By Nick Jewell, Mendelsohn Neil Chan
Published on 2024年9月13日
Retail and commercial banking is evolving fast. Traditional banks are facing fierce competition from digital-native fintech companies and digi-banks. These new players are winning customers with innovative product offerings, deep personalization, and exceptional service levels.
Fifth Third Bank, a prominent regional bank headquartered in Cincinnati, Ohio, has been a leader in the financial services industry since its establishment in 1858. With a diverse portfolio that includes retail banking, commercial banking, consumer lending, and wealth management, the bank serves millions of customers across the Midwest and Southeast United States. As the banking landscape sees increasing adoption of AI and digital technologies, Fifth Third faced the significant challenge of managing vast amounts of data across multiple, diverse systems while ensuring regulatory compliance and maintaining high data quality standards.
Fifth Third Bank has embarked on a comprehensive data governance journey to address these challenges. The bank’s primary objectives were to unify its data management practices across multiple business units, integrate new technologies with existing legacy systems, and cultivate a strong data-driven culture. By focusing on these areas, Fifth Third Bank aims to enhance its operational efficiency and scalability, providing customers with a more personalized and reliable banking experience.
Similarly, another prominent commercial bank lost market share and struggled to keep pace. However, by implementing a comprehensive data governance program and an enterprise-wide data catalog initiative, the bank reversed its fortunes, achieving a 6% increase in market share within a year.
This blog explores how establishing solid data governance fundamentals was the key to turning the ship around and transforming the concept of data governance from a defensive compliance-led program into an offensive, value-driven movement that used data governance fundamentals to develop a competitive data culture.
To address its market share challenges, the commercial bank initiated a data governance program to align everyone with a universal understanding of core metrics. The first step was implementing an enterprise-wide data catalog, creating a 'single pane of glass' for the organization’s data assets. The catalog automated the extraction of metadata across 100,000+ tables and 10,000 BI reports, serving as the bank’s internal version of Google and Wikipedia, significantly reducing the time spent searching for data assets.
The impact was immediate and impressive. The catalog reduced time spent searching for data assets, saving an estimated 180,000 hours of full-time equivalent (FTE) labor productivity. More importantly, it enabled the bank to streamline bureaucratic processes and foster an agile, governed self-service analytics ecosystem. This empowered non-technical business users, accelerating decision-making processes.
This move was crucial in transforming the bank’s operations. With a centralized data governance framework, the bank could finally align on a universal understanding of core metrics and KPIs. This alignment produced more accurate reporting, which, in turn, led to faster and more informed decision-making, with a 50% reduction in time-to-market for new banking products.
Retail banks, including Fifth Third Bank, have traditionally operated with fragmented data systems across various business units—retail banking, commercial banking, wealth management, etc. This fragmentation poses significant challenges for data governance, as inconsistent data management practices can lead to inefficiencies and errors.
Fifth Third Bank has recognized the need to centralize its AI and data operations to streamline governance and improve consistency across the organization. By unifying data across all business units, Fifth Third enhances the consistency and quality of its data and enables scalability. This centralization allows for a more seamless integration of AI technologies across the bank, ensuring that AI-driven insights and decisions are based on a comprehensive, unified view of the data.
For example, centralizing customer data from various touchpoints (e.g., mobile banking, branch visits, credit card transactions) into a unified platform allows Fifth Third to deliver more personalized banking experiences through AI while ensuring that all data complies with the bank’s stringent governance standards.
Research from McKinsey validates the centralization approach, especially regarding initiatives involving generative AI (GenAI). In their paper “Scaling GenAI in Banking,” they report that 70% of organizations following a highly centralized operating model for AI report that their models are now in production instead of more decentralized teams that remain in experimental delivery phases:
McKinsey says the highly centralized approach to GenAI has shown the best results.
Meanwhile, the commercial bank's existing data management practices were also fragmented, leading to inconsistent metrics and inefficient processes that hampered its ability to respond quickly to market demands. This inconsistency slowed decision-making and created a culture of uncertainty and hesitation across the organization.
