Self-Service Business Intelligence
Program Overview
Director, Enterprise BI
Cummins, Inc
May 2015
2
Build a sustainable future for all stakeholdersCummins - Profitable Growth
Strong Shareholder ReturnProfits Grow Faster Than
Revenues
0.0%
5.0%
10.0%
15.0%
20.0%
25.0%
$0
$5
$10
$15
$20
2007 2008 2009 2010 2011 2012 2013 2014R
evenue (
$ B
illio
n)
Revenue and EBIT
Revenue
EBIT %
1 EBIT excludes restructuring charges in 2009 and 2014 (in Power Generation), and the gains from the divestiture of two businesses and flood insurance recovery are excluded from 2011. Also, Q2‘12 EBIT excludes $6 million pre-tax
additional gain from the divestiture of two businesses in 2011, and Q4’12 EBIT excludes $52 million in restructuring charges.
$20 BSales in 2014
Record Level
Revenues
3
Power Generation Components DistributionEngine
Global Power LeaderCummins Complementary Businesses
Cummins Market Applications
4
Light-Duty
Automotive & RV
Mining, Marine, Oil & Gas,
GovernmentConstruction & Agriculture Stationary Power
RailMedium-Duty
Truck & Bus
Heavy-Duty
Truck
Global Power Leader
countries and territories employees worldwide
Develop, design and manufacture products on continents
Business Units
Regional Organizations
Corporate Functions
Engine Applications
Market Segments
Why Self-Service BI?
The Most Effective Way to Deploy Analytics to a Global Organization
Self Service BI at Cummins
Why Self-Service Business Intelligence?
Our current business intelligence and analytical platforms
are not delivering on our business needs
Current State Microsoft Self-Service BI Platform
Business requests for new data analysis
require analyst engagement and IT
development to source and deliver data
Existing data sources are directly accessible
by business users; minimal to no analyst
and IT work needed
Sourcing of new data sources into the data
warehouse requires modeling, coding, and
testing
New data sources can be rapidly sourced
into ad hoc data area for quick access;
formal modeling into warehouse if needed
Multiple end user tools for analytics requires
substantial licensing, maintenance, and
training
Common set of analytics tools focuses
investments in training and increases
reusability across Cummins
Limited capacity for analyzing of ad hoc and
what if scenarios for exploring business data
Full featured suite of data analysis tools that
build on and around a widely-understood
Excel platform
Self-Service BI: Guiding Principles
Focus on the end users, think like an analyst
Build and foster an analytical organization
Reduce IT complexity
Business owns the data
Work together to govern the data and process
Drive BI technology innovation
Self-Service BI Distinctions
Introduces a new analytical paradigm
– Rapid prototyping, agile deployment
Connect to any data source, connect any data source
Spreadsheets with unlimited columns, unlimited rows
Eliminates costly ETLs, replaced by Extract, Load, then Transform
– Faster development, easier to reconcile data, easier to adapt to changes
Leverage analysts and business community
– Reduce dependency on 3rd party developers
Grass-Roots deployment, word of mouth communication
What is Self-Service BI?
Cummins uses the Microsoft
Business Intelligence built in
Azure to provide:
– Business friendly reporting,
analytics, and modeling tools
– Web front end for scheduling
data refreshes and migrating
analytical models
– Flexible and scalable
Analytical platform
– Ability to connect to any data
set (Source Systems, Data
Marts, Data Warehouse,
Teradata, Hadoop, etc.)
Source Systems and Processes across CMI
Engine Test Bed &
Performance
Manufacturing Control Systems
Sales Order Management
ProcessingEtc.
Operational Reporting on MES
Operational Reporting on Sales
Orders
Data Transformation and Conformation to Enterprise Model
Data Quality Assurance and MonitoringMinimal Transformation
EnterpriseData
Warehouse
Self-ServiceAnalytics Data
Data Data Data
Customer Product
SalesEtc. EDW
Self-Service Reporting Tools
Excel Reporting & Analysis PowerViews & PowerPivots Visualization & Analysis Tools Reporting & Analysis Storage
Sou
rce
Tran
sfo
rmSt
ore
Co
nsu
me
Model
Microsoft Self-Service BI Stack
Why Self-Service BI Benefits the Business
Business accountability and responsibility of the
data they already own
SSBI improves time to market (introducing agile
processes)
– New engagements: empirically gives the business
their data up to 27 weeks faster
– Change requests: empirically realized by the
business up to 34 weeks faster
SSBI allows data cleansing without impacting
source data
Business
Effective data driven decisions
Faster time-to-market
Quality data on which to make
decisions
Decision
DecisionTime to Data Analysis
AnalysisTime to Data
Why Self-Service BI Benefits IT
Security and auditing is streamlined
– The infrastructure and environment isolation in place
– The business owns the responsibility for the user-level security and auditing of their data and access rules
IT has governance visibility into the environment
– Complete records of who is using the system, what and when they have accessed, etc.
