MGS4020_03.ppt/Feb 19, 2013/Page 1 Georgia State University - Confidential MGS 4020 Business...
-
Upload
archibald-baldwin -
Category
Documents
-
view
216 -
download
3
Transcript of MGS4020_03.ppt/Feb 19, 2013/Page 1 Georgia State University - Confidential MGS 4020 Business...
MGS4020_03.ppt/Feb 19, 2013/Page 1Georgia State University - Confidential
MGS 4020
Business Intelligence
Ch 2 – Decision & Decision MakersCh 4 – Modeling Decision Processes
Feb 19, 2013
MGS4020_03.ppt/Feb 19, 2013/Page 2Georgia State University - Confidential
Agenda
Direct Marketing
Decisions and Decision Makers
Decision Tree
MGS4020_03.ppt/Feb 19, 2013/Page 3Georgia State University - Confidential
Example of a Decision-Making Process
Stimulus
DecisionMaker
ProblemDefinition
AlternativeSelection
Implement
Opportunities,Feedback,
Threats External Pressures and
Personal Values
Bias, Risks, Costs, and
Assumptions
Frame of Reference
Reframing, Strategy, Creativity
Acceptance, Evaluation, and
Control
MGS4020_03.ppt/Feb 19, 2013/Page 4Georgia State University - Confidential
Decision Making Process Basics
• Stimulus – External forces that the Decision Maker must evaluate to determine if he or she perceives a problem
• A problem is defined simply as the perception of a difference between the current state of affairs and a desired state.
• Decision Maker – Evaluates the stimulus to determine if he or she perceives a problem
MGS4020_03.ppt/Feb 19, 2013/Page 5Georgia State University - Confidential
Decision Making Process Basics
• Problem Definition – Decision Maker must define or frame the problem before an alternative solutions can be searched
• Alternative Selection – Evaluating the alternative solution to determine the best decision
• Implementation – Putting the solution into production
• Creating Consensus
• Negotiation
• Strategizing
• Politicking
• Intense Planning
MGS4020_03.ppt/Feb 19, 2013/Page 6Georgia State University - Confidential
Classification of Decision Makers& Decision Style
Classification of Decision Makers• Individual Decision Maker• Multiple Decision Maker• Group Decision Maker• Team Decision Maker• Organizational Decision Maker• Meta-Organizational Decision Maker
Decision Style• Problem Context• Perceptions of the Decision Maker• Personal Values
MGS4020_03.ppt/Feb 19, 2013/Page 7Georgia State University - Confidential
Decision Style Classification
Value Orientation
Co
gn
itiv
e C
om
ple
xity
StructureNeed for Structure
ComplexityTolerance for
Ambiguity
Logical Task/Technical
Relational People/Social
Analytical Conceptual
Directive Behavioral
Analytical
Directive
Conceptual
Behavioral
MGS4020_03.ppt/Feb 19, 2013/Page 8Georgia State University - Confidential
Decision Forces
A Decision Maker must balance various forces and constraints that act on a problem context in formulating a decision.
• Personal / Emotional ForcesFeelings, Health, Security, Rewards, etc.Personal and Emotional forces can reinforce or debilitate a decision makers ability to make a sound decision
• Economic / External ForcesSocietal Values, Competitive Pressures, Consumer Demands, New Technology, etc.The ultimate decision is often adjusted to account for these external forces.
• Organizational ForcesOrganizational Culture, Polices and Procedures, Resources Allocation, etc.What is the Risk Tolerance of the Organization?Is it a Conformity or Innovative Culture?
• Contextual and Emergent ForcesTime Requirements, Motivation to reach a decision, Skills Inventory etc.
MGS4020_03.ppt/Feb 19, 2013/Page 9Georgia State University - Confidential
Why are decision so hard?