A senior analyst at the commercial bank highlighted the problem: "Our agility and time-to-market was inhibited because we couldn't totally trust our metrics. Something as basic as loan approval time would show different numbers depending on who was looking at it or which report you checked because the rules or formulas were inconsistent."
Integrating new data catalogs and governance frameworks with legacy systems is a significant challenge in any data governance initiative. Fifth Third Bank has addressed this by taking a flexible, phased approach to integration.
Instead of a complete overhaul, the bank has integrated data governance practices incrementally, ensuring that each system—customer relationship management (CRM), transaction processing, or AI platforms—was aligned with the new governance standards.
For example, when introducing a new AI-driven credit scoring system, Fifth Third ensured that it could seamlessly integrate with the existing customer data platform, maintaining consistent data quality and governance standards. This approach minimized disruptions and ensured the bank could leverage new technologies without compromising data integrity or governance.
Alation offers 100+ enterprise connectors to ensure tight integration with banking systems
Banks of all sizes regularly face data-related challenges that slow their ability to innovate. For example, internal acronyms, abbreviations, and jargon embedded in data sets make it difficult for analysts to comprehend information. Moreover, reverse engineering formulas in critical dashboards is time-consuming, especially when the original developer is unavailable, perhaps due to reassignment or leaving the organization.
With advanced capabilities, including AI and machine learning, Alation analyzes behavioral metadata to guide employees using their data assets. By linking similar data sets based on thematic and semantic associations, Alation helps create new data products, streamlining report generation. As a result, the time required to develop new reports, dashboards, and analyses for business stakeholders can be significantly reduced. For example, our commercial bank saw reduced delivery times of up to 33%, from six weeks to four weeks.
Data quality and governance are critical in retail banking, especially as banks increasingly rely on AI-driven decision-making. Yet, AI systems are only as good as the data they are trained on. At Fifth Third Bank, the importance of high-quality data has been amplified by regulatory scrutiny and the need for precise, reliable AI outputs. For instance, when using AI to assess creditworthiness or detect fraudulent activity, poor data quality could lead to biased outcomes or missed threats, damaging customer trust and putting the bank at risk of regulatory penalties.
Fifth Third Bank implemented robust data governance practices to address these challenges, ensuring that data used in AI applications is accurate, consistent, and compliant with all necessary regulatory requirements. This approach helps mitigate risks associated with AI, such as unintended bias or decision errors, by ensuring that all data feeding into AI systems is rigorously managed and of the highest quality.
The Alation Open Data Quality Framework enables seamless integration with various data quality tools, ensuring continuous monitoring and improvement of data health across the enterprise.
One of the commercial bank's most significant issues was report proliferation—multiple dashboards and reports that duplicated or misrepresented the same metrics, leading to confusion and inefficiency. Using Alation’s lineage capability, data leaders could identify and decommission these duplicate reports, helping the bank eliminate redundancy.
Alation automated the extraction, classification, tagging, and documentation of crucial KPIs and business metrics. This automation reduced the time spent on data reconciliation and validation by 50%, allowing analysts to focus on more value-adding tasks.
Additionally, Alation provided analytics to optimize the bank’s data warehouse costs. Leaders used Alation to identify tables and analytic assets that had not been queried recently and created audit reports around them, allowing the bank to manage its data more efficiently and reduce unnecessary costs.
Fifth Third Bank has leveraged data governance to drive efficiency and productivity across its core banking functions, such as loan origination, customer onboarding, and risk management.
For instance, by ensuring that data is consistent and easily accessible, the bank has reduced the time needed for credit assessments and approvals, speeding up customer service delivery and reducing operational costs.
Fifth Third Bank has minimized errors and rework by integrating data governance into its day-to-day operations, further boosting productivity. What does this look like in practice? It starts with processes that automate routine data validation tasks, freeing up employees to focus on more complex, value-added activities, ultimately leading to improved customer service functions and satisfaction.
