57
Andy Hayler CEO, The Information Difference June 3, 2009 Master Data Management: Avoiding the Potholes

Master Data Management - download.101com.comdownload.101com.com/pub/tdwi/Files/Informatica0603092.pdf · •Width Product Spec Product Sub Group Product ... • A Unilever executive

  • Upload
    vandiep

  • View
    222

  • Download
    3

Embed Size (px)

Citation preview

Andy Hayler

CEO, The Information Difference

June 3, 2009

Master Data Management: Avoiding the Potholes

Sponsor

Speakers

Andy Hayler

CEO,

The Information Difference

Michael DesteinSolutions Marketing Director,

MDM,

Informatica

4

Master Data Management: Avoiding the Potholes

Andy HaylerCEO, The Information Difference

© The Information Difference Ltd 2009

5

Agenda

• What is Master Data Management (MDM) ?

• The business case for an MDM program

• From the trenches: MDM in action

• MDM potholes

• Benefits that can be achieved

• Lessons

© The Information Difference Ltd 2009

6

What Is Master Data?

Data that is shared between computer systems

Examples: product, customer, asset, location,...

Master Data Management is the process of managing master data from an enterprise viewpoint

It is not just a technology. MDM includes the data governance structures and processes to support the lifecycle of master data

© The Information Difference Ltd 2009

7

Locations

CDI & PIM & MDM

Products (PIM)

Customers (CDI) Suppliers

Business lines

Materials

Assets

Brands SKUs

Channels

RegionsDepartments

Employee Classes

Units of Measurement

Product Groups

Branches

Currencies

Others

Organisations

Reference

© The Information Difference Ltd 2009

8

How Many Systems Generate Master Data?

Source: 2008 Information Difference Survey (115 large companies took part)

Overall # of Systems

Generating Customer data?

Generating Product Data?

Median Over 100? Highest

15

6

9

13% 2,000

13%

11%

300

768

© The Information Difference Ltd 2009

9

• Partial views scattered across enterprise

> In applications, data warehouses—even spreadsheets, etc.

• Inconsistent formats, codes, definitions

• Slow to reflect market consolidation, reorganizations, and

other business changes

• Data changes are uncontrolled—often made redundantly and

inaccurately

Why is Master Data Management So Difficult?

CRMMasterData

PartnerMasterData

SCMMasterData

ERP

ERPDW

MasterData

DWMasterData

Product

Brand

BrandSub Group

• Size

Product Usage

Target Industry

Product

Segment

Catalog

Line Item

Finished Product

• Height

• Length

• Width

Product Spec

Product Sub Group

Product Group

Product Manager

• Colour

Product

© The Information Difference Ltd 2009

10

How Well-Managed is Your Master Data?

Perception

• 1 definition of ―Margin‖

• Market 20,000 products

• 20,000 customers

• Analysts analyze information(source: customer study)

Reality

• 23 definitions of ―Margin‖

• Market 5,000 products

• 6,000 customers

• Analysts spend 60% of their time gathering information

© The Information Difference Ltd 2009

11

The Business Case for MDM

• Shell downstream attributed much of a 1% margin improvement ($140M annually) gained to managing master data and acting on clear analytics

• A Unilever executive has estimated that consistent operational master data is worth $200M/year in procurement savings alone

• BP‘s estimated that improved master data management is worth to them at ‗conservatively‘ $400M per year

This is clearly a problem worth tackling

© The Information Difference Ltd 2009

12

Home Depot Canada On-line Retailing

Company-wide product master data management – as well as product safety data, shipping and delivery information

STEP product repository is linked to multiple internal systems such as order management, and external systems such as shipping and delivery.

Customers have extensive ability to research products, and in many cases to purchase and have them delivered.

Following the Stibo STEP implementation, on-boarding a new product is reduced to just 10 days on average.

Analytic back-end to the system allows monitoring of the on-line behaviour of customers, allowing new insights

Much improved data quality

Home Depot needed to expand its retail presence from stores to on-line retailing. On-boarding new products currently takes an average of 10 weeks.

On-line retailing requires significant additional product information to that traditionally stored in internal systems to enable customers to research and purchase.

The Canadian market requires that descriptive information is stored in French as well as English.

