the urgent need for enterprises to adopt effective scaling strategies. In this context, traditional product development methods are no longer sufficient to address rapidly changing technological environments. The concepts of Scaling up (vertical expansion) and Scaling out(horizontal expansion) have become key strategic considerations:

Scaling up: Enhancing the performance of a single system (such as increasing CPU cores, memory, optimizing software algorithms)
Scaling out: Expanding system capabilities through distributed architecture or multi-node collaboration

Currently, AI scaling laws face the challenge of diminishing marginal returns, forcing AI laboratories to change their development paths. Traditional scaling methods and expectations are encountering bottlenecks, requiring new thinking paradigms.

The core assumption of this framework is: The key to successful product development lies in simultaneously leveraging Scaling up (reducing unit costs) and Scaling out (distributing and undertaking innovation risks), and through the integration of AI and big data, making cost optimization and risk prediction more precise. This paper proposes an innovative dynamic system architecture that integrates Six Sigma methodologies with AI technology, providing systematic guidance framework for hardware and software product development.

2. Dynamic System Architecture Design

2.1 System Architecture Overview

Based on the provided system architecture diagram (Figure 1), this study proposes a dynamic system containing four core levels, forming a complete closed-loop optimization system:

2.1.1 Business Strategy Pyramid (Upper Architecture)

The pyramid structure contains four strategic levels from top to bottom:

Goal (Ultimate Objective) - Enterprise overall strategic objectives

↓

Value Chain - Core value creation activities

↓

Customer Need & Relationship - Market-oriented strategy

↓ Business Process & Key Activity - Execution-level activities

2.1.2 Dual-Loop Dynamic Optimization Model (Core Mechanism)

The system core adopts a dual-loop design, with two loops intertwining to form an infinity symbol (∞) structure:

Left Loop: DMAIC Cost Saving Cycle

Define → Measure → Analyze → Improve → Control
Objective: Achieve Cost Saving through systematic improvement

Right Loop: Design Innovation Risk Taking Cycle

Design → Failure → Mode → Analyze → Effect
Objective: Achieve Risk Taking through risk assessment

Convergence Zone: CLCA Continuous Assessment The two loops form a Closed-Loop Continuous Assessment (CLCA) mechanism at the central convergence point, ensuring dynamic balance between cost saving and risk taking.

2.1.3 Flanking Support System (Left and Right Pillars)

Left Pillar: DMEDI Innovation Development Process

Define – Clarify innovation objectives
Measure – Establish evaluation indicators
Explore – Technology possibility exploration
Develop – Solution development
Implement – Solution implementation

Right Pillar: NUDD Differentiation Strategy Framework

New – Degree of technological innovation
Unique – Degree of market differentiation
Difficult – Difficulty for competitors to imitate
Different – Degree of product differentiation
Yield Rate – Mass production feasibility assessment

2.1.4 AI and Big Data Support Infrastructure Layer (Bottom Enablement)

Agentic AI and Data System Support Base

Intelligent Agent AI System: Autonomous decision-making, learning optimization, predictive analysis
Big Data Platform: Real-time data collection, processing, analysis, and visualization
System Integration Interface: Connecting upper-level business logic with lower-level technical architecture

2.2 System Architecture Operation Mechanism Analysis

2.2.1 Vertical Integration Mechanism (From Strategy to Execution)

The system architecture adopts top-down strategic decomposition and bottom-up data-driven bidirectional integration:

Strategic Level to Execution Level Transmission:

Goal Level: Set overall enterprise objectives (such as market share, profitability targets)
Value Chain Level: Identify key value activities (such as R&D, manufacturing, marketing)
Customer Level: Clarify customer needs and relationship management strategies
Process Level: Design specific business processes and key activities

Execution Level to Strategic Level Feedback:

Collect execution-level data through Agentic AI systems
Big data analysis provides strategic adjustment recommendations
Real-time feedback mechanism ensures consistency between strategy and execution

2.2.2 Horizontal Coordination Mechanism (Dual-Loop Dynamic Balance)

DMAIC Cost Saving Loop Operation Logic:

Define → Measure → Analyze → Improve → Control

↓ ↓ ↓ ↓ ↓

Goal Setting → Data Collection → Root Cause Analysis → Improvement Implementation → Continuous Monitoring

