William Mcknight - Master Data Management

download William Mcknight - Master Data Management

of 39

Transcript of William Mcknight - Master Data Management

  • 8/3/2019 William Mcknight - Master Data Management

    1/39

    UnderstandingMaster Data

    anagemen an eBenefits

    William McKnight

    Partner, Information Management

    Lucidity Consulting GroupFebruary, 2009

  • 8/3/2019 William Mcknight - Master Data Management

    2/39

    Agenda MDM Introduction

    MDM Justification and Outcomes

    MDM Architecture Approaches

    Organiztional Roles for MDM Success

  • 8/3/2019 William Mcknight - Master Data Management

    3/39

  • 8/3/2019 William Mcknight - Master Data Management

    4/39

    ArchitectureWhat Is MDM?

    MDM:MDM: The organization,management and

    ETL

    Modeling

    distribution of corporatelyadjudicated, high-qualityinformation with

    widespread use in theorganization for businessgain

    Data Qualityand 3rd party

    data

    MDM ROI

    Organization

    Workflow

    From Justifying and Implementing Master Data Management by William McKnight, DM Review April 2006

  • 8/3/2019 William Mcknight - Master Data Management

    5/39

    MDM Subject Areas Customers (CustomerData Integration)

    Products (ProductInformation Management)

    Parts (Parts Data

    Partners Policies

    Stores Geography Org Calendar

    #1

    #2

    anagemen Product (Product and

    Content DataManagement)

    Vendors

    Warehouse Sales Hierarchy

    Forrester sized the worldwide MDM market at $1.1 billion in 2006 andForrester sized the worldwide MDM market at $1.1 billion in 2006 and

    forecasts that market will grow to $6.7 billion by 2010forecasts that market will grow to $6.7 billion by 2010

  • 8/3/2019 William Mcknight - Master Data Management

    6/39

    Characteristics of MDM Subject

    Areas

    Areas of high interest to the corporation

    Areas with high impact on many projects

    Areas of potentially high controversy as to itsil

    Areas with diverse input to its build

    Areas with difficulty assigning stewardship for

  • 8/3/2019 William Mcknight - Master Data Management

    7/39

    Challenges to Implementing

    MDM Lack of centralization

    Lack of documentation Lack of cross-departmental working relationships

    Lack of data stewardship

    Conflicting master files Data misunderstandings

    Lack of metadata

    Inconsistent lexicon Other priorities

    Lack of subject-area culture

    Poor Data Quality

  • 8/3/2019 William Mcknight - Master Data Management

    8/39

    Tangible Deliverables of

    MDM

    Data Model suitable for enterprise needs Master data publish and subscribe mechanisms;Options:

    O erational master data mana ement hub

    Synchronize from existing operational masters

    Role-based workflow for business process todevelop master data

    Improved data quality Syndicated data take-on

    Conformed Dimensions for the Enterprise

  • 8/3/2019 William Mcknight - Master Data Management

    9/39

    MDM Justification andOutcomes

  • 8/3/2019 William Mcknight - Master Data Management

    10/39

    Investments in MDM

    As an enabler, means improved projectresults

    projects

    Project benefits are indirect

    MDM also enables lower operating costs

  • 8/3/2019 William Mcknight - Master Data Management

    11/39

    MDM Business Objectives Vary By

    IndustryIndustry Scenario

    Banking

    Customer Intimacy: Effective direct marketing &Retention

    Compliance, Privacy and Risk Management

    Retail

    Channel alignment & time to market

    Strategic product sourcing, GDSN and EPC

    Customer ex erience lo alt & direct marketin

    Communications,Media & Utilities

    Channel convergence

    New Product Introduction Service creation

    Customer experience, loyalty & direct marketing

    Public Sector Case management

    Single view of Taxpayer / Citizen / Terrorist

    High-Technology &Manufacturing

    Improve order accuracy and reduce inventory

    Hierarchy Management capabilities for B2B

    Product introduction, compliance and sourcing

  • 8/3/2019 William Mcknight - Master Data Management

    12/39

    MDM Manages Master Data as a

    Resource to Numerous CompanyProjects

    6

    8

    10

    -6

    -4

    -2

    0

    2

    4

    MDM

    Costs

    Project 1 Project 2 Project 3 Project 4 Project 5

    MDM Project Cost

    Data Management savings per

    project

    ROI

    Hypothesis: 5 projects / year, 40% of costs of data management in each project, Reduction: 50%, 10%

