Betsy Burton

1
Data Warehouse and Business Intelligence Manage
Costs and Enhance Value

• Betsy Burton

2
High-Cost Failures
Data Warehouse
Service
Marketing
Billing
Sales
Support
Analytical Side of CRM

• Integrated
• Transformed
• Complete
• Cleansed

3
Key Issues
1. What are the components of the TCO model for a
data warehouse implementation? 2. What best
practices can locate hidden costs and reduce the
TCO of a data warehouse? 3. Can the TVO of a data
warehouse be determined before implementation?
4
TCO for a Data Warehouse Implementation

• Technology Costs
• Software, hardware and maintenance
• Servers
• Desktops
• Storage
• Telecommunications
• ETL tools
• Data cleansing
• BI tools/applications

• People Costs
• Internal staff
• Customers
• ESP/IS staff
• Vendor staff
• Internal support
• External support
• Education
• Training (EU and IS)
• Skills

• Process Costs
• Strategy
• Tactics
• Skills
• Operations
• Business functions

Phase
Strategize
Execute
Manage
Evaluate
Costs
Years
3
1
2
5
A Data Warehouse Sample Budget
1TB of Raw Data With 3TB Disk Storage
Support and Maintenance
Data Warehouse Platform
584,841
1,564,473
ETL Platform (Hardware and Software)
11
30
15
782,237
3
Misc. Software
152,500
8
33
DBMS Software
IT Staff/Services
425,000
1,756,994
Total 5,266,045
6
Data Warehouse Staffing
DBA 13
System Admin. 5
DA 3
QA 5
ETL 45
Production Services9
Storage Management 2
Architects 2
Business Analysts 16
7
Data Preparation (ETL) Components
Three Approaches to ETL
Poor Persons Approach

• Low implementation costs
• High utilization costs

Iterations

• Repetitive costs
• Latency of phase delivery

Subject Area by Subject Area

• Higher upfront costs
• Durable, flexible
• Low-latency application delivery

8
Growth of Data Warehouse Implementations
Number of Queries vs. Query Complexity
Complexity of Queries
Number of Queries
Year 1
Year 2
Year 3
Year 4
Year 5
9
Data Warehouse Readiness Assessment

• High-level, full-time manager in charge
• Support of various business areas
• IT/user classification map (e.g., Type A, B, C)
• Culture to change business process as a result of
  analysis

Corporate Commitment

• IT perceived as strategic, not a necessary evil
• History of the execution of IT deployments and
  benefits
• IT has strong reputation among users

Importance of IT and Linkage to Corporate Strategy

• Expertise available internally or outsourced
• Full-time commitment of resources to effort
• Support of application support teams

Skill Set Availability

• Success may deviate from corporate standard
• Technology will not solve all issues
• Tactical products need replacement strategy

Viability of Vendors and Technology
10
Leverage a Shared-Service Model

• Does Not Work for Everything
• Helps to manage IT infrastructure costs
• System administration, 24-hour coverage is easier
  to obtain
• Provide a broader breadth of experience and
  perspective
• Select and manage external service providers for
  competence or economy, provided this does not
  inhibit innovation
• Manage to increase productivity, balanced with
  opportunity

11
Case Study Implementation Review
Large Financial Institution
Sources

• Results of Review Lead to
• Campaign management work is brought in-house
• More than 6 million per year eliminated 16
  percent of total BI costs
• Three- to four-week turnaround for list
  processing is eliminated
• Now measure campaigns against current data
• Added 30 percent more data, including some
  additional history
• Daily scoring is performed

Data Warehouse
X
Data Marts and BI Applications
Data Sent to External Service Providers
12
Vendor-Suggested Configurations Are Often Wrong

• The Challenge of Negotiation
• Issues
• Overconfiguration can you keep discount levels?
• Underconfiguration will the vendor put skin
  in the game?
• Incremental use a pay as you grow clause
• Challenges
• Obtaining vendor references for configuration
  comparisons
• Balancing predicted growth vs. the requirement
• Obtain line-item pricing to improve ability to
  compare
• Trends
• Server costs are dropping 30 percent per year
• Disk storage is dropping 35 percent to 45 percent
  per year
• Credit for DBMS software in replacement
  situations
• Competitive situations are obtaining 50 percent
  to 60 percent discounts
• Negotiation in support and maintenance costs

13
Architecture Benefits of a Data Warehouse and BI
Competitiveness
Key trends globalization, virtualization,
transparency
Business Organization Functionality Infr
astructure Transactional
Performance management, front-office/back-office
information culture (inclusion, exclusion) user
types BI methodology skills and BI department
Leverage
BI applications strategic, operational,
analytical
QR standard, canned, ad hoc, user driven
Effectiveness
BI platforms, EBIS, data mining, IT-centric/user
driven
Data warehouse
ODS
Efficiency
ETL
Integration brokers
Real Time
Point-in-Time
14
Examples of ROI

• Traditional ROI Does Not Work
• Prevent 2 million in claims fraud
• Direct-mail campaign provides 10 million in
  product lift
• Customer retention increased 3 percent
• Reduced out-of-stock condition to increase sales
  by 6 percent

• New Type of ROI Evaluation (TVO)
• Unavailability of the data warehouse means a loss
  of 200,000 to 1.5 million per day due to fraud
  risk, credit exposure and customer satisfaction
• Decrease in time to deploy new applications by
  400 percent
• A combination of customer understanding and
  supplier control helped increase profitability by
  23 million per year

15
The Cost of Alternatives
16
The Cost of Alternatives Diminishing Returns and
Low TVO
The ROI From an Application Is Not Static.

• Caution
• After initial value, return will decrease
• Business process change to accommodate results
• New data analysis required to return to the
  original ROI
• Cost of new data infrastructure
• Latency in providing new infrastructure
• Low TVO

17
BI So Many Tools, So Much Cost?
Does Cost Per Seat Make Sense for BI Tools?
Production Servers xx.xx Development/Test
Servers xx.xx Server Support Costs
xx.xx Training Developer xx.xx Administrator
xx.xx End User xx.xx FTE Security
xx.xx New Release xx.xx Administration
xx.xx Software License Costs xx.xx Total
x,xxx.xx
?
?
?
Business Value
?
?
?
?
?
Tool Functionality
?
?
?
?
18
Total Value of Opportunity
Low Total Cost
SQL Server Data Mart
SQL Server DW
DB2 Data Mart
AnalyticalAgility
AnalyticalRigidity
Teradata DW
DB2 DW
Oracle Data Mart
Oracle DW
Teradata Data Mart
As of January 2003
High Total Cost
19
Recommendations

• The number of components that compose the data
  warehouse TCO is large, and enterprises need to
  understand these to perform adequate cost
  management.
• Enterprises that do not coordinate efforts and
  leverage initiatives will spend an order of
  magnitude more on the various BI initiatives.
• Focus on where spending should be to obtain the
  peak value for the enterprise as a whole.
• Use shared resources, where it makes the most
  sense, to contain the cost of a data warehouse
  implementation.
• Enterprises need to look at the TVO of a data
  warehouse implementation and not the ROI of a
  single application.