Cracking the Data

Modeling Code
A Comprehensive Guide To
Data Readiness For Banks

Inspiring Better Banking

1 | External Document © 2024 EdgeVerve Systems Limited
www.finacle.com
 Data is the new goldmine – rich with insights that can propel an organization to new heights
of efficiency and innovation. Banks, with their vast troves of datasets, are uniquely positioned to
harness this potential and unlock transformative opportunities. However, despite this potential,

Data is expanding at most banks are far from fully capitalizing on their data assets.

an unprecedented Recent research by Infosys Finacle and Qorus reveals a significant gap – only 17% of banks rated
themselves highly in their ability to extract value from data, analytics, and AI. This struggle is often

pace, but are banks rooted in the complexity and fragmentation of data sources, each with its own structure and
definitions. Such disarray leads to errors, inconsistencies, and inefficiencies, all of which impede

truly ‘data-ready’? banks’ ability to manage and leverage data effectively. Moreover, as data continues to expand,
the reliance on outdated, ad-hoc data models has become unsustainable, eroding operational
efficiency, decision-making capabilities, and the ability to seize emerging opportunities.

To turn the tide, banks must first lay the groundwork by becoming truly ‘data-ready’. This requires
embracing modern, gold-standard data modeling practices that offer a unified framework for
organizing, defining, and managing data. By doing so, banks can build a solid foundation that not
only supports data-readiness but also positions them to fully exploit the wealth of opportunities
their data holds.

This thought paper explores the critical role of data models, best practices in data modeling,
the adoption of data modeling standards like BIAN, and the implementation of advanced
data platforms. It serves as a comprehensive guide for banks aiming to establish a robust data
infrastructure, enabling them to achieve data-readiness and fully leverage the potential of
their data assets.
 Table of Contents 1. The Data Conundrum: Understanding limitations with current
systems better

2. Understanding Anatomy of a Data Model

3. How Streamlined Data Modeling Makes a Difference?

4. Finding Success with Data Models: A Reference Architecture Blueprint

5. How a Proven, Next-Gen Platform can Unleash the Potential of

Data Modeling for Banks

3 | External Document © 2024 EdgeVerve Systems Limited

 The Data Conundrum: Understanding limitations
with current systems better
Case in Point: Loan officers often find themselves sifting through data definitions and formats further impede integration and analysis,
documentation scattered across various sources, such as core banking making it difficult to gain meaningful insights.
systems, credit bureau reports, and customer relationship management Data Quality and Integrity: Poor data quality manifested in errors,
(CRM) software. Each of these systems may use slightly different data duplicates, and missing values undermines the reliability of
formats and terminology, making the task of extracting and consolidating any insights derived from it. Without a standardized approach to data
information both time-consuming and labor-intensive. For instance, one modeling, ensuring data accuracy becomes an uphill battle.
system might list “income” as “gross salary,” while another refers to it as
Operational Inefficiencies: The manual effort required to extract,
“annual income.” These inconsistencies in data definitions add another layer
transform, and load data from multiple sources is both
of complexity, leading to frustration and, worse, inaccurate assessments.
time-consuming and prone to errors, impacting banks’ agility and
The absence of a unified data model not only hinders efficiency but also
competitivenesss.
jeopardizes the accuracy of lending decisions.
Increased Costs: Managing disparate data sources and resolving data
The scenario above vividly illustrates how inadequate data management
quality issues often demands substantial investment in data cleansing,
systems can become a significant bottleneck, stifling operational efficiency.
integration, and reconciliation efforts.
The key challenges associated with current approaches include:
These challenges underscore the need for a clear, holistic understanding
Data Silos: Data is often trapped within isolated systems, limiting its of data encompassing its type, quantity, interrelationships, organization,
accessibility and preventing a holistic view of operations. Inconsistent storage, and application. One of the primary objectives of data modeling is
to provide this clarity.

4 | External Document © 2024 EdgeVerve Systems Limited

 Understanding Anatomy of a Data Model
A data model is a blueprint that organizes data elements and defines their interrelationships, enabling efficient access, manipulation, and retrieval of
information. It also standardizes system components, such as entities, associations, and requirements, ensuring consistency across the organization.

A well-constructed data model, often visualized through a diagram, allows both business and technical teams to collaboratively design how data will
be stored, accessed, shared, updated, and leveraged. This collaborative approach is crucial for creating robust databases and software applications that
align with business needs.