    MDM maintenance costs

  • 8/3/2019 William Mcknight - Master Data Management

    13/39

    MDM (Banking, Customer)

    Reduced Overexposure

    Trade Financing

    $35 MM

    ACME

    BankEurope

    ACMEBank

    Customer

    Corp.Europe

    Global CreditLimit $50 MM

    CustomerCorp.

    Lack of Single View of the Customer across silos leads to routineLack of Single View of the Customer across silos leads to routineoverexposure. ACME Bank should have denied one of the trade creditoverexposure. ACME Bank should have denied one of the trade credit

    facilities or demanded a higher interest rate.facilities or demanded a higher interest rate.

    BankUS

    Trade Financing$45 MM

    Total Credit Extended = $80 MM ACME Corp isoverexposedby $30 MM

    Corp.US

  • 8/3/2019 William Mcknight - Master Data Management

    14/39

    MDM (Customer) in Marketing

    Improves Response and Reduces CostCustomer Example

    24% Duplication rate+ 13% wrong addressx 4 marketing campaigns per year

    x $0.50 Cost per customer= $1.2 MM Total annual marketing waste

    Customer Example1.4 million prospects not reached

    = 14,000 potential new customers missedx $983 lifetime revenue per customer= $14 MM Total potential revenue missed

    Product Example30% item data error rate in retail files

    X 25 minutes of manual cleansing perSKU per year

    = $60-$80 cost per error+ 60% error rate for all invoices

    generated= $40-$400 per invoice to reconcile

    errors= 3-5% of total revenue

  • 8/3/2019 William Mcknight - Master Data Management

    15/39

    MDM Improves Cross-Sell and

    RetentionEconomics of Cross-Selling in Banking

    Customer Households

    2 Products, we earn an averageof $176/year

    9+ Products, we earn an average

    Wholesales Customers

    3 Products, we earn $5,000

    Every transaction is collected and combined with personal data that the customerEvery transaction is collected and combined with personal data that the customer

    provides. The bank comes up with prospective offerings just at the right time toprovides. The bank comes up with prospective offerings just at the right time tocoincide with a lifecoincide with a life--changing event.changing event.

    Source: Wells Fargo Investor presentation

    of 795+/year

    Current Average= 4+products/household

    8 Products, we earn $58,000 14 Products, we earn $285,000

    Current Average= 5 products percustomer

  • 8/3/2019 William Mcknight - Master Data Management

    16/39

    How to Attain Business

    Qualification Usually a matter of the best estimates of those who

    should know

    Through the project justification for projects with alarge component being master data

    Throu h measurin the ro ect cost and risk of

    managing master data

  • 8/3/2019 William Mcknight - Master Data Management

    17/39

    MDM With Focus On the

    Customer Master Enabling Projects

    Clustering customers by commonalities

    Ranking customers by important measures, suchas profit or spend

    Customer Lifetime Value (CLV) ,

    is a Straight-line projection of the last n years of profits or Average projected CLV of a representative customer segment

    ValueChurn ManagementMarketing Expense Management

    Targeted Marketing Cross- and Up- Selling

  • 8/3/2019 William Mcknight - Master Data Management

    18/39

    MDM ArchitectureApproaches

  • 8/3/2019 William Mcknight - Master Data Management

    19/39

    Information Architecture

    Separate MDM

    SyndicatedData

    OperationalEnv

    Master data can be

    managed here

    Master data can bemanaged here(synchronization)

    Master data canbe managed here

    hub

    Data Warehouse

  • 8/3/2019 William Mcknight - Master Data Management

    20/39

    Product

    CRMERP

    Master Data Landscape

    Hub database

    Customer Part

    Promo

    Brand

    PartsDatabase

    Marketing

    Database

  • 8/3/2019 William Mcknight - Master Data Management

    21/39

    MDM Architecture Approaches: Data

    Warehouse as Master (Default)

