2. 2
Master Data Management is a
cornerstone for data-driven processes
Know Your
Customer
Know Your
Products
Know Your
Suppliers
3. 3
3
MDM DEFINITION
Master data management (MDM) is the process of creating a single point of reference for
highly shared types of data, including customer, products, suppliers, sites, organizations
and employees.
Master data management requires companies to create a single view of their shared
master data asset. It then links together multiple data sources, and ensures the
enforcement of policies for accessing and updating the master data, handling data quality
and the routing of exceptions to people.
This “data stewardship” capability allows the lines of businesses to take ownership of the
content they need for their data centric processes. Once a single view is created, that
data can be operationally applied, and eventually in real-time, to business problems and
opportunities.
MDM is a strategic initiative for data-driven organization seeking to improve business
results such as better customer service, increasing cross-sell and up-sell revenue, and
streamlining supply chains.
4. 4
The journey from Data Integration to Information
Governance
From a fully IT driven model…
…to a federated and collaborative
responsibility model
IT Lines of
Business
Evolutionpath
From Data Management… …to Information Governance
5. 5
The Business cases for MDM
M&A and
restructuring
010101011010101010101
010101101010101010101
010101010101010101010
101101010101010101010
101011010101010101010
101101010101010101101
010101010101010101101
0 1 0 1 0 1 0 1 0 1
360°
Views
Managed Data
Accuracy
Collaborative
Data
Governance
Information
Accessibility
Information
Accountability
MDM
Platform
Governance,
Risk Compliance
and fraud mgmt.
Just-in-time and lean
operations
Customer
centric
processes
Customer
Experience
Management
Time to market
6. 6
MDM : why change? why now? And how ?
Source : Gartner 2014 survey Enterprise Information and MDM
MDM is a hot topic
•in top 3 initiative for 50% of IT execs
There is a urgent need to refresh current
processes linked to master data
•Ratings of the current capability: 3,6 on 7 ; average for 79%;
poor for 21%
A lot of companies have engaged, but most are at
early steps
• 61% still on planning/prototyping phases
Only 49% have a clear business case
• and 31% through an ROI model
7. 7
Typical challenges during MDM planning cycle
Lack of a solid
Business Case
Lack of readiness
Unclear
Roadmap
Misalignment
between
stakeholders
Unclear
requirementsUndefined
Roadmap
Many MDM initiatives
get stuck in their
planning phase
8. 8
So Where to start your journey to data governance ?
Define your business needs and your roadmap
Set up your stewardship organization
Design the platform
Engage your
MDM programs
9. 9
Some misconceptions on MDM
Misconception Key success factor
Massive IT Project
(Think Big, Start Big)
Incremental program with
engagement from Lines of Business
MDM & integration
as separate disciplines
(Start Small, Stay Small)
Total data integration capability for
current and future needs
A standalone application
(Siloed Approach)
A real time platform to operationalize
the master data
Golden record is only based on
systems of record like CRM
(Soon to be Outdated)
There will always be new sources of
data to give you a better 360 view of
customer--- social, mobile,
clickstreams….
11. 11
Modeling your data
Key steps to consider
• Creating the data model
• Defining the business rules
• Defining Data Validation controls
• Defining the roles , and the security
Modeling
Managingthe
dataquality
Enablingstewardship
Integrating&
propagatingthedata
Operationalizing
themasterdata
12. 12
Organizing for MDM: Defining the implementation
Style
MDM
ERP
CRM
COTS
DWH
Consolidation
MDM
ERP
SFA
CRM
DWH
Centralized
MD
M
CRM
E-
Commerc
e
Marketin
g
DWH
Coexistence
MDM
ERP
SFA
CRM
DWH
RegistryLess Intrusive
Most MDM Configuration
Most ESB Configuration
Less Intrusive
Standard MDM Configuration
More Intrusive
Standard MDM Configuration
Optional ESB Configuration
Most Intrusive
Moderate MDM Configuration
Required ESB Configuration
13. 13
Modeling best practices
Functional
Engage heavily the LOBs in the designing effort
Reach consensus ASAP on the data definition of
golden record
Start at the core and keep it simple, then expand
Make the model as self explanatory as possible
for the business users, and document your
business glossary
Create your own primary key
Manage the design and validation phase
carefully, as changing a data model at run time
once the data is populated may be a tedious
exercise
Leverage views and roles for usability
Value:
➜ Establish sustainable foundations for your
MDM model
➜ Establish the cornerstone for collaboration
(Stewardship and IT integration)
Technical
Create an internal permanent key for Master
Data records
Define modeling standards and respect them
Use a graphic Case tool for the design
Establish naming rules
Reuse definition, rules and patterns
Anticipate the performance impact of
controls, enrichment and propagation rules
14. 14
Managing the Data Quality
Key steps to consider
• Data Profiling
• Collect the referential to enriching the data
• Defining parsing, standardization, validation
• Defining the matching and survivorship
• Building Address validation rules
Modeling
Managingthedata
quality
Enablestewardship
Integrating&
propagatingthedata
Operationalizing
themasterdata
15. 15
Taking care of the most precious “resource”
in a citizen community: the children
Challenge:
Need a single view of a child to provide top quality
services and value for money on a one to one basis
for the local government’s 210 000+ children and
their family
Why Talend:
• MDM masters the cross references between
public services (education, social care…) and
orchestrates data governance to effectively
match, merge and un-merge incoming records.