Components of a Data Model

Entity Attributes Entity Relationship

A specific object Fields that provide additional A depiction of how different
representing a well-defined details about an entity. Each entities are connected or
singular concept, such as entity has one or more attributes, related to one another within
accounts, transactions, or with the primary key being the the model.
customers. attribute that uniquely identifies
the entity.

5 | External Document © 2024 EdgeVerve Systems Limited

 Types of Data Models
Data models are typically categorized based on their level of abstraction:

Conceptual Model
Also known as domain models, these high-level representations conceptualize how database tables
or entities will be structured. The conceptual model illustrates the relationships between entities,
driven entirely by business requirements from stakeholders and end-users. Although abstract, this
model offers business stakeholders a clear view of how the entities will be designed to meet business
objectives.

Logical Model
This model delves deeper into the details, specifying entities and their attributes, as well as the
relationships between them. The logical model includes information about primary and foreign keys,
offering a less abstract and more detailed view of the domain’s structure. It bridges the gap between
business concepts and technical implementation.

Physical Model
The physical model provides a concrete schema for how data will be physically stored within
a database. It offers a final design that can be implemented as a relational database, detailing
associative tables, relationships between entities, and the primary and foreign keys that maintain
those relationships. Data modeling tools, such as Erwin, can generate Data Definition Language
(DDL) scripts directly from physical data models, streamlining the implementation process.

6 | External Document © 2024 EdgeVerve Systems Limited

 How Streamlined Data Modeling Makes a Difference?
Streamlined data modeling provides a clear framework that allows business stakeholders, developers, data architects, and analysts to visualize and
understand the intricate relationships within a database or data warehouse. This clarity not only improves collaboration but also delivers a wide range of
benefits, including:

Potential Use Cases: By clarifying data relationships and dependencies, Enhanced Collaboration: A clear data model improves
data modeling helps identify business opportunities and informs the information sharing between developers and business
development of data marts, dashboards, and machine learning models. intelligence teams, fostering stronger collaboration and
Error Reduction: A well-defined data model acts as a blueprint, alignment.
reducing errors in software and database development, and ensuring Accelerated Database Design: By providing a clear conceptual
consistency and reliability across systems. and logical view of data, data modeling speeds up the database
Enhanced Consistency: It promotes uniformity in documentation and design process.
system design throughout the organization, strengthening Scalability: A robust data model accommodates future
overall data governance. growth and changes in data requirements without causing
Improved Performance: Optimized data structures and relationships major disruptions.
lead to better application and database performance, enhancing Compliance Adherence: Ensuring data consistency and integrity
operational efficiency. through standardized models helps organizations meet data
Simplified Data Mapping: Data modeling eases the process governance and regulatory requirements.
of mapping data between different systems and applications,
streamlining integration efforts.

7 | External Document © 2024 EdgeVerve Systems Limited

 Leveraging Streamlined Data Modeling in Banking:
Business Use Cases

Fraud Detection: Developing a data model that integrates behavioral patterns,

transaction history, and external fraud intelligence helps banks identify fraudulent
activities, money laundering, and other suspicious behaviors.

Loan Risk Assessment: Creating a data model that evaluates loan applicants’
creditworthiness by considering factors such as income, employment history, debt-to-
income ratio, and repayment history ensures more accurate risk assessments.

Customer 360 View: Building a comprehensive customer profile by integrating data

from various sources such as accounts, transactions, demographics, and interactions
enables personalized products, services, marketing, risk assessment, and
customer service.

Enterprise Revenue Management: Building a comprehensive data model can help

banks create a custom revenue management capability by capturing relevant data
such as customer information, transaction and account details to build customer-
specific charging, billing, and collection reports.

Asset Liability Management: Developing a data model, banks can establish a

comprehensive liquidity and interest rate risk management solution to deliver an
enterprise-wide view of all on-balance sheet and off-balance sheet exposures, ensuring
better funding and liquidity decisions.

8 | External Document © 2024 EdgeVerve Systems Limited

 Finding Success with Data Models:
A Reference Architecture Blueprint
The following architectural blueprint provides a guide for banks to build robust
data management systems and implement an effective data modeling strategy.