    DW is Master

    DW provides unique ability for DQ

    a ency an s r u on are ssues

    Query master data in the DW

    Unless its an active, operational DW, this is

    seldom sufficient

  • 8/3/2019 William Mcknight - Master Data Management

    22/39

    MDM Architecture Approaches:

    Separate MDM Hub New MDM database into operational systems

    May or may not be data entry system

    To move data Flat file export (CSV, XML, other) and load

    EII

    oo n up

    Web Services API

    Can be Scheduled or triggered, full or subset ofsubject area

    ETL/EAI/EII MDM data to operational systems and DWthat need it

    Requires MDM understanding of target systems

    Portal to MDM database for master data query

  • 8/3/2019 William Mcknight - Master Data Management

    23/39

    MDM Architecture

    Approaches: Synchronization

    Identify location of master data within (mostly)operational environments

    No specified data entry system Master data moved throughout environment

    Portal to virtual master data database

    Less performing than physical hub approach

  • 8/3/2019 William Mcknight - Master Data Management

    24/39

    Wither the Data Warehouse?

    Cannot access data in the operational

    environment Need for integrated data

    Need to have data modeled for access andanalytics somewhere in the organization

    Need a place with data quality

    Need a place for corporate master data

  • 8/3/2019 William Mcknight - Master Data Management

    25/39

    Elements of a Master File Most elements available for the subject area

    Data is documented and well-understood

    Data is predictable

    Data currency is high

    Data is atomic level

    You usually have little to no choice about what

    is the master file, but you may have to dealwith many deviations from best practices inthose files

  • 8/3/2019 William Mcknight - Master Data Management

    26/39

    Reconciling Customer Across

    Multiple SourcesSource #1 (TX)Source #1 (TX)

    SSN_NOSSN_NO X(9)X(9)Cust_NOCust_NO X(10)X(10)Div_eff_dtDiv_eff_dt X(10)X(10)

    Source #2 (OK)Source #2 (OK)Pol_IDPol_ID 9(9)9(9)Cst_NOCst_NO X(10)X(10)Stt_dtStt_dt X(8X(8

    Source #3 (LA)Source #3 (LA)Cust_IDCust_ID X(10)X(10)Claim_IDClaim_ID 9(9)9(9)Beg_dtBeg_dt 9(8)9(8)

    MDMMDM

    CLM_IDCLM_ID Dec(15)Dec(15)CUST_ID Dec(13)CUST_ID Dec(13) EFF_DTEFF_DT DATEDATE

    MEMBERMEMBER CLAIMCLAIM GROUPGROUP

  • 8/3/2019 William Mcknight - Master Data Management

    27/39

    Data Usabilit andQuality

  • 8/3/2019 William Mcknight - Master Data Management

    28/39

    Investments in DataQuality

    DQ = Absence of Intolerable Defects Investments Yield Cleaner Data

    us ness o ec ves canno e me w ouquality data in support

    Data Quality Returns are in the improved

    efficacy of projects targeting businessobjectives

    Data Quality should be an integral part of

    most projects

  • 8/3/2019 William Mcknight - Master Data Management

    29/39

    Data Quality Rule Categories: 1Category Example

    Referential

    integrity

    Order is placed for customer # 123,

    who exists in the customer tableUniqueness Only one customer exists with

    customer # 123

    Cardinality We may expect to find between oneand three addresses for a customerno more and no less

    Subtype/supertyp

    e constructs

    A customer can be divorced or not, but

    only divorced customers have a valuein the ex-spouse name field

    Value

    reasonability

    Last-year purchases should not be >

    $500,000 or < 0

  • 8/3/2019 William Mcknight - Master Data Management

    30/39

    Data Quality Rule Categories: 2Consistent Value

    Sets

    In a Name field, we would not expect

    to find certain characters such as %and $

    Formatting No period after middle initial

    Data derivation While one system may contain discount

    ,

    percentage, a simple divisioncalculation, could be a unrelatedcalculation

    Completeness Sales data covers each day frombeginning to current date

    Correctness William mcnight instead of WilliamMcKnight

    Conformance to a In the gender column we would expect

  • 8/3/2019 William Mcknight - Master Data Management

    31/39

    Data Quality Scoring

    Scoring defines how well your data meets business expectations

    Possibilities

    a va ues , , :