• Complex Data Integration and Data Quality load
routines provide sophisticated fuzzy matching.
Value:
Improved public service provided for child
protection, through a shared knowledge of each
child situation and context
* For Internal Use Only
16. 16
Data Quality best practices
Functional
Know your data before starting the design:
content, availability volume, typology, reliability,
reference data
Understand the information supply chain: who
creates, imports, update, consumes (and
when/where…)
Establish strong collaboration with stewards in
charge of manual resolution to fine tune your
matching algorithms iteratively
Define business and project metrics to be
monitored over time, in order to size the data
stewardship efforts and to show the progress
Value:
➜ Illuminate the data quality problems and its
impact for lines of business
➜ Establish clear metrics for measuring the
progress and success of the MDM program
Technical
Use a data profiling tool
Integrate the data quality rules as
gatekeepers in your data integration process
Understand the constraints and objective
that are behind the matching policies,
including performance, impact of
mismatches, cost of manual efforts…
Anticipate the need for adjustments,
including for undoing redoing data resolution
activities
17. 17
Synchronizing with the existing systems in
batch or real time
Key steps to consider
• Batch/real time, Bulk or incremental load,
propagation : defining the integration
policies
• Integrating with applications: internal, cloud
based, external
Modeling
ManagingtheDataQuality
Enablestewardship
Integrating&
propagatingthedata
Operationalizing
themasterdata
18. 18
Challenge:
Support hyper growth of members in a non profit
and highly regulated healthcare market
Re-engineering customer facing processes
Use case: Re-engineering member relationship
in a heavily regulated environment
Key capabilities need:
Start with strong Data quality and data reconciliation
capabilities
Manage external data standards and connect in real
time with exchanges in the healthcare industry
Implement workflow driven processes for customer
facing activities (on-boarding, claims, billing…)
Value:
• Compliance (with HIPAA regulations)
• Scalable processes to meet hyper growth (+250%
members acquisition rate)
• Lower TCO and automated processing
19. 19
Integration best practices
Functional
Define the integration architecture and the decision
criteria to inform data integration scenarios for each
source and targets
Design the integration layer as a moving object that
will have to evolve on a regular basis, with its own
lifecycle (new systems to connect, upgrades…)
Use design mechanisms like publish and subscribe or
Master data services to avoid dependencies
between system and have clear segregation of
duties
Value:
➜ A shared service to bring trusted data across
your IT trough a well defined and rapid to
deploy process
➜ Manage change info your MDM program and
take advantage into new sources of data and
accelerate the roll-out of new applications
Technical
Invest on productivity and change
management tools, since this makes a
substantial part of your TCO
Identify the volume now…and for the future
Identify the MDM multiple environments
Define procedures for Delivery between
environments
Integration
Services
Data Staging
MetaData
Repository
Web Layer
Hybris
TCP/IP - Kereberos
Legend
Customer Data Management – Static Architecture
Integration
Services
Batch
Adaptors
Real-time
Adaptors
Real time
data
services
File based
Master
Repository
@ComRes
ACDS
Pega
Tracs
Vision
Data Quality Services
Talend Integration Platform
Parsing
& enrichment
(Experian)
Matching
Services Batch data
services
Data Layer
Master Data
Governance
Talend
Administration
Data Quality
Dashboard
Migration
Adaptors
Standardisation
Services
IntegrationLayerActive
Directory
SOAP over JMS
GetCustomerDetailsCore
GeCustomerinteractions
CreateCustomer
UpdateCustomer
PublishCustomer
GetCustomerEngagements
GetCustomerProfile
SearchCustomer
MatchCustomer
PublishCustomerMerge
IntegrationLayer
MatchCustomerBulk
SOAP over Http
Talend ESB
20. 20
Engage your Lines of Businesses
Key steps to consider
• Organize data stewardship tasks by roles
• Managing the day to day tasks related to
master data
• Accessing and authoring the master data
• Defining the workflows for collaborative
authoring
Modeling
ManagingtheDataQuality
Enable
stewardship
Operationalize
themasterdata
Operationalize
themasterdata
21. 21
Monetizing content and increasing
ARPU in the media industry
Challenge:
Deliver 28,000 hours of multimedia content
monthly from 340 content providers targeting
75 million households
Why Talend:
• Flexibility and rapid implementation time
• Unified integration platform with
embedded data quality, ESB and Business
Process Management
Value:
Decreased costs and time for adding new
content to the movie catalog
Re-engineer the billing process to meet
compliance mandates and drastically
reduce cost and time of operations
* For Internal Use Only
22. 22
Best practices for Data Stewardship
Functional
Define and document the data governance
policies (incl inventories roles, permissions,
workflows)
Make sure that the lines of businesses are
engaged and accountable
Define clear roles & tasks for data stewards and
define their working environment and workflows
accordingly ;
Engage the data stewards early in the project,
well before the training and roll-out phase
Value:
➜ Engage the lines of business in the success of
data centric initiatives
➜ Organize for a MDM roll-out and continuous
improvement
Technical
Integrate the people driven tasks related to
data authoring, validation and correction
into the overall landscape, rather than as a
separate flow
Target the right environment for the right
roles (designers, data stewards, authors and
contributors, end users)
23. 23
To BPM or not to BPM ?