One strategic objective for the commercial bank was to improve credit decisions to increase the Lifetime Value (LTV) of underbanked customers and reduce risk through better credit scoring models. By leveraging internal banking data, credit bureau data, partner data, and fund transfer data, the bank aimed to identify upselling opportunities and evaluate the possibility of default among current credit holders.
With Alation, stakeholders across various bank roles—from data engineers to loan officers—could access the same accurate, timely data, leading to better-informed decisions and improved credit scoring models. This initiative helped the bank increase lifetime value while mitigating risk and reducing potential losses.
For our commercial bank, data privacy and cost management were also top priorities for the bank’s governance teams. With a significant volume of Personally Identifiable Information (PII) and sensitive data fields to manage, manual processes were time-consuming and prone to errors. With the Alation platform, automated processes for coding, tagging, and masking PII data ensured 100% compliance with regulatory standards such as GDPR, CCPA, PDPA, and DPDP.
Alation’s tagging and data classification syncs seamlessly with platforms like Snowflake, ensuring consistent data management across financial services organizations.
With a framework in place and tangible benefits to hand, the commercial bank’s vision for the future is ambitious. It aims to streamline bureaucratic processes further to deliver an agile yet governed self-service analytics environment, driving significant service level improvements on critical metrics such as Customer Satisfaction (CSAT), Loan Approval Time, and First Contact Resolution Rate. With an improved data governance framework, the bank intends to compete for market share by quickly identifying innovative product mix and product-bundling opportunities.
Likewise, Fifth Third Bank understands that a strong data culture is essential for effective data governance that scales successfully. The bank has worked to build a data-driven culture by making data governance a shared responsibility across the organization. This includes rebranding traditional roles and concepts to make them more relatable and engaging for employees—for example, transforming the traditional term "data steward" into "data management maven" to reflect the importance of these roles in the bank’s success.
How can you measure the effectiveness of such a cultural shift? Fifth Third uses metrics like data quality scores, adoption rates of data management tools like Alation, and the frequency of data governance training sessions. By aligning these metrics with business goals, the bank ensures that its data governance efforts directly contribute to broader organizational objectives, such as improving customer satisfaction, reducing risk, and enhancing operational efficiency.
Selecting the right data governance platform is critical beyond simply choosing a tool based on check-box functionality or industry analyst positioning. A data governance solution must closely align with your organization’s unique needs and long-term goals.
Even a platform with well-regarded capabilities can become a source of frustration if it doesn’t integrate well with existing systems or if the user adoption process is too complex. Beyond evaluating a platform's technical features, look to ease of use, training requirements, and the overall impact on your team’s productivity. A steep learning curve and high costs associated with rollout and adoption can often be missed during a purely technical vendor evaluation, leading to significant failures when deployed to the enterprise. To this end, many organizations organize a vendor bakeoff process to roadtest the most important functionality for your use cases before selection.
While the banks featured in this article have adopted a plan to centralize and unify their AI-ready data in platforms such as Snowflake or Databricks, there’s a crucial distinction between a policy-driven, centralized approach and a more user-friendly, less-invasive decentralized model for data governance.
While a centralized approach for AI initiatives might seem ideal for maintaining strict controls, it can lead to rigidity and operational inefficiencies when this centralization is applied to the wider data ecosystem. If centralization is taken too far, everyday users can often perceive it as inflexible, leading to a growth in so-called “shadow IT”. On the other hand, platforms that emphasize data discovery and collaboration may offer the flexibility needed to develop a more agile and responsive data governance environment.
In highly regulated retail or commercial banking industries, rigorous compliance trumps all other needs at the risk of financial penalties or sanctions. However, a democratized approach to data governance is often a long-term success that empowers the organization to work more effectively with data. Unlike the “command and control” style of governance enforcement, which never truly captures the spirit of innovation or a thriving data culture, a democratized approach to data governance is often a long-term success that empowers the whole organization to work more effectively with data.
