Skip to content
The Definitive Source To Digital Transformation

2025 April 20th Update: Data Products, Engineering Practices, and AI Operationalization

This update introduces greater depth on operationalizing digital transformation and AI, focusing on how data is managed as a product, the engineering practices required for robust implementation, and specific models for delivering AI services. We delve into the concept of Data Products as reusable assets, explore modern engineering disciplines like DevOps and MLOps, and detail frameworks for understanding AI service delivery and evaluating data quality for AI.

Additionally, the update clarifies strategic scoping using a Domain-Based approach, provides more detail on federated data governance structures, and incorporates empirical findings linking reported transformation efforts to market valuation. New concepts like the AI Project Capabilities Triumvirate and specific AI integration patterns offer practical guidance for implementation teams.

New

Section titled “New”
  • Digital Business Design: Emphasizes DT as a deliberate leadership configuration of people, process, and technology to create unique value, moving beyond simple restructuring. Ch 1
  • Domain-Based Scoping: Introduces identifying and prioritizing specific business ‘domains’ (based on journeys, capabilities, functions) using Value vs. Feasibility criteria as a method for defining transformation scope alongside the Core/Adjacent/Frontier model. Ch 2
  • Federated Data Governance Model: Details this common governance structure, explaining the roles and responsibilities of central data offices/councils (setting policy) versus local business units (implementation, stewardship). Ch 8
  • Data Products: Significantly expands on this concept, defining characteristics (Discoverable, Addressable, Trustworthy, Self-describing, Interoperable, Secure) and common consumption archetypes (Applications, Analytics, BI, Sandboxes, External Sharing). Positions data products as key enablers for scaling data value and supporting architectures like Data Mesh. Ch 7, Ch 8
  • Data Quality Dimensions for AI: Presents three key dimensions for assessing data quality specifically for AI use: Technical Correctness, Match with Reality, and Reputation of Data. Ch 8
  • Data Compartmentalization Framework: Introduces a 3D framework (Sensitivity, Domain, Jurisdiction/Org Structure) for managing data access control effectively in complex environments like data lakes used for AI. Ch 8
  • Advanced Data Privacy Techniques: Details specific methods for handling sensitive attributes: Not Copying Sensitive Data, Anonymization, and Veiling (Pseudonymization), explaining their trade-offs. Ch 8
  • Ontology Components (Is-ness vs. About-ness): Provides a detailed explanation and examples of these fundamental metadata concepts for classifying information assets within an ontology. Ch 9
  • AI Project Capabilities Triumvirate: Defines the essential skill sets needed for AI projects: Data Science, Data Engineering, and the bridging role of the AI Translator. Ch 5
  • AI Model Integration Patterns: Details three primary architectural patterns for deploying AI models into operational systems: Precalculation, Model (Re)implementation, and Encapsulated AI Component (via AI Runtime Server). Ch 5
  • Modern Engineering Practices: Adds dedicated sections covering:
    • DevOps & xOps: Principles (Flow, Feedback, Learning) and related disciplines (DevSecOps, MLOps, DataOps). Ch 15
    • Code Quality & Technical Debt: Practices (Version Control, Frameworks, Standards, Testing, Simplicity) and the importance of managing technical debt. Ch 15
    • CI/CD: Continuous Integration and Continuous Deployment pipelines for faster, reliable delivery. Ch 15
    • Developer Productivity: Enhancing efficiency via Internal Developer Platforms (IDPs), Sandboxes, Self-Service, and Standardized Tooling. Ch 15
    • Production Grade: Ensuring Control/Auditability, Security/Scalability/Availability, and Monitoring/Observability for deployed solutions. Ch 15
    • Security (“Shift Left”): Integrating security throughout the SDLC (DevSecOps). Ch 6, Ch 15
  • AI Service Delivery Models (Factory, Shop, Mall, Boutique): Introduces a framework categorizing AI service delivery based on Customization and User Involvement levels. Ch 18
  • AI Service Attributes (Pivotal, Core, Peripheral): Presents a model for classifying AI service attributes based on their importance to acceptability and value delivery. Ch 18
  • Designing for Reuse (“Assetization”): Details the concept of designing reusable solution packages (“recipes”) containing implementation steps, modular technology, and support roles to accelerate scaling across the enterprise. Outlines scaling approaches (Linear Waves, Exponential Waves, Big Bang) and assetization levels (Feature, Application, Platform). Ch 13
  • Performance Architecture & Stage-Gates: Defines a comprehensive performance management architecture with three KPI families (Value Creation, Pod Health, Change Management) and the use of Stage-Gate (L0-L5) reviews for tracking initiative progress. Ch 13
  • Operationalizing Risk Management: Outlines practical steps for embedding risk management: Risk Triage, Policy Review Cadence, Operationalizing Policies (Controls, Talent, Automation), and Raising Awareness. Ch 16
  • Jupyter Notebooks & AutoML: Introduces these specific tools used in data science and AI development for experimentation, communication, and automating parts of the ML workflow. Ch 21, Ch 21

Updated

Section titled “Updated”
  • Strategy (Unified Approach): Reinforced the critique of separate “digital strategies” and formalized the hierarchy (Business -> Digital -> Data -> Monetization). Ch 2
  • Leadership Roles (Data Monetization Owners): Added specific operational owner roles (Process, Product, Information Solution Owner) crucial for driving data monetization initiatives. Ch 3
  • Digital Platform (Asset Focus): Explicitly positioned the digital platform as the central hub for accessing curated data assets and shared data monetization capabilities. Ch 6
  • Data Governance (Federated Model & Roles): Provided more detail on the structure and roles (Owner, Steward, Analyst) within a federated data governance model. Clarified link between governance and enabling Data Democracy. Ch 8
  • Data Management (Data Products/Roadmap): Integrated the Data Product concept as a sophisticated approach to managing reusable data assets. Introduced a three-level data roadmap approach (Foundational, Expansion, Optimization). Ch 8
  • Data Catalog (Features): Added details on advanced Data Catalog features like collaborative tagging, lineage visualization, and crowd-intelligence. Ch 8
  • Risk Management (AI Landscape Mapping & Mitigation): Added a structured process for mapping the AI risk landscape (Threat Actors, Assets, Threats, Likelihood/Impact, Mitigation). Incorporated specific AI mitigation techniques (System Hardening, Governance, Compartmentalization, Sensitive Attribute Handling, Probing Detection, Cloud-AI Risks). Ch 16, Ch 16
  • Value Measurement (Linking DT to Market Value/Earnings): Incorporated empirical research findings on the correlation (or lack thereof) between reported DT activities and financial metrics like market capitalization and future earnings (ROA3Y), highlighting contextual dependencies. Ch 17
  • Talent (AI Translator Role Definition): Consolidated descriptions and clearly defined the AI Translator role, highlighting necessary skills. Ch 10
  • User Perceptions of AI: Added user segmentation (Skeptics, Novices, Explorers, Power Users) to discussions on cultural resistance and embedding norms. Ch 11, Ch 13
  • Change Management (Performance Arch & TO Role): Integrated the detailed performance management architecture (Value, Pod Health, Change KPIs) and the central coordinating role of the Transformation Office (TO). Ch 13

Removed

Section titled “Removed”
  • No major content sections were removed in this update. The focus has been on adding new detailed frameworks for operationalizing data and AI, incorporating modern engineering practices, and refining existing concepts with more practical implementation guidance.