Design Innovation Risk Taking Loop Operation Logic:

Design → Failure → Mode → Analyze → Effect

↓ ↓ ↓ ↓ ↓

Innovation Design → Failure Identification → Pattern Analysis → Impact Assessment → Effect Verification

Dual-Loop Convergence Mechanism (CLCA):

Balance Point Identification: Dynamically monitor the optimal balance point between cost saving and risk taking
Resource Allocation Optimization: Dynamically allocate resources between two loops
Decision Support: Provide data-based decision recommendations

2.2.3 Flanking Support System Coordination

Left DMEDI Coordination with Main Loop:

DMEDI focuses on breakthrough innovation (Scaling Up technical breakthroughs)
Forms exploration-exploitation balance with DMAIC
Identifies new technology opportunities through Explore phase

Right NUDD Coordination with Main Loop:

NUDD evaluates market differentiation potential (Scaling Out market expansion)
Forms innovation-validation cycle with risk taking loop
Ensures innovation manufacturability through Yield Rate

2.3 DFMEA Risk Assessment Integration

Integrating DFMEA into the dynamic system, achieving risk taking through the following mechanisms:

2.3.1 Risk Identification

AI Assistance: Historical failure data analysis, predicting potential failure modes
Expert System: Combining domain knowledge with AI recommendations

2.3.2 Risk Assessment

Dynamic RPN Calculation: Real-time update of Risk Priority Number
Scenario Analysis: Risk impact assessment under multiple scenarios

2.3.3 Risk Mitigation

Preventive Measures: AI-recommended proactive prevention actions
Contingency Plans: Automatically triggered emergency response mechanisms

3. Scaling Strategy and Cost-Risk Balance

3.1 Architecture-Based Scaling Strategy Implementation Framework

According to the design philosophy of the system architecture diagram, the implementation of Scaling up and Scaling out strategies follows the following integration framework:

3.1.1 Scaling Up Strategy: Vertical Performance Enhancement Path

Implemented through DMAIC Cost Saving Loop:

Define Phase: Based on Goal level strategic objectives, define specific performance improvement indicators
- System throughput improvement targets
- Resource utilization optimization targets
- Unit cost reduction targets
Measure Phase: Establish Agentic AI monitoring system
- Real-time performance data collection
- Resource consumption pattern analysis
- Bottleneck identification and tracking
Analyze Phase: Big data analysis supports root cause analysis
- AI-driven performance bottleneck analysis
- System load pattern identification
- Optimization opportunity quantitative assessment
Improve Phase: Technology upgrade and optimization implementation
- Hardware resource expansion (CPU, memory, storage)
- Software algorithm optimization
- System architecture reconstruction
Control Phase: Continuous monitoring and dynamic adjustment
- Automated performance tuning
- Anomaly detection and response
- Benefit tracking and assessment

Role of Left DMEDI Support System:

Explore Phase: Explore breakthrough technologies (quantum computing, edge AI, etc.)
Develop Phase: Develop innovative solutions
Implement Phase: Integrate breakthrough technologies into existing systems

3.1.2 Scaling Out Strategy: Horizontal Expansion Path

Implemented through Design Innovation Risk Taking Loop:

Design Phase: Based on Customer Need level, design distributed architecture
- Microservice architecture design
- Modular system planning
- Cross-platform compatibility design
Failure Phase: Identify potential failure points in distributed systems
- Network communication failure risks
- Data consistency risks
- Load balancing failure risks
Mode Phase: Analyze failure modes and impacts
- System availability impact assessment
- Data integrity risk assessment
- User experience impact analysis
Analyze Phase: Risk quantification and priority ranking
- RPN (Risk Priority Number) calculation
- Cost-benefit analysis
- Risk acceptance assessment
Effect Phase: Implement risk mitigation measures
- Fault tolerance mechanism design
- Backup system construction
- Monitoring and warning systems

Role of Right NUDD Support System:

New & Unique: Evaluate the innovation and uniqueness of new architecture
Difficult & Different: Analyze the difficulty for competitors to replicate
Yield Rate: Ensure the practicality and stability of distributed systems

3.2 CLCA Dynamic Balance Mechanism Implementation

3.2.1 Operation Logic of Dual-Loop Convergence Area

CLCA (Closed-Loop Continuous Assessment) serves as the core coordination mechanism of the system architecture, achieving dynamic balance between cost saving and risk taking.