Some form of data governance will likely already be in place to meet statutory regulations, so platform selection is rarely a green-field exercise. Ultimately, deciding to switch platforms or stay with an existing one hinges on clearly understanding how well the solution can deliver on its promises. Cost considerations are significant, but they must be weighed against the potential risks of inadequate governance or prolonged integration issues.
As alternatives are explored, ensure that any platform offers a tangible mix of both short-term savings and long-term value. This will support your organization’s growth and compliance needs without sacrificing usability or support. The right choice will empower your team to achieve operational efficiency and strategic innovation, turning data governance into a competitive advantage.
This blog has explored the data governance journeys of both a retail and commercial bank, leading to significant improvements in their internal processes and their ability to compete in their respective markets.
These outcomes can be attributed to several key strategies for data governance programs:
Take a People-Centric Approach: By focusing on the human element of data governance, including rebranding roles and gamifying data management tasks, Fifth Third has been able to engage employees at all levels and make data governance an integral part of the organizational culture.
Design for Scalability and Flexibility: The bank’s data governance framework is designed to be scalable, allowing for growth and the integration of new technologies without sacrificing data quality or governance standards.
Centralize to Unify: Fifth Third has improved consistency and scalability in its AI and data management operations by centralizing data across business units, leading to more accurate and reliable insights.
Note: Many organizations are also exploring a compelling alternative approach: decentralized data products, often called a data mesh. In this model, instead of just centralizing data into a single repository, data ownership is distributed across various business domains, each responsible for its own data products. Proponents of data mesh argue that this decentralized model can enhance innovation and responsiveness, although this approach can also introduce challenges in maintaining data consistency and governance without careful design.
Integrate Incrementally: Don’t look for a big bang, top-down heavy approach to data governance. Instead, delivering incremental integration of governance practices with existing systems has allowed Fifth Third to maintain business continuity while enhancing its data governance capabilities.
Automate the Mundane: By automating the extraction, classification, and documentation of critical data, the commercial bank significantly reduced manual effort and errors. This led to a 50% reduction in data reconciliation and validation time, freeing analysts to focus on strategic tasks.
Aim for Continuous Improvement: Fifth Third regularly revisits and refines its data governance practices to ensure they remain aligned with business goals and regulatory requirements, ensuring long-term sustainability and success.
Through these strategies, financial services institutions of all sizes can successfully navigate the complexities of data governance, transforming data governance from a defensive ‘compliance’ necessity into a competitive driver of value and business success.
Having laid the foundation, our next step is translating these concepts into actionable steps. This approach will allow us to validate the framework, refine our strategies, and drive tangible outcomes that align with the organization’s data governance objectives.
Let’s explore the five critical activities for implementing data governance in retail banking.
Organize related data sources into domains that align with key business processes or capabilities. For instance, establish a Customer Domain encompassing CRM, marketing automation, and customer support data; a Product Domain integrating product catalogs, inventory, and sales data; and a Financial Domain managing accounting, sales, and budgeting information. This approach ensures data domains reflect the business landscape accurately, enhancing data accessibility and relevance for end-users.
An example of a hierarchical two-layer data domain structure is shown below, focusing on how a retail bank’s lending business can be organized into logical data domains.
(Pro tip: We typically recommend only going to a maximum of 3 levels deep; anything beyond that can be too granular)
A sample data domain structure for a retail banking lending business, organized into two tiers.
Alation is uniquely equipped to streamline data organization by structuring information into key domains that mirror your business’s core processes and capabilities. This domain-centric approach ensures that data and knowledge are accurately categorized, making it easier for users to find and utilize relevant information and strengthening governance practices. Below is an example of how the theoretical framework above can be applied in practice in the Alation Data Intelligence Platform:
Using Alation to organize data and assets into domains and subdomains for effective search, discovery, and governance.
Now that we’ve created our data domain structure, the next step is identifying the most important KPIs and metrics to assess and monitor business performance in each domain and subdomain.
Measures are numerical data points that can be aggregated, calculated, and analyzed to provide insights into business activities. They are the metrics that answer questions like “how much,” “how many,” and “how often” and are essential for creating reports, dashboards, and other analytical tools.