Factors that determine the relative difficulty of a decision
• Structure
A Structured Problem vs. An Unstructured Problem
• Cognitive Limitations
Miller’s study of cognitive limitations resulted in the “7 slots” theory
• Uncertainty
Must rely on Subjective Probability
• Alternatives and multiple objectives
The more alternative solutions to choose from, the more difficult the
decision
MGS4020_03.ppt/Feb 19, 2013/Page 10Georgia State University - Confidential
Simon’s Model of Problem Solving
Intelligence Design Choice
ImplementationOutcomeReality
OfSituation Success
Failure
Model Validation Solution Testing
MGS4020_03.ppt/Feb 19, 2013/Page 11Georgia State University - Confidential
Rational Decision Making
Optimal Solution vs. Acceptable Solution
• Most economic theory built on the concept of individuals always seeking
the optimal solution
• Impractical because it is cost prohibitive to search, analyze and compare
every possible alternative to determine which is optimal.
• Simon suggested that we tend to “simplify reality” by searching for
solutions that meet our preconceived notion of an acceptable solution
given the problem context.
MGS4020_03.ppt/Feb 19, 2013/Page 12Georgia State University - Confidential
The Process of Choice
• While quantitative models can be used to compare and evaluate the alternatives, the Decision Maker is always faced with some uncertainty and must make a judgmental decision.
• “If you choose not to decide, you still have made a choice.”
Cognitive Processes
• The need for a DSS comes from the limits of the Decision Makers
cognitive abilities.
- Cognitive Limitations- Perception- Judgment
MGS4020_03.ppt/Feb 19, 2013/Page 13Georgia State University - Confidential
Agenda
Direct Marketing
Decisions and Decision Makers
Decision Tree
MGS4020_03.ppt/Feb 19, 2013/Page 14Georgia State University - Confidential
Decision Tree
Buy Stock
Do Not Buy Stock
Price goes up
Price goes down
Gain
Loss
Loss/gain nothing
MGS4020_03.ppt/Feb 19, 2013/Page 15Georgia State University - Confidential
Decision Tree
Buy Stock
Leave money in savings
Return > 4 %
Return < 4 %
Reach Objective - 40%
Miss Objective - 60%
Return > 4 %
Return < 4 %
Reach Objective - 70%
Miss Objective - 30%
MGS4020_03.ppt/Feb 19, 2013/Page 16Georgia State University - Confidential
Decision Tree – Activation Test
SkyMiles Enrollment
Message A
Returned within xx days
Message B
Returned within xx days
Did not return within xx days
Message C
Did not return within xx days
If Vc xx, send
Message D
Graduate to “SOW”
Did not return within xx days
If Vc < xx, no more
messages
Graduate to “SOW”
If Vc xx, send
Message D
If Vc < xx, no more
messages
MGS4020_03.ppt/Feb 19, 2013/Page 17Georgia State University - Confidential
Decision Tree - Activation Test
ChannelsEnrollmentMessage E/F
flied within xx days
Message G
flied within xx days
Did not fly within xx days
Message H
Did not return within xx days
If Vc xx, send
Message J
Graduate to “SOW”
Did not return within xx days
If Vc < xx, no more
messages
Graduate to “SOW”
If Vc xx, send
Message J
If Vc < xx, no more
messages
MGS4020_03.ppt/Feb 19, 2013/Page 18Georgia State University - Confidential
Decision Tree - Retention / SOW Test
HURDLE
SkyMiles w/ x flies last year, fly x+y
this yearMessages R*
Returned within xx days
Non-SkyMiles w/ x lx days
since last tripMessage P**
Did not return within xx days
If Vc xx, send
Message Q
Next promotion
(responsive)
Did not return within xx days
If Vc < xx, no more messages (non-responsive)
Next promotion
(responsive)
If Vc xx, send
Message S
Returned within xx days
If Vc < xx, no more messages (non-responsive)
MGS4020_03.ppt/Feb 19, 2013/Page 19Georgia State University - Confidential
Decision Tree - Reactivation Test
RATE OF trip
SkyMiles w/ xx days since last tripMessages L,M,N,O*
Returned within xx days
Non- SkyMiles w/ xx days
since last tripMessage P**
Did not return within xx days
If Vc xx, send
Message Q
Next promotion
(responsive)
Did not return within xx days
If Vc < xx, no more messages (non-responsive)
Next promotion
(responsive)
If Vc xx, send
Message Q
Returned within xx days
If Vc < xx, no more messages (non-responsive)
MGS4020_03.ppt/Feb 19, 2013/Page 20Georgia State University - Confidential
Probability
The Three Requirements of Probabilities:
1. All Probabilities must lie with the range of 0 to 1.
2. The sum of the individual probabilities equal to the probability of their union
3. The total probability of a complete set of outcomes must be equal to 1.
MGS4020_03.ppt/Feb 19, 2013/Page 21Georgia State University - Confidential
Decomposing Complex Probabilities
Severe Winter70%
Sales > 25,000 units
Sales <= 25,000 units
80%
20%
Sales > 25,000 units
Sales <= 25,000 units
50%
50%
Moderate Winter30%
Probability [ Sales > 25,000 units ] = ( .70 X .80 ) + ( .30 X .50 )= .56 + .15 .= .71 .
MGS4020_03.ppt/Feb 19, 2013/Page 22Georgia State University - Confidential
Agenda
Direct Marketing
Decisions and Decision Makers
Decision Tree
MGS4020_03.ppt/Feb 19, 2013/Page 23Georgia State University - Confidential
Direct Marketing Campaign Platform
ACQUIRE
RETAIN
REACTIVATE
“FIRE”
STORE DIFFERENT CHANNELS
A C T I V A T I O N P R O M O T I O NA C T I V A T I O N P R O M O T I O N
E-mail Address
Vehicles:
• Statements
• Newsletters
• Inserts
• Direct mail
• Personalized kits
• Telephone
Vc Cost to reactivateIf:
Vc < Cost to reactivateIf:
Ugly Postcard???
TestArea
• POS
• Partners
• Advertising
Vehicles:
• Direct Mail
• Statements
Triggered Promotions
highest value
customers
lowest value
customersdowngrade
trigger *
(for example)Days since last purchase = X
X = 30 days for PTNM
X = 60 days for GOLD
X = 120 days for CLUB
Direct Marketing Campaign Platform
PURCHASED
NO PURCHASE
PURCHASE
* < 1 purchase in last 12 mo
If : Time since inactive = X, and
Point balance > X
MGS4020_03.ppt/Feb 19, 2013/Page 24Georgia State University - Confidential
Communication “Variables”
Vehicles
= Kits
= Statement
= Telephone
= Direct Mail (USPS)
Message / Offer (incentive)
• Hurdle (SOW)
› trip x get y
• Next trip (Re-Activation)
› Rate of trip triggers
• Points (double/flat?)
• Miles (front & back-end)
•Other
Creative Execution
• Can test several executions tailored to clusters/segments
Timing/Frequency
• Monthly (statements)
• Repeat/Follow-up Mailings
MGS4020_03.ppt/Feb 19, 2013/Page 25Georgia State University - Confidential
“Measuring Effectiveness: Lift/Gains Chart
Percent of population targeted
Percent of potentialresponders captured
100
1000
90
45
45
Targeting
Random mailing
MGS4020_03.ppt/Feb 19, 2013/Page 26Georgia State University - Confidential
Example Direct Mail Optimization
• Using multivariate model we are able to maximize profit while minimizing costs
• In comparison to methodology used last year model savings = $XXX
– Savings attributable to reduced mailing to achieve last years result (variable cost savings).