3.2.2 Key Role of Agentic AI and Big Data Support in CLCA

Intelligent Decision Support:

Real-time Data Fusion: Integrate operational data from dual loops
Trend Prediction: Predict market changes’ impact on balance points
Autonomous Adjustment: Automatically adjust weight parameters based on environmental changes
Anomaly Detection: Identify system operation anomalies and trigger human intervention

Learning Optimization Mechanism:

Reinforcement Learning: Learn optimal balance strategies from historical decision results
Transfer Learning: Apply successful experiences to new application scenarios
Federated Learning: Share learning outcomes across organizations (under privacy protection)

4. Challenges and Limitations

4.1 Technical Challenges

4.1.1 Data Quality Issues

Challenge: AI and big data analysis effectiveness highly depends on data quality
Solution: Establish data governance framework to ensure data accuracy and completeness

4.1.2 System Complexity

Challenge: Integrating multiple methodologies may increase system complexity
Solution: Phased implementation, gradually building organizational capabilities

4.2 Organizational Challenges

4.2.1 Cultural Transformation

Resistance: Traditional organizations may resist new working methods
Strategy: Change management and continuous training

4.2.2 Talent Requirements

Need: Personnel with cross-disciplinary knowledge
Response: Combine internal training with external recruitment

4.3 Economic Considerations

4.3.1 Initial Investment

Cost: AI infrastructure and system construction require substantial initial investment
Return: Reduce initial risks through phased implementation

5. Conclusion

5.1 Major Findings

In the era of rapid AI development, integrating Scaling up and Scaling out strategies with cost saving and risk taking as core considerations in design thinking can effectively improve the development efficiency and success rate of hardware and software products. Key findings include:

Dynamic Balance Mechanism: Cost saving and risk taking are not opposing relationships, but can achieve dynamic balance through systematic design
AI Enablement Effect: Integration of AI and big data significantly improves the efficiency and accuracy of Six Sigma methodologies
Cross-domain Applicability: The proposed framework shows good applicability in hardware, software, and cross-domain applications

5.2 Theoretical Contributions

Methodology Integration: First systematic integration of Six Sigma DMAIC, DFMEA with AI technology
Dynamic System Architecture: Proposed product development dynamic system framework adapted to the AI era
Balance Strategy Model: Established dynamic balance model between cost saving and risk taking

5.3 Future Directions

Industry Specialization: Develop customized implementation guides for different industries
International Comparative Research: Conduct cross-national comparative research to verify framework universality
New Technology Integration: Research integration possibilities of emerging technologies

擴展式與擴充式設計思維：人工智慧驅動的硬體與軟體開發中的成本節省與風險承擔

摘要

在人工智慧快速演進的時代，硬體與軟體產品的開發必須同時兼顧 Scaling up（垂直擴充） 與 Scaling out（水平擴充） 的策略，以達成最佳的 成本節省（Cost Saving） 與 風險承擔（Risk Taking） 平衡。本文結合 六標準差設計方法（DMAIC、DMEDI、DFMEA） 與 AI + 大數據支援基礎，提出一個動態系統架構，涵蓋商業策略、價值鏈、顧客關係與流程關鍵活動，作為產品設計與決策的核心依據。

關鍵字： 擴展策略、六標準差、設計思維、成本優化、風險管理、人工智慧、大數據分析

1. 引言

當前AI基礎設施市場正經歷前所未有的成長，快速成長凸顯了企業對於有效擴展策略的迫切需求。在此背景下，傳統的產品開發方法已不足以應對快速變化的技術環境。Scaling up（垂直擴充）與 Scaling out（水平擴充）的概念成為關鍵策略考量：

