Transcript EricSiegel.pps - Prediction Impact

Predictive Analytics for Business and Marketing
Hosted by Prediction Impact, Inc.
in association with the Emetrics Summit
________________________
Eric Siegel Ph.D.
Prediction Impact, Inc.
[email protected]
(415)385-1313
© 2007 Prediction Impact, Inc. All rights reserved
Predictive Analytics for Business and Marketing
Case Study: Direct Response
1. Company overview:


National veterans organization
Non-profit organization; fund raising
2. General Objectives



Find good donors
Recapture “lapsed” donors
Find “high dollar” donors
•

There is often an inverse correlation between likelihood to respond and dollar
amount of gift.
Cost-effective fund raising—net revenue
© 2007 Prediction Impact, Inc. All rights reserved
Predictive Analytics for Business and Marketing
Problem-Solving Session Template
1. Overall strategy

Outline of initiative and its primary phases
2. Predictive modeling approach


Prediction statement/goal
Data required
•
•


Applicable segments
Predictors
Deployment: How the model will be integrated or otherwise made use of
Business case
3. Evaluation



KPIs (a la business goals)
Final AB test; control group
Baseline method for comparison
4. Challenges and bottlenecks anticipated


Organizational
Technical
© 2007 Prediction Impact, Inc. All rights reserved
Predictive Analytics for Business and Marketing
Problem-Solving Session #1: Retention
•
•
•
•
•
Business: Web-o-Rama House of Metrics (worhm.com)
Year: 2012
Type: B-2-B
Description: Web analytics services for small to medium
businesses
Customer Breakdown:
–
–
100k free subscribers
30k premium subscribers
•
•
•
Credit card auto-bill monthly
High churn rate: 35% per year
Problem: Attrition rate has increased 20% since last
quarter, while conversions have remained the same.
Another team is working on increasing conversions.
© 2007 Prediction Impact, Inc. All rights reserved
Predictive Analytics for Business and Marketing
Summary of Killer Apps
•
Increased profit with response modeling for direct marketing
–
–
•
Increased customer retention by predicting defection
–
–
•
Dynamic, behavior-based content selection
From AB selection to ABC...Z selection
Increased sales by predicting cross-sell opportunities
–
–
•
Retaining tenured customers
Converting first-time customer
Increased response with targeted content
–
–
•
Increase response rate
Decreased spending
Recommendations engine
Collaborative filtering
Increased net worth by predicting customer lifetime value (LTV)
–
–
Higher valued acquisitions
Optimized retention targeting
© 2007 Prediction Impact, Inc. All rights reserved
Predictive Analytics for Business and Marketing
Jellybeans Brain Teaser
• Good: red red red red red red red red red red red
• Good: blue yellow green orange
• Bad: black moive magenta beige turquoise fuisha
© 2007 Prediction Impact, Inc. All rights reserved
Predictive Analytics for Business and Marketing
How a Crystal Ball Works
© 2007 Prediction Impact, Inc. All rights reserved
Predictive Analytics for Business and Marketing
Predict This!
•
•
•
•
•
•
•
Introduction
Data Overload
How Modeling Works
Statistical Models in Fashion
Modeling Methods
Deployment & Results
Conclusions
© 2007 Prediction Impact, Inc. All rights reserved
Predictive Analytics for Business and Marketing
Can Computers Think
© 2007 Prediction Impact, Inc. All rights reserved
Predictive Analytics for Business and Marketing
Next Steps for you
•
Moving towards a predictive analytics initiative
–
–
–
•
List your top 3 to 5 business optimization objectives
Match the list of killer apps to these objectives
Scope the data collections requirements
Other courses
–
–
–
The Modeling Agency's Level II and III Training
Tools courses, such as Salford Systems
These courses go beyond business apps to include science and
engineering
© 2007 Prediction Impact, Inc. All rights reserved
Predictive Analytics for Business and Marketing
Recommended References Sources
•
•
•
•
•
•
Prediction Impact’s bi-annual email newsletter: Case studies, articles, events.
Click “subscribe” at www.PredictionImpact.com.
www.KDnuggets.com: A comprehensive online reference and newsletter.
The UCI KDD Database Repository (kdd.ics.uci.edu): the most popular site
for datasets used for research in machine learning and knowledge discovery.
But all the core references such as this are found under KDnuggets, above.
The Cartoon Guide to Statistics, L. Gonick and W. Smith.
Data Mining: Practical Machine Learning Tools and Techniques with Java
Implementations, E. Frank and I.H. Witten.
Database marketing books for preliminary steps towards modeling, and for a
more holistic, less technical marketing viewpoint:
– Strategic Database Marketing, Arthur Hughes
– See JimNovo.com for his “Drilling Down” book (first 9 chapters for free)
– Email Marketing by the Numbers, Chris Baggott
© 2007 Prediction Impact, Inc. All rights reserved
Predictive Analytics for Business and Marketing
Predictive Analytics and Data Mining

Services:
• Defining analytical goals & sourcing data
• Developing predictive models
• Designing and architecting solutions for model deployment
• "Quick hit" proof-of-concept pilot projects

Training programs:
• Public seminars: Two days, in San Francisco and other locations
• On-site training options: Flexible, specialized
• Instructor: Eric Siegel, Ph.D., President, 15 years of data mining, experienced
consultant, award-winning Columbia professor
• Training participants: Boeing, Corporate Express, Compass Bank, HewlettPackard, Liberty Mutual, Merck, MITRE, Monster.com, NASA, Qwest, SAS, U.S.
Census Bureau, Yahoo!
© 2007 Prediction Impact, Inc. All rights reserved
Predictive Analytics for Business and Marketing
Predictive Analytics and Data Mining
Applications:
Verticals:








Response modeling for direct
marketing
Product recommendations
Dynamic content, email and ad
selection
Customer retention
Strategic segmentation
Security
– Fraud discovery
– Intrusion detection
– Risk mitigation
– Malicious user behavior
identification
Cutting-edge research for
groundbreaking data mining
initiatives






Online business: Social networks,
entertainment, retail, dating, job
hunting
Telecommunications
Financial organizations
A fortune 100 technology company
Non-profits
High-tech startups
Direct marketing, catalogue retail
© 2007 Prediction Impact, Inc. All rights reserved
Predictive Analytics for Business and Marketing
Predictive Analytics and Data Mining
Team of several senior consultants:
•
•
•
•
•
Experts in predictive modeling for business
and marketing
Relevant graduate-level degrees
Communication in business terms
Complementary analytical specialties and
client verticals
Published in research journals and
industrials
Extended network of many more:
•
•
Eric Siegel, Ph.D., President
Prediction Impact, Inc.
San Francisco, California
[email protected]
(415) 385-1313
For our bi-annual newsletter, click
“subscribe”:
www.PredictionImpact.com
To receive notifications of training
seminars:
[email protected]
Closely collaborating partner firms
East coast coverage
© 2007 Prediction Impact, Inc. All rights reserved