• History of Data Management
• Business Drivers for implementation of data governance • Building Data Strategy & Governance Framework
• Data Management Maturity Models
• Data Quality Management
• Metadata and Governance
• Metadata Management
• Data Governance Stakeholder Communication Strategy
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Why data governance is the new buzz?
1. Why Data Governance is the
new Buzz?
By,
Abhishek Prasad
Senior Data Governance & Strategy Consultant
Email: abhishek.prasad@Osthus.com
OSTHUS GmbH
2. Slide 2
Agenda
• History of Data Management
• Business Drivers for implementation of data governance
• Building Data Strategy & Governance Framework
• Data Management Maturity Models
• Data Quality Management
• Metadata and Governance
• Metadata Management
• Data Governance Stakeholder Communication Strategy
CONFIDENTIAL
3. Slide 3 CONFIDENTIAL
1960s
ADAPSO introduces the
concept of “Data
Management” for training &
quality assurance metrics.
1950s
Computers were
slow, clumsy & labour
intensive to operate.
1970s
SQL was developed by
Edgar F. Codd with focus on
relational databases.
1998
NoSQL was introduced by
Carlo Strozzi
gained popularity late 2005
2002
Cloud Data Management
picks up the trend
with players like Salesforce &
AWS.
2018
Deep Learning , AI,
Augmented Reality are now
part of the curriculum at
most of the universities.
History of Data Management
UK Banking
Crisis-73-75
Wall Street
Crash of
1929
Dot Com
bubble
US Subprime
crisis’07-’10 &
EU Debt
Crisis’19
Corona
virus
20201980s
DAMA cam into existence &
worked towards framing data
mgmt. standards, literacy
IDC predicts
By 2025 users
will be
creating 463
billion GB
data/ day.
Note: “ADAPSO”-Association of Data Processing Service Organizations
4. Slide 4 CONFIDENTIAL
Data Drivers & Challenges - Business stakeholder perceptions
CONFIDENTIAL
Business: “It costs too much to run our data estate!”
True: Direct data costs i.e. the costs to run the numerous databases & data
warehouses that support the business generally runs in millions for big organizations.
Business: “IT is too slow to deliver new products & services”
True: Due to the number of disparate & duplicate data sources, all new requests pay
an implicit “change tax” – especially requests that require multi-region or are global in
nature.
Business: We have poor data quality
False: Data quality is very subjective i.e. what is acceptable to one business function
is unacceptable to another.
5. Slide 5 CONFIDENTIALCONFIDENTIAL
Business: “How do I store all this research data economically so that it does not get
lost?”
True: Cost of data archiving is high. E.g. Physical Storage vs Cloud Storage.
Data Drivers & Challenges - Business stakeholder perceptions
Business: “We can’t find the data we need i.e. where do I go for e.g.
VaccinesClinical Trails data?”
True: Historically both the business and IT have been structured by business
functionproducts and also regionally, which has lead to a proliferation of processes and
support IT & data architecture. On an average 80% of the time spent on an analysis
project is spent on 'cleaning' our data - filling in missing information and confirming
meaning.
Business: “I need to standardize my data?”
True: JVs/ Alliances/ Collaboration approach allows researchers to tackle huge
projects but different instruments produce different data and different scientists record
data in different ways and formats.
6. Slide 6 CONFIDENTIALCONFIDENTIAL
Business: “I can’t locate my data when I need it?”
True: 100,000 genomes project aims to sequence 100,000 human genomes in just 5
years. 13 regional health service groups are contributing, made up of thousands of
health care professionals who then depend on multiple partners for sequencing, analysis
and storage. Data Science resources are expensive.
Data Drivers & Challenges - Business stakeholder perceptions
Business: “Organizational Inertia around data ownership and accountability”
True: Data Ownership can be defined if you know the authorized data source,
consumer and distributor across domains.