Key Characteristics of Measures:
Quantitative Data: Measures are always numerical values, such as sales revenue, profit, costs, number of units sold, or average order value (AOV).
Aggregatable: Measures can be summed, averaged, counted, or otherwise aggregated to provide meaningful insights at various levels (e.g., by time period, geography, product).
Contextualized by Dimensions: Measures gain context and meaning when analyzed alongside dimensions (e.g., time, location, product categories) to show trends and patterns.
Time Intelligence: Measures are frequently analyzed with a time component, such as Year-over-Year (YoY) growth, Year-to-Date (YTD), Month-to-Date (MTD), or Rolling periods, providing insights into performance trends over time.
In our retail banking example, metrics that are relevant to the loans or credit-related domains of the business include:
Loss Given Default
Segmented LGD
Recovery Rate
Exposure at Risk
Probability of Default
Delinquency Rate
Loss Rate
Credit Utilization %
Debt to Income Ratio
Loan to Value Ratio
Alation goes beyond being a mere repository for KPI and metric documentation; it enriches this information by integrating both technical metadata and practical business context. For example, in addition to core components like the metric’s definition and formula, Alation enhances this data with metadata classifications and tags, creating a richer, more navigable resource.
It also identifies and lists the key fields required to calculate each metric from specific datasets, bridging the gap between technical details and business relevance. This holistic approach empowers users to fully understand and leverage their data, driving more informed decision-making across the organization.
Using Alation to document and organize critical KPIs and metrics ensures consistent team usage.
Cataloging domain-specific jargon, acronyms, and abbreviations is essential because database and data warehouse tables often contain fields with technical or obscure labels that business users do not easily understand.
These non-business-friendly terms can lead to confusion, misinterpretation, and errors when analyzing data, as users struggle to connect the data fields to their real-world business context. By cataloging and clearly defining these terms, organizations can bridge the gap between technical data and business understanding, making data more accessible, interpretable, and valuable for decision-making. This practice enhances data transparency and empowers users to confidently leverage data for insights without getting lost in unfamiliar terminology.
For our retail banking example, we collect common acronyms and abbreviations found in the loans and credit business units, such as:
TXN = transaction
AMT = amount
GL = general ledger
CCY = currency
CASA = current account savings accounts
APR = annual percent rate
EAD = exposure at default
LGD = loss given default
LTV = loan-to-value ratio
DTI = debt to income ratio
Alation transforms complex banking-related acronyms, abbreviations, and shorthand naming conventions into clear, business-friendly titles, making data more accessible and interpretable for everyone in the organization.
By automating this process, Alation ensures consistency and clarity across tens of thousands of tables and fields, breaking down silos and enabling seamless communication between technical and non-technical stakeholders. This scalability not only saves time but also enhances data comprehension, fostering a culture where insights are easily understood and actionable by all users, regardless of their technical expertise.
Alation delivers suggested descriptions across thousands of columns and tables using ALLIE AI for intelligent curation at scale.
Identifying business terms and concepts is crucial in ensuring that data across the organization is consistently understood and applied. A well-defined business glossary is a centralized repository for standardizing terminology, providing clear definitions, and establishing a common language across departments. This reduces ambiguity and miscommunication, enabling all stakeholders—from data analysts to business leaders—to interpret data similarly.
For instance, it’s common for a metric to have varying definitions across different data domains. A typical example is “Days Past Due” (DPD), which refers to the number of days a borrower has missed or delayed payments, such as loan monthly installments or credit card dues. Documenting and highlighting these subtle differences, especially among various credit products, is critical to avoid confusion when reporting these figures to management or regulators.
In our lending domain, common examples of business term definitions include:
Collateral: Assets pledged by a borrower to secure a loan, which can be seized by the lender in case of default.
Delinquency Rate: The percentage of delinquent loans within a portfolio, serving as an indicator of credit risk within the portfolio.