Cloud Storage Repositories

Data Warehouse: A structured data management system that stores current
and historical data from multiple sources (e.g., point-of-sale systems,
marketing databases) for analysis and reporting. Ideal for business intelligence (BI)
and data mining, data warehouses excel at handling complex queries on large
datasets but struggle with new data formats like audio or text.
Data Lake: An unstructured repository that stores vast amounts of raw data in
its native format. Data lakes allow for easy storage and access without predefined
schemas, making them flexible but less structured, lacking built-in support for
ACID transactions.
Data Lakehouse: A hybrid architecture that combines the flexibility of data lakes
with the data management capabilities of data warehouses, supporting both
business intelligence and machine learning. By merging these systems, data
lakehouses offer a unified platform for accessing complete, up-to-date data.

9 | External Document © 2024 EdgeVerve Systems Limited

 Other Data Storage Architectures

Data Warehouse Data Lake Data Lakehouse

Delta Lake: An open storage layer
that unifies data types for transactional,
analytical, and AI use cases. It supports
streaming and batch operations, making it a BI Reports BI Reports Data Machine BI Reports Data Machine
Science Learning Science Learning
foundation for scalable, cost-effective
data lakehouses.

Data warehouses Data warehouses

Data Mesh: A decentralized architecture Metadata and

that integrates data from various business Governance Layer
ETL
lines for analytics, following principles like ETL

domain-oriented ownership, data as a ETL Data Lake

product, and federated governance. Data Lake

Structured Data Structured, Semi-structured and Unstructured Data

Data Processing Systems

Online Analytical Processing (OLAP) and Online Transaction Processing efficient analytics, with availability being a lower priority.
(OLTP) are two primary data processing systems used in data science: OLTP (Online Transaction Processing): Designed for real-time
transaction processing, OLTP systems prioritize data-write operations,
OLAP (Online Analytical Processing): Optimized for complex queries and
ensuring integrity in high-frequency, high-volume transactions.
trend analysis, OLAP systems prioritize data-read operations for

10 | External Document © 2024 EdgeVerve Systems Limited

 Layers of Data Storage
Bronze Layer: The landing zone for raw, Improve Data Quality
immutable data from external sources,
maintaining historical transactions.
Raw Data Bronze Silver Gold
Silver Layer: A validated, enriched
Batch BI
version of data, where merging or
appending from various sources occurs
to create a comprehensive overview.
Gold Layer: Project-specific Raw Filtered, Cleaned, Business-Level
Integration Augmented Aggregates
databases where data is transformed
for consumption, including activities like
filtering, aggregating, and answering Streaming ML
specific questions.
“Landing zone” for raw data, no Define structure, enforce schema, Deliver continuously updated, clean
schema needed evolve schema as needed data to downstream users and apps

11 | External Document © 2024 EdgeVerve Systems Limited

 BIAN Standards: Framework for interoperability
Since a siloed product has its own database structure, there is need
of canonical structure that can be used as a translation hub for
Party
any other database structure. The Banking Industry Architecture
Network (BIAN), an ecosystem of leading banks, technology
providers, consultants, and academics, was created to establish,
promote and provide a common framework, and serve as a world-
class reference point for banking interoperability issues. This Accounts
standardized approach eliminates inconsistencies and ambiguities,
ensuring data accuracy and reliability.

Taking inspiration from the BIAN framework, it is possible to create

interoperable data models that may be defined for a certain use Interest Details Charges Limits Transaction Balance Details
case and vertical but can subsequently be used in other use cases
needing similar insights. This allows banks to quickly adapt to
changing market conditions and introduce new products and
Operational Accounts Deposits Loans
services with speed.

Ultimately, BIAN standards contribute to a more efficient, A sample ER diagram based on BIAN for Core Banking
effective, and customer-centric banking ecosystem by achieving
standardization and interoperability.

12 | External Document © 2024 EdgeVerve Systems Limited

 Getting it Right: Some key considerations
Alignment with Business Needs: Ensure data models align with business
goals and industry standards such as BIAN for consistency within the
financial industry.
Data Quality Management: Implement rigorous data quality checks
throughout the data modeling lifecycle.
Stakeholder Engagement: Involve key stakeholders from different
departments to ensure clear communication and alignment with
objectives.
Iterative Development: Continuously refine and maintain data models to
reflect changes in the business environment.
Selecting the Right Tools and Platform: Choose the appropriate data
modeling approach and tools based on business needs, data types, and
desired platform architecture.