7. Slide 7 CONFIDENTIAL
Data Governance for SML (Small, Medium, Large Enterprise)
Data Governance is the process of formalizing & setting standards, defining rules,
establishing policy and implementing oversight.
“Life isn’t about waiting for the storm to pass. It’s about learning to dance in the rain.”-Non
Invasive Approach is something similar
Note: OSTHUS is certified partner to conduct DCAMv2 maturity assessment.
8. Slide 8 CONFIDENTIAL
Data Governance Framework
Data Governance
define the data strategy
Policies & Rules
of Engagement
Mission
Data Governance
Organization
Data Management
execute the IT and business processes that touch data
Management
define the business strategy
Data Management
Maturity Model
Data Stakeholders Data Council Data Stewards
WHOWHENWHY
Level 1:
Set up Data
Council
Level 2:
Agree Data
Scope
Level 3:
Capture Data
Assets: Data
Elements,
Sources & Flows
Level 4:
Implement
Data Quality &
Ownership
Level 5:
Publish Data
Controls
Level 6:
Review Data
Controls &
Transition to
RTO
Ongoing
Effort:
Maintain Data
Quality & Data
Assets
Quality Monitoring
Data Modelling
Ref. & Master Data
WHAT
Goals Metrics/Success
Measure
Value Identification &
Funding
Focus Areas Data Rules, Definitions & Policies
to achieve
HOW
Control Mechanisms
Decision Rights
AccountabilitiesInitiatives and
priority use-cases
9. Slide 9 CONFIDENTIAL
Data Governance is relevant for all (Small, Medium, Large Enterprise)
Data Governance components are rather domain agnostic and regardless of the
company’s size, data related challenges are almost always identical e.g.
1) Data Quality issues & high total cost of ownership
2) Regulatory Requirements such as GDPR, BCBS239, MIFiD2, GxP, SOX
3) Legacy system & Data Integration
4) No Traceability/Findability of data (Lack of FAIR implementation).
Large
Small
Medium
Note: FAIR stands for findability, accessibility, interoperability, and reusability
10. 10
CONFIDENTIAL
Data Governance Framework
Data you
can find
quickly
Data you
can
understand
Data that is
factually
correct
Data that
is
complete
Data that is
consistent
Data that you
can trace
(RAS)Register
AuthoritativeSources
EnterpriseData
Model
DataQuality(DQ)
Reporting
DataContracts.
DataIntegration
Framework
DataLineage
EnterpriseData
Catalogue
ReferenceData
Services
BusinessGlossary
DataDictionary
DataQuality(DQ)
Management
DataScience
andAnalytics
Data as an
asset
Recommended Approach to Enterprise Data Strategy
11. Slide 11
Identification of Key Challenges & Questions
• In many process-driven organization , data is not managed as an asset
• Business information needs & regulatory pressure
• X-functional data sharing and collaboration
• Data standardization and quality
• How far should we go with Reference and Master Data Management?
• Assumption that “IT will solve it”
Agreement on Guiding Principles & Policies
• E.g., FAIR, data accountability, standardization
• Target data and IT architecture
Practical Implementation with Concrete and Coherent
Actions Roadmap
• Prioritization of data assets, applications and a use-case pipeline
Moving from process/application-centric to a data-centric organization requires a
good data strategy
Data Strategy: Moving to a Data-Centric Organization
12. Slide 12
Illustration of hybrid (central + federated) DG Organizational Bare-
bone Structure
A DG bare-bone structure is a minimalist view
that is often recommended as a bottom-up
approach.
Anything over and above this recommendation
is aspirational:
1. Setup Division Data Council
2. Setup Division Data Office (identify
people and start as a working group to get
started and)
3. Identify Division Data Office Roles
Othe
r DG
Tea
ms
Othe
r DG
Tea
ms
Other
DG
Teams
Enterprise Data Council
Division Data Council
Division Data Office
Data Governance Manager Technical Data Steward
Data Platform Owner
Data Consumers
Data Content Owner
Business Data Steward Data Architect
Business IT
Chief Data Office
full-time role part-time role Enterprise Scope
Working Group
13. Slide 13 CONFIDENTIAL
What is a Data Management Maturity Model?
