Big Data and Marketing - CESGcesg.tamu.edu/wp-content/uploads/2015/01/Venky... · 4 What is Big...
Transcript of Big Data and Marketing - CESGcesg.tamu.edu/wp-content/uploads/2015/01/Venky... · 4 What is Big...
![Page 1: Big Data and Marketing - CESGcesg.tamu.edu/wp-content/uploads/2015/01/Venky... · 4 What is Big Data? •Large and complex data with challenges to collect, curate, store, search,](https://reader035.fdocuments.us/reader035/viewer/2022071103/5fdc72c37ed20344002853f9/html5/thumbnails/1.jpg)
![Page 2: Big Data and Marketing - CESGcesg.tamu.edu/wp-content/uploads/2015/01/Venky... · 4 What is Big Data? •Large and complex data with challenges to collect, curate, store, search,](https://reader035.fdocuments.us/reader035/viewer/2022071103/5fdc72c37ed20344002853f9/html5/thumbnails/2.jpg)
Big Data and MarketingProfessor Venky Shankar
Coleman Chair in Marketing
Director, Center for Retailing Studies
Mays Business School
Texas A&M Universityhttp://www.venkyshankar.com
Copyright © Venkatesh Shankar. All rights reserved.
![Page 3: Big Data and Marketing - CESGcesg.tamu.edu/wp-content/uploads/2015/01/Venky... · 4 What is Big Data? •Large and complex data with challenges to collect, curate, store, search,](https://reader035.fdocuments.us/reader035/viewer/2022071103/5fdc72c37ed20344002853f9/html5/thumbnails/3.jpg)
Agenda
• What is Big Data?
• The Big Deal about Big Data
• Marketing Problems
• Models/Tools
• Applications
• My Research
• Future Outlook
Copyright © Venkatesh Shankar. All rights reserved. 2
![Page 4: Big Data and Marketing - CESGcesg.tamu.edu/wp-content/uploads/2015/01/Venky... · 4 What is Big Data? •Large and complex data with challenges to collect, curate, store, search,](https://reader035.fdocuments.us/reader035/viewer/2022071103/5fdc72c37ed20344002853f9/html5/thumbnails/4.jpg)
What is Big Data?
Copyright © Venkatesh Shankar. All rights reserved.
![Page 5: Big Data and Marketing - CESGcesg.tamu.edu/wp-content/uploads/2015/01/Venky... · 4 What is Big Data? •Large and complex data with challenges to collect, curate, store, search,](https://reader035.fdocuments.us/reader035/viewer/2022071103/5fdc72c37ed20344002853f9/html5/thumbnails/5.jpg)
4
What is Big Data?• Large and complex data with challenges to
collect, curate, store, search, transfer, analyze,and visualize
• Requires massively parallel software running onthousands of servers
• Characterized by 4Vs: Volume, Velocity, Variety,and Veracity
• Propelled by the rise of SMACIT (Social, Mobile,Analytics, Cloud, Internet of Things)
• Analysis requires models to accommodate 4Vs
© Shankar, Berry, Dotzel. All rights reserved.
![Page 6: Big Data and Marketing - CESGcesg.tamu.edu/wp-content/uploads/2015/01/Venky... · 4 What is Big Data? •Large and complex data with challenges to collect, curate, store, search,](https://reader035.fdocuments.us/reader035/viewer/2022071103/5fdc72c37ed20344002853f9/html5/thumbnails/6.jpg)
The Big Deal about Big Data
Copyright © Venkatesh Shankar. All rights reserved.
![Page 7: Big Data and Marketing - CESGcesg.tamu.edu/wp-content/uploads/2015/01/Venky... · 4 What is Big Data? •Large and complex data with challenges to collect, curate, store, search,](https://reader035.fdocuments.us/reader035/viewer/2022071103/5fdc72c37ed20344002853f9/html5/thumbnails/7.jpg)
The Big Deal about Big Data
“In God we trust, all others bring data.”
- William Deming
Copyright © Venkatesh Shankar. All rights reserved. 7
![Page 8: Big Data and Marketing - CESGcesg.tamu.edu/wp-content/uploads/2015/01/Venky... · 4 What is Big Data? •Large and complex data with challenges to collect, curate, store, search,](https://reader035.fdocuments.us/reader035/viewer/2022071103/5fdc72c37ed20344002853f9/html5/thumbnails/8.jpg)
The Big Deal about Big Data• Internet data volume: 670 exabytes in 2014; 2.6 exabytes daily
• Business data double every 1.2 years
• 500,000 data centers
• By 2020, 1/3 data through cloud, 35 zettabytes of data
• Market for big data applications
– $17B 2015; $50B in 2017 (BMO Capital market report)
– $200B business (McKinsey & Co)
• Companies investing in Big Data will outperform others by 20%(Gartner)
• Volume, Velocity (Walmart: 1M transactions/hr-2.5K terabytes)
• IBM spearheading it
• Emergence of data science/data scientist: 4.4M needed by 2015(Gartner)
Copyright © Venkatesh Shankar. All rights reserved. 8
![Page 9: Big Data and Marketing - CESGcesg.tamu.edu/wp-content/uploads/2015/01/Venky... · 4 What is Big Data? •Large and complex data with challenges to collect, curate, store, search,](https://reader035.fdocuments.us/reader035/viewer/2022071103/5fdc72c37ed20344002853f9/html5/thumbnails/9.jpg)
Big Data Social Application
Copyright © Venkatesh Shankar. All rights reserved. 9
![Page 10: Big Data and Marketing - CESGcesg.tamu.edu/wp-content/uploads/2015/01/Venky... · 4 What is Big Data? •Large and complex data with challenges to collect, curate, store, search,](https://reader035.fdocuments.us/reader035/viewer/2022071103/5fdc72c37ed20344002853f9/html5/thumbnails/10.jpg)
Marketing Problems for Big Data
Copyright © Venkatesh Shankar. All rights reserved.