13 | External Document © 2024 EdgeVerve Systems Limited

 How a Proven, Next-Gen Platform can Unleash the
Potential of Data Modeling for Banks
As banks continue to navigate the complexities of a data-driven world, the organization. They allow data to be segmented into domain-specific
need for a robust, Next-Gen Platform to harness the full potential of data marts, providing precise insights tailored to business needs, and
modeling becomes increasingly critical. To truly unlock the value of data, enhancing overall data management.
banks must invest in an end-to-end integrated platform that combines
Seamless Data Pipeline:
advanced data management capabilities with seamless integration across
A seamless pipeline automates data capture, extraction, and delivery,
various systems and processes.
ensuring data remains accurate and reliable. Real-time processing
Below are some characteristics of a Next-Gen Data Platform: capabilities enable quick responses to market changes, keeping data
current and actionable.
Integrated, Modular Data Lakehouse:
This architecture combines the strengths of data lakes and data Layered OLTP-OLAP Architecture:
warehouses, offering flexibility and scalability to handle all data types This architecture optimizes both transactional processing and analytical
efficiently. It unifies storage and processing, reducing complexity and queries, ensuring data is continually refined and ready for use. It supports
making data management more streamlined for banks. high-volume transactions and complex analyses, enabling informed
decision-making and operational efficiency.
BIAN-Inspired Data Models and Domain-Based Data Marts:
BIAN-inspired models ensure consistency and standardization across the

14 | External Document © 2024 EdgeVerve Systems Limited

 Finacle Data Platform – Functional architecture
Capture/Emit Stream Curate Consume

Finacle Data Platform Rest APIs

OLTP
Finacle/Non - Finacle Applications

Finacle/Non - Finacle Applications

database Finacle Data Pipeline Business dashboards
and reports
Real- time
Files Ingest Clean Transform
event AI/ML models

Connectors
streaming Tailored data
APIs/ Deduplication | Pre-modeling checks | Embedded analytics
extraction | Metadata Consistency checks Model selection |
webhooks Real- time management | Pre - | Formats Aggregation | Normalization Enterprise analytics
event sourcing processing | Encoding engine
Event Batch feeds Query Engine – SQL, JSON, Spark Developer apps
streams
Finacle Data Lakehouse Downstream
Others systems
Unify Store Access
Structured and BIAN data In memory domain Synthetic datasets
…. specific data-marts
unstructured data sets models ..

Data Observability
Dashboards | Notifications | Alerts | Log streaming
Data Governance
Classification | Quality frameworks | Metadata management | Lineage tracking | Policy enforcement | Audit | Compliance
Data Security and Privacy
Access control | Encryption | Authentication | Data retention | Dynamic masking | Zero trust

External Logos used are registered trademarks of the respective organizations

15 | External Document © 2024 EdgeVerve Systems Limited

 How a Leading Indian Private Bank Streamlined Pricing
and Billing Leveraging a Next-Gen Data Platform

The bank was looking to create a personalized solution for billing customers based on
their ongoing relationship. To do this, it needed to capture specific data that would help to
assign value to every customer relationship.

Enter Next-Gen Data Platform

To manage the huge accounts metadata and transactions-related data required for
analysis, a silver layer data model was created to address customer-specific requirements.
The Finacle Data Platform created a BIAN-inspired pricing and billing data model
to capture relevant data for analysis and to deliver insights to create flexible pricing
capabilities.

The physical data model generated DDL scripts to quickly design and create database
schema. The silver layer data model enabled data layer reusability so that the model could
be extended to other use cases.

16 | External Document © 2024 EdgeVerve Systems Limited

 In Conclusion
Data modeling provides banks with a structured approach to organize,
manage, and leverage their data effectively, enhancing operations
like customer service and fraud detection. Banks must now focus
on adopting robust data modeling frameworks, guided by industry
standards to unlock the full potential of their data. Finacle’s Next-
Gen Data Platform is an end-to-end, integrated platform that helps
banks unleash the true data potential. Powered by BIAN-inspired
data models, Data Lakehouse, Data Pipeline, and robust architectural
Author Contributor
capabilities unlocks new dimensions of insights-driven banking. Get in
touch with us today to learn more about how Finacle’s Next-Gen Data Amit Kumar Sharma Arjun Singh
Platform helps banks become truly “data-ready”. Product Mangement Content Marketing
Infosys Finacle Infosys Finacle

17 | External Document © 2024 EdgeVerve Systems Limited

 [email protected]

www.finacle.com

www.linkedin.com/company/finacle

twitter.com/finacle

. . . . . . . . . . . . . . . . . . . . . . ........................................................................................................................................................ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .............................................................................................................................................................................................................................................

For more information, contact [email protected] www. finacle.com

. . . . . . . . . . . . . . . . . . . . . . ........................................................................................................................................................ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .............................................................................................................................................................................................................................................