– Complete visibility on the creation of data models and all connections to source systems
Gold standard Data Warehouse can be developed over time based on actual usage
IT can focus on the infrastructure and providing a service to the business
IT
Security
Governance
Planning for the future
HighlightedSuccess Stories
Parts Pricing and Analytics
Category: Large entity, high visibility
What: Expensive, inflexible 3rd party system delivered no result
How: Designed a Self-Service solution that the Parts business teams
(Pricing, Product Management, and Analytics) now support
Value: Enabled the Parts business to analytically price 80,000 parts and
begin building the history to achieve optimal pricing
User Experience Third-Party Proprietary Self-Service Analytics
Parts Pricing
Parts BI
$6.5M
$1.4M$0*
Parts Consulting Fees $2M $300,000
Implementation Time 3.5 Years 12 Weeks
Parts Priced 0 80,000
Year 1 Revenue $0 $33M
Corporate HR Analytics
Category: Large scope
What: Manual process with continual churn and no analytical capabilities
How: Designed a Self-Service solution that allowed for automated global business efficiencies
Value: “In my 4 years at Cummins, this is the first time that we have successfully moved forward from a system perspective on analytics. This is very exciting, the possibilities are huge!”
- Brian Hamilton, Director - HR Reporting & Analysis
User Experience Past Experience Self-Service Analytics
Efficiencies Gained Manual Reporting Workforce Analytics
BI Engagement No Capability/Support Capable in 4 Weeks
AOP Process Manual Automated
Decision Cycle Time No Ad-Hoc Responses 5 Minutes
Data
Quality/Governance
1 Day 5 Minutes
Distribution BU Global Inventory
Category: Do it yourself, never had BI AOP funding
What: Manual inventory and cleansing collection process against disparate systems. Error prone, time consuming, and demanded user compliance.
How: Designed a Self-Service solution that pulls and aggregates the data automatically from multiple ERP systems
Value: Automatic data aggregation allowed for the resolution of many customer down incidents. Additionally, it provides for a secure data environment that reduces user error and enhances work satisfaction (Cummins employees can now work strategically to benefit the business rather than spend their time with data entry).
User Experience Past Experience Self-Service Analytics
Efficiencies Gained 46 Support Personnel 1 Person Part-Time
Process Compliance Chronic Weakness Automated
Data QualityManual Entry -
Worse than SourceSame as Source
Enterprise Remedy Analytics Category: Global IT Production Support group supporting multiple BI
reporting applications, reporting requests, and data warehousing services
What: Analytics capability on large volume of support requests (1k/mo)
How: Pull daily tickets direct from source into the MSBI environment for
visibility to SLA performance and all other KPIs
Value: Significantly improved ability to focus improvement work, reduce
support costs, improve customer satisfaction
User Experience Past Experience Self-Service Analytics
Availability of data Limited by vendor Full access to all
relevant data
Reporting interval Monthly per vendor Daily refresh in MSBI
Ability to drill down None Full
Visualizations on data Minimal Unlimited
Engine Warranty Analysis Category: New capability, cross-BU/cross-functional unsatisfied need
What: Existing Big Data solution was limited and not delivering end-user
data or linking with other CMI data sets
How: Built a small Hadoop solution in Microsoft Azure within 3 weeks,
including reprocessing all engine INSITE logs for 3 years
Value: Extracted and delivered all engine information to Engineering,
Reliability, and Six Sigma teams to use in their investigations. Data used
in one Six Sigma project with projected $5.6M in savings.