Scaling up：提升單一系統的效能（如增加 CPU 核心、記憶體、優化軟體算法）
Scaling out：透過分散式架構或多節點協作擴展系統能力

目前AI擴展定律面臨邊際效益遞減的挑戰，迫使AI實驗室改變發展路線，傳統的擴展方法和期望正遭遇瓶頸，需要新的思維模式。

本架構的核心假設是：成功的產品開發關鍵在於同時利用 Scaling up（降低單位成本）與 Scaling out（分散與承擔創新風險），並透過AI與大數據的整合使得成本優化與風險預測更加精準，提出一個創新的動態系統架構，整合六標準差方法論與AI技術，為硬體和軟體產品開發提供系統性的指導框架。

2. 動態系統架構設計

2.1 系統架構總覽

基於提供的系統架構圖（圖1），本研究提出的動態系統包含四個核心層次，形成一個完整的閉環優化系統：

2.1.1 商業策略金字塔（上層架構）

金字塔結構由上而下包含四個戰略層次：

Goal（最終目標）- 企業整體戰略目標

↓

Value Chain（價值鏈）- 核心價值創造活動

↓

Customer Need & Relationship（顧客需求與關係）- 市場導向策略

↓

Business Process & Key Activity（業務流程與關鍵活動）- 執行層面活動

2.1.2 雙迴圈動態優化模型（核心機制）

系統核心採用雙迴圈設計，兩個迴圈相互交織形成無窮符號（∞）結構：

左迴圈：DMAIC成本節省循環

Define（定義） → Measure（測量） → Analyze（分析） → Improve（改善） → Control（控制）
目標：透過系統性改善達到Cost Saving（成本節省）

右迴圈：設計創新風險承擔循環

Design（設計） → Failure（失效分析） → Mode（模式識別） → Analyze（分析） → Effect（效果評估）
目標：透過風險評估實現Risk Taking（風險承擔）

交會融合區域：CLCA持續評估 兩個迴圈在中心交會處形成**Closed-Loop Continuous Assessment（CLCA）**機制，確保成本節省與風險承擔的動態平衡。

2.1.3 側翼支援系統（左右支柱）

左側支柱：DMEDI創新開發流程

Define（定義） – 明確創新目標
Measure（測量） – 建立評估指標
Explore（探索） – 技術可能性探索
Develop（開發） – 解決方案開發
Implement（實施） – 方案落地執行

右側支柱：NUDD差異化策略框架

New（新穎性） – 技術創新程度
Unique（獨特性） – 市場差異化程度
Difficult（困難度） – 競爭者模仿難度
Different（差異性） – 產品區隔程度
Yield Rate（良率） – 量產可行性評估

2.1.4 AI與大數據支援基礎層（底層賦能）

Agentic AI and Data System Support Base

智能代理AI系統：自主決策、學習優化、預測分析
大數據平台：即時數據收集、處理、分析與視覺化
系統整合介面：連接上層業務邏輯與底層技術架構

2.2 系統架構運作機制分析

2.2.1 垂直整合機制（從戰略到執行）

系統架構採用由上而下的戰略分解與由下而上的數據驅動雙向整合：

戰略層面向執行層面的傳導：

Goal層：設定企業整體目標（如市場占有率、獲利率目標）
Value Chain層：識別關鍵價值活動（如研發、製造、行銷）
Customer層：明確顧客需求與關係管理策略
Process層：設計具體業務流程與關鍵活動

