1. Analytics at Work
Smarter Decisions, Better Results
Tom Davenport
Babson College
Deloitte Analytics Symposium
October 7, 2010

2. The Worst of Times for Decisions?
► Decision processes and outcomes are often
bad!
► The body of knowledge on what works is often ignored
► Decisions take too long, get revisited, involve too many or few
► Many bad outcomes in the public and private sectors
► Little measurement/progress/accountability
► Weak ties between
data/information/knowledge inputs and
decisions
► If we’re not getting better at decision-making,
much of IT’s work is called into question
► Data warehousing, analytics, reports, ERP, knowledge
management, etc.
2 | 2010 © All Rights Reserved. Thomas H. Davenport – Analytics at Work

3. The Best of Times for Decisions?
► Analytics and algorithms
► Intuition and the subconscious
► ―The wisdom of crowds‖
► Behavioral economics and ―nudges‖
► Neurobiology
► Decision automation
► …Etc.
3 | 2010 © All Rights Reserved. Thomas H. Davenport – Analytics at Work

4. Analytics at Work—The Big Picture
Analytical Capability Organizational Context Desired Result
Data
Enterprise
Analytical Culture Better
Leadership And Business Decisions!
Targets Processes
Analysts .
Systematic Review
4 | 2010 © All Rights Reserved. Thomas H. Davenport – Analytics at Work

5. What Are Analytics?
Optimization “What’s the best that can happen?”
Predictive Modeling/ “What will happen next?” Predictive and
Forecasting
Prescriptive
Randomized Testing “What happens if we try this?”
Analytics
Degree Statistical analysis “Why is this happening?”
(the “so what”)
of Intelligence
Alerts “What actions are needed?”
Query/drill down “What exactly is the problem?” Descriptive
Analytics
Ad hoc reports “How many, how often, where?”
(the “what”)
Standard Reports “What happened?”
5 Thomas H. Davenport – Analytics at Work 5

6. Levels of Analytical Capability
Stage 5
Analytical
Competitors
Stage 4
Analytical Companies
Stage 3
Analytical Aspirations
Stage 2
Localized Analytics
Stage 1
Analytically Impaired
Thomas H. Davenport – Analytics at Work
6

7. Analytical Competitors
Old Hands, Turnarounds, Born Analytical
 Marriott — Revenue management
 UPS — Operations and logistics, then customer
 Progressive— risk, pricing
• Harrah’s — Loyalty and service
• Tesco — Loyalty and internet groceries
• MCI/Worldcom— Cost identification and reduction
• Capital One— “information-based strategy”
• Google — page rank, advertising, HR
• Netflix— customer preference algorithms
7 Thomas H. Davenport – Analytics at Work

8. Analytical Competitors or Companies
Across Industries
Financial Services Consumer Products Hospitality/ Entertainment
• Wellpoint • E&J Gallo • Harrah’s Entertainment
• Progressive Insurance • Mars • Marriott International
• Barclays Bank • Procter & Gamble • New England Patriots
• Capital One • Boston Red Sox
• Royal Bank of Canada • AC Milan
Industrial Products Pharmaceuticals
• Astra Zeneca Retail
• CEMEX • Amazon
• Merck
• John Deere & Company • Tesco
• Vertex
• Wal-Mart
Telecommunications Transport • JCPenney
• O2 • FedEx eCommerce
• Rogers Telecom • Schneider National • Yahoo
• Cablecom • United Parcel Service • Ebay
• Expedia
8 Thomas H. Davenport – Analytics at Work 8

9. The Analytical DELTA
DELTA = change
Data . . . . . . . . breadth, integration, quality
Enterprise . . . . . . . .approach to managing analytics
Leadership . . . . . . . . . . . . passion and commitment
Targets . . . . . . . . . . . first deep, then broad
Analysts . . . . . professionals and amateurs
9 | 2010 © All Rights Reserved. Thomas H. Davenport – Analytics at Work

10. Data
 The prerequisite for everything analytical
 Clean, common, integrated
 Accessible in a warehouse
 Measuring something new and important
10 | 2010 © All Rights Reserved. Thomas H. Davenport – Analytics at Work

11. New Metrics / Data
Wine Chemistry Optimized revenue Smile Frequency
11 | 2010 © All Rights Reserved. Thomas H. Davenport – Analytics at Work

12. Enterprise
 If you’re competing on analytics, it doesn’t make
sense to manage them locally
 No fiefdoms of data
 Avoiding “spreadmarts”—analytical duct tape
 Some level of centralized expertise for hard-core
analytics
 Firms may also need to upgrade hardware and
infrastructure
12 | 2010 © All Rights Reserved. Thomas H. Davenport – Analytics at Work

13. Leadership
 Gary Loveman at Harrah’s
 “Do we think, or do we know?”
 “Three ways to get fired”
 Barry Beracha at Sara Lee
“Our CEO is a real  “In God we trust, all others bring data”
data dog”
Sara Lee  Jeff Bezos at Amazon
executive
 “We never throw away data”
13 | 2010 © All Rights Reserved. Thomas H. Davenport – Analytics at Work

