AI is the engine, Data the fuel, and Data Governance the refinery

Businesses today are captivated by AI, investing significantly in technology as if it's a silver bullet for their challenges. Yet many forget the foundational element crucial to any data-driven approach: good data governance. Without strong data management, investments in cutting-edge technology risk being a false economy. We are currently at the peak of the AI hype curve, and whilst AI is powerful, its true potential is entirely dependent on robust, quality data.

73% of UK organisations say data governance is the cornerstone to better AI implementation

Getting your data governance foundations right isn't just prudent, it’s transformative. It exponentially increases data intelligence, drastically reduces risk, and enhances process efficiency. Remember, technology itself is never the ultimate goal. The real objective is consistently great user experiences and tangible business outcomes.

"You Can't Build a House on Sand"

Data Governance: The Bedrock of Your Business

Data governance ensures the quality, consistency, security, and accessibility of data across your organisation. It's vital because every business decision, customer interaction, and technical process relies heavily on data quality. Without effective data governance, businesses risk making decisions based on unreliable information, leading to inefficiencies and missed opportunities.

71% of organisations report having a data governance program in place in 2024, compared to 60% in 2023

The Core Pillars of Your Data Governance Framework

A robust Data Governance Framework typically includes:

• Data Quality Management: Ensures reliability and accuracy of your data, enabling better decision-making.

• Data Security and Compliance: Protects data against breaches and ensures regulatory compliance, reducing legal risks and safeguarding reputation.

• Data Stewardship and Ownership: Clearly defined responsibilities for data management and improvement, fostering accountability.

• Data Lifecycle Management: Effective control of data throughout its lifecycle, enhancing operational efficiency.

• Metadata Management: Documenting and organising your data landscape, significantly improving discoverability and usability.

• Data Architecture and Integration: Ensures data flows effectively between systems, reducing complexity and improving agility.

Each of these elements directly contributes to a healthier, more agile organisation.

75% of UK businesses waste an average of 14% of revenue on bad data quality

AI Success Hinges on Data Governance

Good AI models, as any data-driven technology, are entirely dependent on the data they're trained on. Without proper governance, data quality issues can propagate and compromise AI outcomes. Poor data quality means unreliable AI predictions, while unchecked biases in data can lead to unfair or discriminatory model outcomes.

44% of UK businesses reported that incomplete, missing data is their most common problem

Another governance challenge is the traceability of AI decisions - the infamous "Black Box" problem. Effective data governance ensures transparency and traceability, providing confidence and compliance in AI-driven decisions.

Data Governance done right (click to expand)

Procter & Gamble

P&G faced significant challenges managing data across global and regional operational systems with 48 SAP instances in various downstream application servers. Their data governance team deployed data quality software to improve the quality and control of their master data, which included over 32 unique SAP instances and billions of records.

This resulted in:

• Improved productivity by automating processes that previously required manual data integration

• Reduced operational risks by minimizing data leakage and duplication

• Actionable insights via dashboards providing timely access to health reports and performance metrics

Uber

Uber, which operates in over 10,000 cities and provides more than 18 million rides daily, stores 256 petabytes of data and processes 35 petabytes every day. They support 12,000 monthly active analytics users who conduct over 500,000 daily queries.

Uber deployed Presto as a distributed SQL query engine to handle their massive data analytics needs, resulting in:

• Federated queries enabling data access from multiple sources

• Workload management for data quality

• Real-time queries for data analytics

• Custom data governance functions tailored to their business needs

JPMorgan Chase

JPMorgan Chase implemented a data mesh architecture to address challenges with dispersed data across different business lines and functions.

In this architecture:

• Data flows are tracked and logged via the mesh catalog

• Data product owners curate data for reporting purposes within dedicated data lakes

• Data consumers access data through the enterprise data catalog This approach improved data tracking, auditing, and cross-enterprise visibility.

Data Governance going wrong (click to expand)

Citigroup's Data Governance Disaster

Citigroup has been involved in a series of compliance failures dating back to 2013, resulting in over $1.5 billion in fines paid to US regulators for lapses in risk management and other key areas. The core issue lies in outdated, fragmented technology and decades of underinvestment:

• In 2020, the Federal Reserve and OCC fined Citigroup $400 million for deficiencies in compliance and data governance

• In May 2024, Citibank was fined £62 ($82.8) million by British financial regulators for failing to catch a $1.4 billion trading error

• In June 2024, they faced another $136 million penalty for inaccurately reporting loans to regulators

• Since 2021, Citigroup has spent over $7.4 billion to overhaul its technology

Business impacts include:

• It takes an average of nine days to open a new account, compared to an industry standard of just three days

• Citigroup's wealth management business has a wallet share of just 13%, significantly lower than the 64% average among competitors

Security Company's S3 Bucket Exposure

A multinational security company left highly sensitive data in one of its Amazon S3 buckets exposed, allowing anyone to access nearly 3 TB of company data without authentication. The company stored approximately 1.5 million files containing information relating to its employees and business operations. The data leak was discovered on October 28, 2021, and included:

• Photographs of ID cards of airport and security staff showing personal information

• Sensitive information in photos of aircraft and luggage in baggage handling areas

• EXIF metadata containing time, date, and GPS locations

Potential impacts included:

• Risk to travelers and airport staff security

• Possibility of criminals creating fake staff ID cards

• Potential sanctions for breaching data protection regulations in Colombia and Peru, including financial penalties up to US$400,000 and US$122,000 respectively

Unity Technologies' $110M Ad Targeting Error

Unity Technologies experienced a significant data quality incident in Q1 2022 when their Audience Pinpoint tool ingested bad data from a large customer, causing major inaccuracies in the training sets for its predictive ML algorithms. This resulted in:

• A loss of approximately $110 million, including direct revenue impact, recovery costs, and delayed launches

• Unity shares dropping by 37%

• Investors "losing faith" in the company's strategy

Practical First Steps: Your Roadmap to Better Governance

Start your journey towards effective data governance with these actionable steps:

• Assess Your Current State: Conduct an audit of existing data governance practices and identify key areas for improvement.

• Establish Ownership: Appoint data stewards and owners who will take accountability for data governance.

• Implement Quick Wins: Prioritise initiatives that deliver immediate visible improvements in data quality or compliance.

• Invest in Training: Equip your team with the necessary skills and knowledge to support your data governance strategy.

• Iterate and Refine: Regularly review progress against defined KPIs and refine your approach based on feedback and evolving business needs.

Recommendations to Prove Your Data Governance ROI

Measure your data governance success just as rigorously as other critical business KPIs:

• Track improvements in data availability and usability, measuring how these enhance productivity.

• Measure the risks mitigated by your data governance initiatives, directly correlating reduced risks to financial or operational benefits.

• Manage data quality issues similarly to software bugs, prioritising resolution based on severity and impact.

• Treat data governance policy breaches with the same urgency as information security incidents, clearly highlighting their significance.

By adopting these metrics, you'll build a robust business case proving the tangible ROI of your data governance efforts.

---

How I can help...

With over 25 years experience in providing Technology and Data strategic thinking, solutions and nurturing technical teams, I can help you with these challenges.

If this engagement resonated with you, or if you have your own challenge you'd like to discuss, feel free to reach out through the usual channels, and we can get the conversation started.

contact@andyryanconsulting.co.uk