Data Readiness: The Backbone of AI Success

KPMG - You Can With AI Series | Episode: Data Readiness
Guest: Daniel Bearinger, Principal Data Modernization Practice, KPMG
Former CDO at Nissan | Background: Software & Data Engineer

Why Data Matters for AI: The Foundation

• Data is the prerequisite for AI/ML and GenAI success, not an afterthought.
• High-quality data that is well understood and contextualized is critical.

The Grocery Store Analogy:

• Products on shelves have labels showing identity, origin, and utility.
• Data must have clear utility: What is this data for? What can it be used for?
• Without understanding, building successful AI is hard.
• Data must be understood, explainable, and empowering.

Data as Team Sport:

• Requires collaboration between business stakeholders and technologists.
• All must understand: What data will be used for? What certifies it as fit for AI? What are outcomes?
• Without high-fidelity, high-quality data at the start, innovation and value cannot be realized.

Enterprise AI vs Generic GenAI: The Data Difference

Retrieval-Augmented Generation (RAG)

• Enables organizations to converse with their data in natural language.
• Users ask questions without needing SQL; ChatGPT-style interfaces grounded in corporate data.

Context Grounding Benefits

• GenAI grounded in corporate data vs generic internet search.
• Data is relevant to user’s business context (finance, supply chain, sales, etc.).
• Time to insights is dramatically shortened; no need for technical expertise.
• Users recognize context faster as it’s their own data.

Understanding Data Readiness: Framework

• Definition: People, Process, and Technology.
• Key Principle: Adoption drives success, not just technology presence.

Core Data Management Functions

Data Governance Foundations

• Data Quality: What makes data fit for use?
• Business Glossary: Clear definitions for data meaning.
• Data Ownership: Who is responsible?
• Standardized Terminology: Ensure consistency.
• Process Definition: Who does what when preparing data?

Data Preparation Lifecycle Evolution

• Before: Manual architect tasks (schema, validation).
• Now: GenAI-augmented automation:
  • Automated labeling/classification
  • Automated metadata enrichment
  • Automated data catalog seeding
  • Quick validation interfaces

The Acceleration Factor

• Before: Weeks/months for readiness.
• Now: Hours/days with right people + GenAI automation.
• Result: Certified, ready data much faster.

Three Primary Organizational Patterns

1. Large ERP Transformation

• Long journey for core finance & ERP orchestration.
• Want more value from data during transformation.
• Requirements: Master data management, data quality standards, critical element flows.
• Solution: Data prep in source systems + augmentation with third-party/internal data.

2. AI/ML & GenAI Initiatives (Most Common)

• Pursuing AI with foundational readiness gaps.
• Gaps: No formal data interaction org, no data stewards/governance, lack of tools/catalog.
• Need: Understand, label, classify, define domain—precursor to AI/ML success.

3. Mature Data-Driven Organizations

• Have data product factory, lakehouse, analytics/data science at scale.
• Goal: Monetization/commercialization—often starts in finance.
• Secondary: Privacy controls, role-based access, wider security.

Biggest Challenges Observed

Challenge 1: Executive-to-Working Level Gap

• DOTS assessment probes perception at all levels.
• Found gaps between senior leaders and working teams—cultural/perception.

Challenge 2: Silos and Lack of Visibility

• Data copied across silos; people unaware of overlapping work.
• Culture, not technology, is the biggest challenge.

Challenge 3: Project Proliferation/Duplication

• Case: Reduced 25 projects to 4 by aligning to data products.
• Solution: Center solutions on data products for efficiency.

Challenge 4: Legacy System Sprawl

• Lack of pruning/retirement of legacy data products.
• Results: Complexity, expense, unclear data landscape.

Challenge 5: Skills and Resources Gap

• More process/resource issue than technology.
• Bottlenecks and prolonged wait times result in lost opportunities.

GenAI Impact on Awareness and Sentiment

Early 2024: Fear, resistance, risk aversion, investment concerns.
Mid-Late 2024: Excitement, rapid adoption, Data Time Machine idea:

• Data wrangling reduced from 40 hrs/week to 5-10 hrs/week.
• 30+ hrs freed for innovation, flywheel pulls projects forward.

Executives:

• Everyone now sees GenAI promise with good data.
• Adoption up due to trusted frameworks, right constructs, market validation.

Recommendations for Leaders (2025+)

1. Active Learning & Monitoring

   • Keep up with advances, vendor releases, industry benchmarks.

2. Build Data Community/Advocacy

   • Data bridges business/IT. Form groups/councils, share social metadata.

3. Create Idea Management Systems

   • “Idea jars” for use cases, unsolved problems, allow tinkering/hackathons.

4. Value Determination Framework

   • Measure impact with “Data Value Chain”, demonstrate ROI.

5. Strategic Vendor/Tool Selection

   • Assess/existing tech, select best-fit tools, reference architecture.

Key Insights Summary

Data as Strategic Asset

• Massive variation in how organizations use/data.
• Exec-working gap; culture + tech transformation needed.

Efficiency Through Alignment

• Project consolidation (25→4) frees resources, boosts innovation.

The Data Time Machine

• GenAI compresses weeks into hours/days, enabling new possibilities.

Human-Centric Innovation

• GenAI at home influences workplace; integration/workflow thinking overtakes dashboards.

Critical Success Factors

• People > Process > Technology
• Governance is business enabler
• Simplification before expansion
• Cross-functional alignment
• Persona-based design
• Active learning
• Value measurement
• Community building

Source: KPMG “You Can With AI” Series | Guest: Daniel Bearinger
Focus: Data Modernization & Enterprise Data Strategy
Date: 2024 | Series: Part of 7-part exploration on AI implementation