Challenge Solution Benefits

Home Depot is the second largest US retailer

© The Information Difference Ltd 2009

13

Shell Oil Products Global Product Catalog

Eliminated duplicate product offerings, reducing global product catalog by 80%

Saved millions in R&D expense via increased efficiency

Global procurement improves customer service

Decentralized R&D led to duplicate localization of products, wasting millions in R&D expense

No way to provide major customers with global purchasing service

Inefficient inventory and sales management

Shell Oil Products is a worldwide supplier of fuel and lubricants

―(the MDM system) has allowed us to manage global products centrally, giving fast access to vital data for local operating units, and ensuring maximum efficiency in R&D."

Iain Pearson, Head of Global Product Management and Supply Chain for Lubricants, Shell

Global product master data management

Define product master data centrally

Subsidiaries map their views to corporate

Transparent views of product catalog information across subsidiaries and customers

Challenge Solution Benefits

© The Information Difference Ltd 2009

14

Provide daily extended supply chain information delivery before 8 a.m. across five time zones

Reduction of costly infrastructure, development and support resources duplication

Unprecedented regional view of the supply chain process, customer demand, planning, and stocks

Gain a clear view of business performance across 34 companies in 19 countries

Harmonize business processes and performance management across Latin America operations for reduced IT and administration costs

Create new economies of scale in the extended supply chain and capitalize on cross-border opportunities

Challenge Solution Benefits

Unilever Latin America Sinfonia Project

Unilever LA is Latin America’s leading manufacturer of consumer packaged goods

“We were delivering 100% of the correct information before 8 a.m. on the very first day, without any performance or quality issues…“

Fernando Rocha, Head of Info Mgt

Supply ChainManagement

Extract accurate, reliable data that adapts efficiently to organization and market change

Use MDM software to provide greater visibility of performance for improved planning and cost efficiency initiatives

Enterprise data warehouse: 12 TB, 1 billion transactions, 2,500 users

© The Information Difference Ltd 2009

Avoiding MDM Potholes

15© The Information Difference Ltd 2009

Lack of Business Engagement

16

Projects get pulled in different directions -IT should not lead MDM projects

© The Information Difference Ltd 2009

Poor Business Cases Risk Funding Crises

17

You don’t want to run out of money part way

© The Information Difference Ltd 2009

Turf Wars

18

When data governance goes bad....

© The Information Difference Ltd 2009

Over Ambition

19

Don’t try to solve everything at once

© The Information Difference Ltd 2009

Data Quality Denial

20

How bad did you say our data quality was?

© The Information Difference Ltd 2009

21

MDM is not a destination, it is a journey

© The Information Difference Ltd 2009

22

Typical MDM Benefits Achieved - Summary

Financial and Regulatory• Improve product margins• Lose fewer sales • Improve effectiveness of marketing spend• Demonstrate to regulators better risk managementCorporate Insight• Increase profitability via better decisionsBusiness Insights• Move to a single business language IT• Lower maintenance (one version of master data)• Faster rollouts of new systems (simpler interfaces)

© The Information Difference Ltd 2009

23

Best Practices

• Develop quantified business case

• Establish Data Governance structures

• Implement iteratively, not via a ―waterfall‖ process

• Budget adequately for data quality

• Obtain adequate input from vendor consultants

• Train personnel fully

© The Information Difference Ltd 2009

24

Think big, start small, evolve, measure

• Information management should be an enterprise initiative/discipline

• Pick a start point that gives you a return on investment

• Have a roadmap and follow it

• Measure progress (e.g. stage by data area)

• Measure quality (e.g. amount of redundancy)

• Be prepared to learn as you go

Business Unit

Sector

Enterprise

4. Definition Agreed

3. Official SourcesIdentified

7. Sustaining Data MgmtProcesses in Place

1. Data SubjectIdentified

8. Data Quality Measuresin Place & Performed

5. Data harmonized

2. Ownership agreed

6. Golden copy activated

Customer

Corporate

Individual

Person

1 2 3 4 5 6 7 8

Material

Finance

Vendor

Product

Location

DATA MANAGEMENT EVOLUTION STEPS

Data Areas

Local

© The Information Difference Ltd 2009

25

Informatica onMaster Data Management

June 2009

26

Agenda

• A key prediction

• Challenges of Master Data Management

• The Foundation for MDM

• What does Informatica do for MDM?

27

50%MDM Initiatives Will

Fail To Achieve Desired Results

28

Challenge of Master Data Management…

29

Diversity

30

Data

Types?

Usage

Styles?

Control?

Business

Value?Scope?

Owner -

ship?