    43%+45%+8%=94%

    . .,

    Gender:

    Multiple rules used to determine overall system score

  • 8/3/2019 William Mcknight - Master Data Management

    32/39

    Data Quality Improvement

    Three Actions to Perform on Data Quality

    Quarantine Data to Prevent Improper Use

    Report on Quality Violations to Raise

    AwarenessChange or Repair Incorrect Data to Conform,

    when possible

  • 8/3/2019 William Mcknight - Master Data Management

    33/39

    Cost/Benefit of Adding Data

    Quality Efforts

    DQ Score investment *

    ROI (return-investment/investment) *

    additionalcost toimprove DQto the level

    Revised Project ROI(return-(investment+DQcost)/ investment+DQ cost)

    . .

    90 400000 175.00% 200000 83.33%

    85 390000 101.92% 75000 69.35%

    80** 380000 64.47% 0 64.47%

    *without considering costing for DQ improvements **Default score will happen without data quality improvements

    BestValue

    Proposition

  • 8/3/2019 William Mcknight - Master Data Management

    34/39

    Or anizational Roles

    for MDM Success

  • 8/3/2019 William Mcknight - Master Data Management

    35/39

    Supporting Programs

    Executive

    Sponsor

    ProgramGovernance

    Program Governance influences data stewardshipProgram Governance influences data stewardship

    and data stewardship is responsible for data qualityand data stewardship is responsible for data quality

    Data Stewardship

    Data Quality

  • 8/3/2019 William Mcknight - Master Data Management

    36/39

    MDM MaturitySelective master data intransactional environmentData stewardship coverssubjects implemented

    Beginnings of DQ programCorp. begins to understandMDM

    Selective master data intransactional environmentData stewardship coverssubjects implemented

    Beginnings of DQ programCorp. begins to understandMDM

    All master data intransactional environmentFeeding ALL systemsneeding the data

    Full DQ programAutomated distributionFull data stewardship

    All master data intransactional environmentFeeding ALL systemsneeding the data

    Full DQ programAutomated distributionFull data stewardshipM

    oreR

    eturnon

    Investment

    Selective data stewardsSelective master data in DWData quality catch as catchcan

    Selective data stewardsSelective master data in DWData quality catch as catchcan

    No Data StewardsApplication-focusedArchitectureNo focus on DQ

    No Data StewardsApplication-focusedArchitectureNo focus on DQ

    Immature MDM Mature MDM

    Low

    Returnon

    In

    vestment

    MDM S S

  • 8/3/2019 William Mcknight - Master Data Management

    37/39

    MDM Steps to Success

    Subject area definition

    Data Stewardship Program

    Subject area prioritization

    Architecture selection Data Quality Program

    Technology selection

    Iterative development

  • 8/3/2019 William Mcknight - Master Data Management

    38/39

    Presented by:

    William McKnight

    Partner, Information Management

    Lucidity Consulting Group(214) 514-1444

    Fax: (800) 886-7033

    [email protected]

    www.luciditycg.com

    Blog: www.b-eye-network.com/blogs/mcknight

    S k Bi

  • 8/3/2019 William Mcknight - Master Data Management

    39/39

    William McKnight

    William functions as strategist, lead enterprise information architect,and program manager for complex, high-volume full life-cycle

    implementations worldwide utilizing the disciplines of datawarehousing, master data management, business intelligence, dataquality and operational business intelligence. Many of his clientshave gone public with their success story. William is a Southwest

    Speaker Bio

    Entrepreneur of the Year Finalist, a frequent best practices judge,

    has authored more than 150 articles and white papers and givenover 150 international keynotes and public seminars. His teamsimplementations from both IT and consultant positions have wonbest practices awards. William is a former IT VP of a Fortunecompany, a former engineer of DB2 at IBM and holds an

    MBA. William is partner of Information Management at LucidityConsulting Group. He may be reached [email protected] or 214-514-1444.