執行層面向戰略層面的回饋：

透過Agentic AI系統收集執行層數據
大數據分析提供戰略調整建議
即時回饋機制確保戰略與執行的一致性

2.2.2 水平協調機制（雙迴圈動態平衡）

DMAIC成本節省迴圈運作邏輯：

Define → Measure → Analyze → Improve → Control

↓ ↓ ↓ ↓ ↓

目標設定 → 數據收集 → 根因分析 → 改善實施 → 持續監控

設計創新風險承擔迴圈運作邏輯：

Design → Failure → Mode → Analyze → Effect

↓ ↓ ↓ ↓ ↓

創新設計 → 失效識別 → 模式分析 → 影響評估 → 效果驗證

雙迴圈交會機制（CLCA）：

平衡點識別：動態監控成本節省與風險承擔的最佳平衡點
資源分配優化：在兩個迴圈間動態調配資源
決策支援：提供基於數據的決策建議

2.2.3 側翼支援系統協調

左側DMEDI與主迴圈的協調：

DMEDI專注於突破性創新（Scaling Up的技術突破）
與DMAIC形成探索-利用平衡
透過Explore階段識別新技術機會

右側NUDD與主迴圈的協調：

NUDD評估市場差異化潛力（Scaling Out的市場擴展）
與風險承擔迴圈形成創新-驗證循環
透過Yield Rate確保創新的可製造性

2.3 DFMEA風險評估整合

整合DFMEA到動態系統中，透過以下機制實現風險承擔：

2.3.1 風險識別

AI輔助：歷史失效數據分析，預測潛在失效模式
專家系統：結合領域知識和AI推薦

2.3.2 風險評估

動態RPN計算：即時更新風險優先順序數值
情境分析：多種情境下的風險影響評估

2.3.3 風險緩解

預防措施：AI建議的主動預防行動
應急計劃：自動觸發的應急回應機制

3. Scaling策略與成本風險平衡

3.1 基於架構圖的Scaling策略實施框架

根據系統架構圖的設計理念，Scaling up和Scaling out策略的實施遵循以下整合框架：

3.1.1 Scaling Up策略：垂直效能提升路徑

透過DMAIC成本節省迴圈實現：

Define階段：基於Goal層戰略目標，定義具體的效能提升指標
- 系統吞吐量提升目標
- 資源利用率優化目標
- 單位成本降低目標
Measure階段：建立Agentic AI監控系統
- 即時效能數據收集
- 資源消耗模式分析
- 瓶頸點識別與追蹤
Analyze階段：大數據分析支援根因分析
- AI驅動的效能瓶頸分析
- 系統負載模式識別
- 優化機會量化評估
Improve階段：技術升級與優化實施
- 硬體資源擴充（CPU、記憶體、存儲）
- 軟體算法優化
- 系統架構重構
Control階段：持續監控與動態調整
- 自動化效能調優
- 異常檢測與回應
- 效益追蹤與評估

DMEDI左側支援系統的角色：

Explore階段：探索突破性技術（量子運算、邊緣AI等）
Develop階段：開發創新解決方案
Implement階段：將突破性技術整合到現有系統

3.1.2 Scaling Out策略：水平擴展路徑

透過設計創新風險承擔迴圈實現：

Design階段：基於Customer Need層，設計分散式架構
- 微服務架構設計
- 模組化系統規劃
- 跨平台兼容性設計
Failure階段：識別分散式系統潛在失效點
- 網路通訊失敗風險
- 數據一致性風險
- 負載均衡失效風險
Mode階段：分析失效模式與影響
- 系統可用性影響評估
- 數據完整性風險評估
- 用戶體驗影響分析
Analyze階段：風險量化與優先級排序
- RPN（Risk Priority Number）計算
- 成本效益分析
- 風險接受度評估
Effect階段：實施風險緩解措施
- 容錯機制設計
- 備援系統建置
- 監控與預警系統

NUDD右側支援系統的角色：

New & Unique：評估新架構的創新性和獨特性
Difficult & Different：分析競爭者複製的難度
Yield Rate：確保分散式系統的實用性和穩定性