User Experience Past Experience Self-Service Analytics
Linkage to CMI Data Not Designed Fully Capable
Effort Realization Continual Churn 3 Weeks
New Capability Cost
Estimate>$270K $87K
Savings * >$5.6M
Engine Warranty Analysis (con’t) Phase I work completed in 3 weeks for under $60K
Over 2600 parameters per engine combined into a single analytical model to enable correlation of engine faults and failure codes to specific components
– Engine service logs pulled from EDW 93M and from CloudOne
– Customer information loaded from multiple sources (ERP’s, EDW, Customer Masters, and National Accounts)
– POLK / VIN data
– Reliability Warranty events
– Expert Diagnostic data
– Work order details/headers
– Warranty campaigns
– Vehicle registrations/OEM info
– Plasma/Genealogy
– Part sales/Product Coverage; some portions of the BOM/SBOM
Self-Service BI: Success Stories
Self Service BI
Deployment Status
Self-Service BI: Program Status
Self-Service BI: Program Status
Self-Service BI: Training Status
$48,960
$88,128$119,808
$13,200
$84,150
$168,300
$0
$20,000
$40,000
$60,000
$80,000
$100,000
$120,000
$140,000
$160,000
$180,000
Low Complexity Medium Complexity High Complexity
Engagement Cost
ADSC SSBI
21.25
38.25
52.00
2.00 12.75
25.50
0.00
10.00
20.00
30.00
40.00
50.00
60.00
Low Complexity Medium Complexity High Complexity
Engagement Time to Delivery (Weeks)
ADSC SSBI
2015 Self-Service Engagements
• Faster time to deliver
• Quicker to business
value
• Higher satisfaction
• What could 6 months
buy the business?
• Initial cost will be higher
only for complex
projects
• 100’s of small projects
vs. 15 high complexity
projects
II
2015 Self-Service Cost Avoidance
Self Service BI - Next Steps
Expand Global Training Program
Formal Communication?
Expand Azure – Europe, Singapore, others
Integration and alignment with traditional BI program
Expand advanced analytics, Machine Learning, IoT
– Tools and processes
Self Service BI – In Summary
Increased Revenue
Decreased IT Costs
Millions in IT Cost Avoidance
Increased User Satisfaction
Global Deployment
Business & IT Teams working together
Appendix
Microsoft Self-Service BI
Microsoft Self-Service BI Architecture
ACDC
PROD Delivery
CED
FTDC
PROD DR
CED
SQL for SP/MDS/SUPPORT
Primary Replica
8vCPU x 32GB
C:80GB, D:300GB, E:100GB
F:100GB, G:300GB
SQL for SP/MDS/SUPPORT
Secondary Replica
8vCPU x 32GB
C:80GB, D:300GB, E:100GB
F:100GB, G:300GB
SP WFE/APP
Active
8vCPU x 32GB
C:-80GB, D:100GB
SP WFE/APP
Active
8 vCPU x 32GB
C:-80GB, D:100GB
SQL for SUPPORT - DEV
Primary (no AG)
8vCPU x 32GB
C:80GB, D:300GB, E:100GB
F:100GB, G:300GB
SQL for SP/MDS/SUPPORT
Secondary Replica
8vCPU x 32GB
C:80GB, D:300GB, E:100GB
F:100GB, G:300GB
SP WFE/APP
Passive
8vCPU x 32GB
C:-80GB, D:100GB
SSAS Tabular
Primary/Query Replica
8vCPU x 64GB
C:80GB, D:100GB, E:50GB,
F:200GB
SSAS Tabular
Secondary/Processing Replica
8vCPU x 64GB
C:80GB, D:100GB, E:50GB,
F:200GB
SSAS Tabular
Tertiary Replica
8vCPU x 64GB
C:80GB, D:100GB, E:50GB, F:200GB
SSAS Tabular - DEV
Primary Replica
8vCPU x 64GB
C:80GB, D:100GB, E:50GB,
F:200GBACDC on-premise servers(see other illustration)
SQL 2012SSAS Tabular
Primary
Cummins CIDC
TFS 2013
Cummins FTDC
Azure-West(Production)
GTM
SQL 2012SSAS Tabular
Secondary(Synchronous)
SQL Agent Sync
SQL 2012RDBS, MDS, DQS
Primary
SQL 2012RDBS, MDS, DQS
Secondary(Synchronous)
SQL Always On Availability Group
SQL 2012Management
Primary
SQL 2012Management
Secondary(Synchronous)
SQL Always On Availability Group
SharePoint 2013Web
Active
SharePoint 2013Web
Active
SharePoint 2013App
SharePoint 2013App
SharePointSQL 2012Primary
SharePointSQL 2012Secondary
(Synchronous Commit)
SQL Always On Availability Group
SQL 2012SSAS Tabular
DR
Azure-East(DR)
SQL 2012RDBS, MDS, DQS
DR
SharePoint 2013WebDR
SharePoint 2013App
SharePointSQL 2012
DR
Azure 2015-03-04
AT&T NetBondMPLS ExpressRoute
AT&T NetBondMPLS ExpressRoute
Cummins NetworkMPLS
SQL Always On Availability Group
Pivot TablesPowerMapPowerView
PowerPivotModel
Enterprise Sync to SSAS Tabular
Microsoft Self-Service BI Architecture