DMMM Article: https://www.linkedin.com/pulse/importance-data-management-maturity-model-your-abhishek-prasad
The Data Management Maturity Model (DMMM) provides guidelines to help organizations
build, improve, and measure their enterprise data management capability. It is a consistent,
organization-wide framework used to implement data management practices. This leads to
data that is accurate, timely, and accessible across the entire organization. The DMMM
follows data strategies, policies, and regulations that align with industry standards.
What does the DMMM provide?
1) A pragmatic step-by-step approach to building, embedding, and measuring data
management capabilities.
2) A standard method each divisional data office can use to track progress.
3) A consistent benchmark to compare data councils (i.e. groups of data stakeholders).
4) A granular list of artifacts to demonstrate evidence and progress.
14. Slide 14 CONFIDENTIAL
DMMM Maturity vs ROI
R
O
I
1
2
3
4
5
6
DMMM : Data Management Maturity ModelROI: cost-benefit analysis
Implement data governance over
current state.
Assess Your
Gaps
Most
organization
achieve
maximum
value
realization for
their Data
Governance
program at
Level 5 and
the journey
thereafter is
at times seen
as
discretionary.
Maturity Levels
15. Slide 15
Data Quality Management defines the processes required to ensure data content is of
sufficient quality to support defined business and strategic objectives.
The Data Quality Management (DQM) component describes the capabilities needed to ensure
that quality objectives are realized
– Data profiling
– Quality measurement
– Defect tracking
– Root cause analysis
– Data remediation
Data Quality Management
Note: Data Quality is NOT the sole responsibility of the Office of Data Management - but instead by all stakeholders within the information ecosystem.
16. Slide 16
Meta Data Management & Governance
Metadata Data provides additional information or context about other data.
1. Business Metadata: Provides context about the data from the perspective of the
business process.
2. Technical Metadata: Used to describe the creation, organization, movement, change
and storage of the data from the perspective of the physical implementation. E.g.
schema, table column, lineage.
3. Operational Metadata: Contains information that is available in operational systems
and run-time environments from the perspective of the process execution. E.g. Data
file size, runtime statistics etc.
17. Slide 17
Meta Data Management & Governance
Critical Data Element: A data element that is aligned to a critical business element and is
deemed materially important.
Master Data: Entities, relationships, and attributes which are identified as critical, shared
across the enterprise, and foundational to key business processes and application systems,
and peripheral and context-providing to the transactional data they manage.
Reference Data: Data that defines the set of allowable values to be used by other data
fields
Transaction Data: Data that defines the set of allowable values to be used by other data
fields
18. Slide 18
DG Stakeholder Communication Strategy
GROUP 1 GROUP 2 GROUP 3 GROUP 4
SENIOR MANAGEMENT
STEERING COMMITTEE
[EXECUTIVE LEVEL]
DATA GOVERNANCE
COUNCIL
[STRATEGIC LEVEL]
DATA DOMAIN
STEWARDS
[TACTICAL LEVEL]
DATA
STEWARDS
[OPERATIONAL LEVEL]
INFORMATION
TECHNOLOGY
[SUPPORT LEVEL]
DATA GOVERNANCE
PARTNERS
[SUPPORT LEVEL]
ORIENTATION COMMUNICATION
PROGRAM PRESENCE & AWARENESS
ON-BOARDING COMMUNICATIONS
GROUPS
COMMUNICATION
TYPES
ON-GOING COMMUNICATIONS
CHARTER AND
PRINCIPLES
ROLE-BASED
ACTIVITIES
GOVERNANCE
DOCUMENTATION
PERFORMANCE
METRICS
ALERTS &
TRIGGERED EVENTS
COUNCIL/MINUTES/
MASS COMM