3.2 CLCA動態平衡機制的實現

3.2.1 雙迴圈交會區域的運作邏輯

CLCA（Closed-Loop Continuous Assessment）作為系統架構的核心協調機制，實現成本節省與風險承擔的動態平衡。

3.2.2 Agentic AI 與大數據支援在CLCA中的關鍵作用

智能決策支援：

即時數據融合：整合雙迴圈的運作數據
趨勢預測：預測市場變化對平衡點的影響
自主調整：根據環境變化自動調整權重參數
異常檢測：識別系統運作異常並觸發人工介入

學習優化機制：

強化學習：從歷史決策結果中學習最佳平衡策略
遷移學習：將成功經驗應用到新的應用場景
聯邦學習：跨組織分享學習成果（保護隱私前提下）

4. 挑戰與限制

4.1 技術挑戰

4.1.1 數據品質問題

挑戰：AI和大數據分析的效果高度依賴數據品質
解決方案：建立數據治理框架，確保數據準確性和完整性

4.1.2 系統複雜度

挑戰：整合多種方法論可能增加系統複雜度
解決方案：階段性實施，逐步建立組織能力

4.2 組織挑戰

4.2.1 文化變革

阻力：傳統組織可能抗拒新的工作方式
策略：變革管理和持續培訓

4.2.2 人才需求

需求：需要具備跨領域知識的人才
應對：內部培訓結合外部招募

4.3 經濟考量

4.3.1 初期投資

成本：AI基礎設施和系統建置需要大量初期投資
回報：透過階段性實施降低初期風險

5. 結論

5.1 主要發現

在AI快速發展的時代，整合Scaling up和Scaling out策略，並以成本節省和風險承擔為核心考量的設計思維能夠有效提升硬體和軟體產品的開發效率與成功率，關鍵包括：

動態平衡機制：成本節省與風險承擔並非對立關係，而是可以通過系統性設計實現動態平衡
AI賦能效果：AI和大數據的整合顯著提升了六標準差方法論的效率和準確性
跨領域適用性：提出的框架在硬體、軟體和跨領域應用中都顯示出良好的適用性

5.2 理論

方法論整合：首次系統性整合六標準差DMAIC、DFMEA與AI技術
動態系統架構：提出適應AI時代的產品開發動態系統框架
平衡策略模型：建立成本節省與風險承擔的動態平衡模型

5.3 未來方向

行業特定化：針對不同行業開發客製化的實施指南
國際比較研究：進行跨國比較研究驗證框架的普適性
新技術整合：研究新興技術的整合可能性

參考文獻

Epoch AI. (2024). Can AI Scaling Continue Through 2030?. Retrieved from https://epoch.ai/blog/can-ai-scaling-continue-through-2030
IBM Security Intelligence. (2025). Trends: Hardware gets AI updates in 2024. Retrieved from https://securityintelligence.com/articles/trends-hardware-gets-ai-updates-2024/
Snell, et al. (2024). AI Scaling: From Up to Down and Out. arXiv preprint arXiv:2502.01677v1.
TechCrunch. (2024). Current AI scaling laws are showing diminishing returns, forcing AI labs to change course. Retrieved from https://techcrunch.com/2024/11/20/ai-scaling-laws-are-showing-diminishing-returns-forcing-ai-labs-to-change-course/
Our World in Data. (2025). Scaling up: how increasing inputs has made artificial intelligence more capable. Retrieved from https://ourworldindata.org/scaling-up-ai
VentureBeat. (2025). Purpose-built AI hardware: Smart strategies for scaling infrastructure. Retrieved from https://venturebeat.com/ai/purpose-built-ai-hardware-the-key-to-scalable-ai-infrastructure/
Taylor & Francis Online. (2023). Integration of Industry 4.0 technologies into Lean Six Sigma DMAIC: a systematic review. International Journal of Production Research.
Harvard Business Review. (2023). How AI Fits into Lean Six Sigma. Retrieved from https://hbr.org/2023/11/how-ai-fits-into-lean-six-sigma
iSixSigma. (2025). How AI Can Be Used in the DMAIC Process. Retrieved from https://www.isixsigma.com/artificial-intelligence/how-ai-can-be-used-in-the-dmaic-process/
Taylor & Francis Online. (2025). DMAIC 4.0 – innovating the Lean Six Sigma methodology with Industry 4.0 technologies. International Journal of Production Research.
Emerald Publishing. (2022). Integration of Six Sigma methodology of DMADV steps with QFD, DFMEA and TRIZ applications for image-based automated inspection system development: a case study. International Journal of Lean Six Sigma, 13(6), 1239-1270.
Six Sigma Institute. (2025). FMEA Six Sigma DMAIC: Failure Mode Effect Analysis. Retrieved from https://www.sixsigma-institute.org/Six_Sigma_DMAIC_Process_Improve_Phase_Failure_Mode_Effect_Analysis_FMEA.php
Academia.edu. (2024). The Future of Six Sigma: Integrating AI for Continuous Improvement. Retrieved from https://www.academia.edu/124464749/The_Future_of_Six_Sigma_Integrating_AI_for_Continuous_Improvement
SixSigma.us. (2025). DMAIC: Approach to Continuous Improvement. Retrieved from https://www.6sigma.us/dmaic-process/