• Other benefits - Customer Behavior, Planning Tool
CUMULATIVE ROI PREDICTED
0%
50%
100%
150%
200%
250%
300%
350%
400%
450%
40 37 34 31 28 25 22 19 16 13 10 7 4 1
Deciles
Hurdle 2000 Ranked List Q1 Last Year Mailed 798,313
Cost & Revenue assumptions Registered 92,523
FIXED COST 185,000$ (Last Year) % Last YearVC 0.41$ (Last Year) Planned Mail 691,951 -13.32%
Cost/Register 5.25$ (Last Year) Plan_Regis. 132,639 43.36%
Revenue/Register 23.79$ (Last Year) Equal Last Years Registration
Mailing 407,029
Multivariate Logistic Regression PREDICTIONS $ Savings 161,746$ PREDICTED PREDICTED Cumulative Cumulative Cumulative Cumulative Cumulative Cumulative Cumulative
Segments CUSTOMERS REGISTRATION RR % COST/REG REVENUE/REG PROFIT/REG ROI Predict RR % Registered Mailed
40 40,703 13,844 34.01% 14.58$ 18.54$ 3.97$ 27.2% 34.0% 13,844 40,703 39 40,703 12,282 30.17% 8.37$ 18.54$ 10.18$ 121.6% 32.1% 26,126 81,406 38 40,704 10,669 26.21% 6.40$ 18.54$ 12.14$ 189.8% 30.1% 36,795 122,110 37 40,700 9,674 23.77% 5.43$ 18.54$ 13.11$ 241.6% 28.5% 46,470 162,810 36 40,707 8,963 22.02% 4.86$ 18.54$ 13.69$ 282.0% 27.2% 55,433 203,517 35 40,702 8,383 20.60% 4.48$ 18.54$ 14.06$ 313.9% 26.1% 63,816 244,219 34 40,703 7,894 19.40% 4.22$ 18.54$ 14.32$ 339.2% 25.2% 71,710 284,922 33 40,705 7,472 18.36% 4.04$ 18.54$ 14.51$ 359.4% 24.3% 79,183 325,627 32 40,702 7,097 17.44% 3.90$ 18.54$ 14.65$ 375.6% 23.6% 86,279 366,329 31 40,700 6,755 16.60% 3.80$ 18.54$ 14.75$ 388.4% 22.9% 93,035 407,029
30 40,708 6,446 15.83% 3.72$ 18.54$ 14.82$ 398.5% 22.2% 99,480 447,737
29 40,703 6,154 15.12% 3.66$ 18.54$ 14.88$ 406.3% 21.6% 105,635 488,440 28 40,696 5,880 14.45% 3.62$ 18.54$ 14.92$ 412.2% 21.1% 111,514 529,136 27 40,711 5,627 13.82% 3.59$ 18.54$ 14.95$ 416.5% 20.6% 117,141 569,847 26 40,701 5,386 13.23% 3.57$ 18.54$ 14.97$ 419.5% 20.1% 122,527 610,548 25 40,702 5,162 12.68% 3.56$ 18.54$ 14.99$ 421.3% 19.6% 127,689 651,250 24 40,701 4,950 12.16% 3.55$ 18.54$ 14.99$ 422.2% 19.2% 132,639 691,951 23 40,707 4,749 11.67% 3.55$ 18.54$ 14.99$ 422.2% 18.8% 137,388 732,658 22 40,699 4,557 11.20% 3.56$ 18.54$ 14.99$ 421.6% 18.4% 141,945 773,357 21 40,709 4,373 10.74% 3.56$ 18.54$ 14.98$ 420.3% 18.0% 146,318 814,066 20 40,697 4,194 10.30% 3.58$ 18.54$ 14.97$ 418.5% 17.6% 150,512 854,763
.. .. .. .. .. .. .. .. .. .. ..5 40,695 2,393 5.88% 4.00$ 18.54$ 14.54$ 363.7% 13.5% 197,711 1,465,309 4 40,706 2,300 5.65% 4.04$ 18.54$ 14.51$ 359.3% 13.3% 200,012 1,506,015 3 40,709 2,196 5.39% 4.08$ 18.54$ 14.47$ 354.9% 13.1% 202,207 1,546,724 2 40,707 2,048 5.03% 4.12$ 18.54$ 14.43$ 350.3% 12.9% 204,255 1,587,431 1 40,705 1,509 3.71% 4.17$ 18.54$ 14.37$ 344.7% 12.6% 205,764 1,628,136
Totals 1,628,136 205,764 12.64% 4.17$ 18.54$ 14.37$ 344.7% 12.6% 205,764 1,628,136
0100
0
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0 10 20 30 40 50 60 70 80 90 100