Charge-Off Rate: The percentage of loans written off as a loss due to non-repayment, indicating the level of bad debt.
Credit Exposure: The total risk a lender faces from borrowers, including all outstanding loans and credit lines.
Stress Testing: A simulation technique used to assess the impact of adverse economic conditions on a borrower’s ability to repay or the overall loan portfolio.
Default Rate: The rate at which borrowers fail to make scheduled loan payments is a key indicator of credit risk.
Alation Business Glossary centralizes the creation of terms such as metrics and other essential definitions.
Documenting relevant regulatory and governance policies in a data catalog ensures compliance and promotes responsible data management across the organization. These policies should be clearly defined, thoroughly documented, and directly linked to relevant data assets, providing users with essential guidance on how data should be handled, shared, and used in line with legal and organizational standards.
To streamline this process, you can create logical groupings called “Policy Categories,” such as Regulatory Policies, Data Sharing Policies, Data Usage Policies, and Data Lifecycle Policies. The data catalog enhances data governance by categorizing and associating policies with specific data assets. It helps users quickly understand the rules and constraints that apply, fostering a culture of compliance and accountability.
Here are examples of relevant policies in the context of the retail banking industry, outlined under the four policy categories:
1. Regulatory Policies:
The Truth in Lending Act (TILA) affects data assets related to loan disclosures, ensuring loan terms, interest rates, and fee transparency.
The Gramm-Leach-Bliley Act (GLBA) Governs the protection of consumer personal information, impacting data handling and privacy for customer accounts.
The Fair Credit Reporting Act (FCRA) regulates the collection, dissemination, and use of consumer credit information, affecting data assets related to credit reporting and loan approvals.
2. Data Sharing Policies:
Customer Consent Requirements: This defines the conditions under which customer data can be shared with third parties, impacting data-sharing agreements and consents.
Interbank Data Sharing Agreements: This governs data exchange between banks for credit assessments or fraud prevention, affecting shared data on customer creditworthiness.
Third-Party Vendor Access Policies: This policy regulates data access permissions for third-party service providers, impacting external data integrations and security measures.
3. Data Usage Policies:
Permissible Use of Customer Data: Defines acceptable use cases for customer data, such as marketing, customer service, or risk assessment, ensuring data usage aligns with customer expectations and consent.
Data Anonymization and Masking Requirements: This section outlines when and how customer data must be anonymized or masked, particularly when used for analytics or testing purposes.
Access Control Policies: Establishes guidelines for who can access specific data assets based on role and purpose, ensuring data is used only by authorized personnel.
4. Data Lifecycle Policies:
Data Retention and Deletion Policies: This policy sets the timelines for retaining and disposing of data, impacting data related to closed accounts, expired loans, or outdated customer records.
Data Archiving Standards: Defines procedures for archiving inactive data while ensuring that archived data remains accessible when needed for compliance or audit purposes.
Policies are documented, governed, and maintained within Alation, and are closely associated with data domains and related data assets
After these core components have been defined and documented, it is time to assemble them and integrate them within the enterprise data catalog. This process involves linking Data Domains that reflect your business structure, defining and associating key metrics with their relevant domains, cataloging domain-specific jargon, acronyms, and abbreviations, and incorporating business glossary terms and concepts to ensure clarity and consistency across the organization. Documenting the catalog's relevant regulatory and governance policies provides a comprehensive view of data rules and standards.
By associating these components based on their domains and relationships, the enterprise data catalog evolves from a static repository of information into a dynamic and interconnected framework representing your entire data ecosystem.
This approach enables users to see not just individual data elements but the full context in which they operate, from business relevance and data lineage to compliance requirements and key performance indicators. The result is a fully integrated data product that serves as a single source of truth, enhancing data discoverability, usability, and governance and ultimately driving data-driven decision-making across the enterprise.
Ready to learn more?
Click here for a guided demo of Alation, the leading data intelligence platform for financial services.
Watch this webinar to explore how Fifth Third Bank successfully transformed its data governance approach using Alation to implement a data mesh strategy.