Functional
➜ Clearly identify the actors
➜ Nominate champions for roles and involve them in
the project to define the processes and activities
➜ Use agile methodologies to define the workflows
and interfaces
➜ Carefully design the users interface
➜ Leverage Business Activity Management for alerts
and continuous improvement
When to use BPM in MDM projects ?
MDM has the lead for data authoring
Lines of businesses are highly engaged
Business users are involved in the authoring
process -> need for guided procedures
There are clear links between MDM and business
processes (e.g.: onboarding a customer/employee,
referencing a product…).
Technical
Make sure you don’t transform your MDM into a
packaged app : separate data and processes in
your design
Keep it simple and anticipate frequent change
since people centric processes are subject
change and to deal with exception much more
frequently that automated processes
Don’t underestimate efforts and time related to
the user interface
Value:
• Re-engineer your processes with a data centric
approach
24. 24
Use case: getting a single view of employee in a
highly distributed organization
Challenge:
• 190000+ employees across 100 countries and
400 subsidiaries)
• No global and up to date view of the employees
at a global level in a highly decentralized
organization
Value:
• shared knowledge of employees at group
level and ability to reach them immediately,
e.g. communication in crisis situations
Key capabilities needed :
• Strong security, lineage and audit capabilities
• Integration to a disparate environment, including
employee directories)
• Workflow based authoring (e.g. : professional
transfer)
25. 25
Making MDM actionable
Key Capabilities
• Integrate Master Data Services real time into
processes
• Bring context into applications such as Big
Data, web or Mobile Applications
Modeling
ManagingtheDataQuality
Enablestewardship
Integrating&
propagatingthedata
Operationalizing
themasterdata
26. 26
Best practices for Operationalizing the Master data
Functional
Identify the touch points where you need to
integrate MDM data services, and prioritize the
roll out interactively.
Define metrics to show the business impact, e.g.
on transformation rates, click rates…
Understand the performance and availability
impact of invoking MDM real time for the
external applications
Define a small set of reusable, well documented
master data services
Connect your master data to your Big Data via
Entity Resolution to boost the relevance of your
bog data analytics
Value:
➜ 360 view are populated at the right time, right
place, when insights or actions are needed.
Technical
Closely integrate this capability into your
existing enterprise service bus capability
Define Service level agreements for the
MDM services and monitor them closely
Create sets of tests cases to industrialize and
automate the testing capabilities
MDM
Business
Applications
Mobile
Applications
Big Data
Web
applications
27. 27
Use Case Bring Actionable Customer Data
across touch points
Challenge:
Drive loyalty and customer retention in an
industry disrupted by digital transformation
Key capability needed:
• Fast & easy collection, cleansing and
reconciling of data for 15 million customers
• Definition of Master data services to bring
customer context and progressive delivery
across touch points in a real time mode
Value:
➜ Improved marketing, sales and service
through knowledge and personalization
➜ Better transformation rates, cross sell/upsell
➜ Multi-Channel consistent Customer
Experience
28. 28
Trends in MDM
Ten priorities to guide organizations into
next generation MDM
1. Multi-domain MDM
2. Multi department, multi application MDM
3. Bi-directional MDM
4. Real time MDM
5. Consolidating multiple MDM Solutions
6. Coordination with other disciplines
7. Richer Modeling
8. Beyond Enterprise Data
9. Workflow and Process Management
10.MDM solutions build atop vendor tools
and platforms
Source : TDWI next generation MDM
Key technologies challenges for next
generation MDM
1. Complex relationships
2. Mobile
3. Social
4. Big Data
5. Time-travel
6. Cloud
7. Action enablement
8. Real time
9. Extreme scalability
10.Proactive, integrated governance
Source : The MDM Institute