14. The Great Divide
Full steam ahead!
• Hire the people
Is your senior • Build the systems
management • Create the processes
team
committed? Prove the value!
• Run a pilot
• Measure the benefit
• Try to spread it
14 | 2010 © All Rights Reserved. Thomas H. Davenport – Analytics at Work

15. Targets
Pick a major strategic target, with a minor or two
TD Bank= Customer service and its impact
Harrah’s = Loyalty + Service
Google = Page rank/advertising + HR
Can also have two primary user group targets
Wal-Mart = Category managers + Suppliers
Owens & Minor = Supply chain managers + hospitals
15 | 2010 © All Rights Reserved. Thomas H. Davenport – Analytics at Work

16. Analysts
Analytical Champions--Own
1%
Lead analytical initiatives
Analytical Professionals—Own/Rent
5-10% Can create new algorithms
Analytical Semi-Professionals—Own/Rent
15-20% Can use visual and basic statistical tools,
create simple models
Analytical Amateurs--Own
Can use spreadsheets, use
70-80% analytical transactions
* percentages will vary based upon industry and strategy
16 | 2010 © All Rights Reserved. Thomas H. Davenport – Analytics at Work

17. The Context: Analytical Culture
• Facts, evidence, analysis as the primary
way of deciding
• Pervasive “test and learn” emphasis where
there aren’t facts
• Free pass for pushbacks—”Where’s your
data?”
• Still room for intuition based on experience
• A focus on action after analysis
• Never resting on your analytical laurels
17 | Thomas H. Davenport – Analytics at Work

18. The Context: Analytical Processes
Defection Risk
Creation
Purchase Order “What is the customer status?”
Creation
Request Global ATP Inventory Forecast
Sales Order “Will this be back in inventory?”
Global ATP Check
Fulfillment Request
Creation &
Release Delivery
Request
Returns per Customer
“What is the customer history?”
CLTV
Delivery “Does this order justify extra efforts?”
Execution
Update Update
Releases ASN
Inventory Accounting Inventory
Delivery Performance
Receives ASN “How effective is our fulfillment
process?”
Source: SAP AG 2006
18 Thomas H. Davenport – Analytics at Work

19. Better Decisions Are the Goal of Analytics
Reports Scorecards
Decisions!
Portals Drill-down
19 | 2010 © All Rights Reserved. Thomas H. Davenport – Analytics at Work

20. A Study of Decisions
► 57 attempts to improve specific decisions
► 90% of companies could name one
► Most decisions were frequent and
Decisions! operational
► Pricing (of consumer goods, industrial goods, government
contracts, maintenance contracts, etc.);
► Targeting of consumers for marketing initiatives (by retailers,
insurers, credit card firms);
► Merchandising decisions by retailers (what brands to buy in
what quantity for what stores, shelf space allocation);
► Location decisions (for bank branches, where to service
industrial equipment)
► Results in “Make Better Decisions,” Harvard
Business Review, Nov. 2010
20 | 2010 © All Rights Reserved. Thomas H. Davenport – Analytics at Work

21. Systematically Making Decisions Better
Identify Inventory
Better
Decisions
Intervene Institutionalize
21 | 2010 © All Rights Reserved. Thomas H. Davenport – Analytics at Work

22. Most Common Decision Interventions
0.9
0.8
0.7
0.6
Frequency Mentioning
0.5
0.4
0.3
0.2
0.1
0
Type of Intervention
22 | 2010 © All Rights Reserved. Thomas H. Davenport – Analytics at Work

23. Multiple Interventions:
Better Pricing Decisions at Stanley
 Pricing identified as one of four key decision domains by CIO
 Pricing Center of Excellence established in 2003
 Adopted several difference pricing methodologies
 Implemented new pricing optimization software
 Regular “Gross Margin Calls” for senior managers
 Offshore capability gathers competitive pricing data
 Some automated pricing systems, e.g., for promotions
 Center spreads innovations across Stanley
 Result: gross margin from 34% to over 40% in six years
23 Thomas H. Davenport – Analytics at Work

24. Closing the Loop: Systematic Review
and Learning
► Tom Brady as ―a student of error‖
► The US Army’s ―After Action Review‖
► Chevron’s ―lookback‖ at all expensive
decisions
► Providence Regional Medical Center in
Everett, Washington
► “The hospital set up an independent panel to investigate
medical mistakes, disclose its findings to the patient, and
voluntarily offer a financial award if warranted. As a result,
Providence has only two malpractice suits pending, compared
with an average of 12 to 14 at other hospitals of similar size.”
(Business Week, Jan. 7, 2010)
24 | 2010 © All Rights Reserved. Thomas H. Davenport – Analytics at Work

25. Keep in Mind
► Five levels, five factors for building
analytical capability
► Data and leadership are the most
important prerequisites
► Make sure your targets are strategic
► Tie all your BI and analytics work to
decisions
► This is not business as usual—there is a
historic opportunity to transform your
industry!
25 | 2010 © All Rights Reserved. Thomas H. Davenport – Analytics at Work