31

Breadth & Redundancy

65% of Global 2000

organizations will deploy

two or more domain-

specific, MDM supporting

technologies

Gartner Predicts

Dec 2009

32

Breadth & Redundancy

65% of Global 2000

organizations will deploy

two or more domain-

specific, MDM supporting

technologies

Gartner Predicts

Dec 2009

Organizations have

5.2CDI solutions on

average.

Customer Data Integration

Phillip Russom, TDWI

Oct 2008

33

How to Manage the Diversity….

34

Lay a foundation of data integration and data quality to give you flexibility to adapt to changes

Data

Integration

1

Basic Data

Quality

2

Data Quality

Platform &

Identity Resolution

3

Single Domain

Hub

4

Cross Enterprise

MDM

5

Forrester MDM Maturity Model

35

Data Integration:The cornerstone of MDM

36

Data Integration Requirements

• Accessing data inside

and outside the firewall

• Bi-directional

transformation

• Multiple latencies

• Process orchestration

• Data Federation

• Metadata visibility

This is not your

father’s ETL!

It’s a sophisticated

data integration platform

37

Why is data quality so important?

38

It’s the master data!

You are going to use it everywhere

It’s your trusted source of truth

You will make decisions using it

39

Ensuring Data Quality for Master Data

• You can’t fix it just once

• You need to monitor and manage it as a process

• It’s not just Address

Correction

• It’s accuracy, consistency, completeness, enrichment, matching and validation

• You need to know the source

of the problems

• You won’t be successful just

writing a few scripts

40

Matching

Identities

41Informatica Confidential

Why are customer data integration projects challenging?

• Identity data is subject to

considerable error and

variation

• Databases about people &

organisations often contain

international data

• Basic search techniques

frequently miss matches or

find too many matches

• Data cleansing is not the

complete answer

EXAMPLES

• Mary Anne, Maryanne

• Easthartford, Hartford East, Hartford

• Browne – Brown

• Johnson, Jhonson

• Hannah, Hamah

• IBM/International Business Machines

• Fedex/Federal Express

• Chris – Christine, Christopher, Tina

42

Hybrid Approach to Identity Matching

Heuristic

Probabilistic

Deterministic

PhoneticLinguistic

Empirical

Use a combination of methods and algorithms to

compensate for different classes of error and

variation present in identity data.

43

Simply put,

MDM without a foundation ofData Integration,Data Quality, andIdentity Resolutionis unstable

44

WHY INFORMATICA?

45

Informatica:The Flexible Foundation for MDM

Data

WarehouseApplications Documents SaaS

Customers

Products

Locations

Assets

Employees

Suppliers

All Master Data Domains

Analytical

Registry

Co-existence

Transactional

All Hub Styles

Buy

Packaged

MDM

Build

Your

Own

All Purchase Models

46

Using Informatica for Master Data Management

Flat File

• PowerCenter drives initial loads of MDM Hub

• Informatica Data Quality to cleanse and monitor the quality of the data

• Informatica Identity Resolution matches and links common entities under a global ID

• PowerCenter supports on-going updates and distributes master data in any latency: batch, message based, real-time

Database

MDM

Hub

47

Open to All MDM Hub Options

CRM

ERP

Data

Warehouse

Custom Built Packaged MDM Existing Systems

48

The Informatica PlatformThe MDM Foundation For These Customers

Build Your Own

Build Your Own

SAP MDMInitiate Systems

Data Warehouse

Oracle MDMSiperian

MDM

49

Data Warehouse

DataMigration

DataConsolidation

Master DataManagement

Data Synchronization

B2B Data

Exchange

Information

Lifecycle

Management

*Source: Gartner EXP (February 2007)

UnstructuredApplication DatabaseOn-Demand B2B

HIPAA

SEPA

NACHA

SWIFT

The Informatica ApproachComprehensive, Unified, Open and Economical platform

50Informatica Confidential

Information Platform for MDM

Flexible foundation for MDM

Proven platform for data integration, data

profiling and data quality

Most accurate matching results for MDM

Improved

Agility

Increased

confidence

Increased

operational

efficiency

51

Next steps…

52

Next steps…

Start with a small

MDM Project now

with data integration,

data quality, and

identity resolution

as the foundation

53

Next steps…

Start with a small

MDM Project now

with data integration,

data quality, and

identity resolution

as the foundation A good place to

start is with a

Customer Registry

54

Questions ?

55

THANK YOU

Questions?

Contact Information

• If you have further questions or comments:

Andy Hayler, The Information [email protected]

• Michael Destein, [email protected]