![Page 11: Big Data and Marketing - CESGcesg.tamu.edu/wp-content/uploads/2015/01/Venky... · 4 What is Big Data? •Large and complex data with challenges to collect, curate, store, search,](https://reader035.fdocuments.us/reader035/viewer/2022071103/5fdc72c37ed20344002853f9/html5/thumbnails/11.jpg)
Marketing Problems• CLV analysis
• Segmentation and targeting
• Marketing mix allocation
• Promotional planning
• Sales force optimization
• Sales force productivity analysis
• Optimal media allocation
• Retail assortment planning
• Product recommendation system
• Cross-category management
• Business model analysis
• Mobile couponing………
Copyright © Venkatesh Shankar. All rights reserved. 11
![Page 12: Big Data and Marketing - CESGcesg.tamu.edu/wp-content/uploads/2015/01/Venky... · 4 What is Big Data? •Large and complex data with challenges to collect, curate, store, search,](https://reader035.fdocuments.us/reader035/viewer/2022071103/5fdc72c37ed20344002853f9/html5/thumbnails/12.jpg)
Big Data Application
Copyright © Venkatesh Shankar. All rights reserved. 12
![Page 13: Big Data and Marketing - CESGcesg.tamu.edu/wp-content/uploads/2015/01/Venky... · 4 What is Big Data? •Large and complex data with challenges to collect, curate, store, search,](https://reader035.fdocuments.us/reader035/viewer/2022071103/5fdc72c37ed20344002853f9/html5/thumbnails/13.jpg)
Big Data Models/Tools
Copyright © Venkatesh Shankar. All rights reserved.
![Page 14: Big Data and Marketing - CESGcesg.tamu.edu/wp-content/uploads/2015/01/Venky... · 4 What is Big Data? •Large and complex data with challenges to collect, curate, store, search,](https://reader035.fdocuments.us/reader035/viewer/2022071103/5fdc72c37ed20344002853f9/html5/thumbnails/14.jpg)
Collaborative Filtering
Copyright © Venkatesh Shankar. All rights reserved. 14
![Page 15: Big Data and Marketing - CESGcesg.tamu.edu/wp-content/uploads/2015/01/Venky... · 4 What is Big Data? •Large and complex data with challenges to collect, curate, store, search,](https://reader035.fdocuments.us/reader035/viewer/2022071103/5fdc72c37ed20344002853f9/html5/thumbnails/15.jpg)
Models: Supervised Learning• Big data ideal for predictive and prescriptive models
• Linear regression, Logistic regression, Decision trees[CHAID]
• Ensembles (Bagging, Boosting, Random forest)
• Neural networks
– Backpropagation algorithm, Gradient checking, Random initialization, Autonomous driving
• Support vector machine (SVM)
• Large scale machine learning systems
Copyright © Venkatesh Shankar. All rights reserved. 15
![Page 16: Big Data and Marketing - CESGcesg.tamu.edu/wp-content/uploads/2015/01/Venky... · 4 What is Big Data? •Large and complex data with challenges to collect, curate, store, search,](https://reader035.fdocuments.us/reader035/viewer/2022071103/5fdc72c37ed20344002853f9/html5/thumbnails/16.jpg)
Models: Unsupervised Learning
• Clustering (BIRCH, Hierarchical, k-means, EM)
• Dimensionality reduction (Factor analysis, PCA)
• Anomaly detection (kNN, Local outlier factor)
• Self organizing maps (SOM)
• Adaptive resonance theory (ARF)
• Latent Dirichlet Allocation (LDA)
• Structured prediction/Graphical/Visualization tools(CRF, HMM, VOS)…….
Copyright © Venkatesh Shankar. All rights reserved. 16
![Page 17: Big Data and Marketing - CESGcesg.tamu.edu/wp-content/uploads/2015/01/Venky... · 4 What is Big Data? •Large and complex data with challenges to collect, curate, store, search,](https://reader035.fdocuments.us/reader035/viewer/2022071103/5fdc72c37ed20344002853f9/html5/thumbnails/17.jpg)
Other Models
• Propensity scoring/matching model
• Stochastic models, MVDP
• Spatial models
• Genetic algorithms
• Agent modeling
• Learning curves
• Recommendation systems
• Social network mapping
Copyright © Venkatesh Shankar. All rights reserved. 17
![Page 18: Big Data and Marketing - CESGcesg.tamu.edu/wp-content/uploads/2015/01/Venky... · 4 What is Big Data? •Large and complex data with challenges to collect, curate, store, search,](https://reader035.fdocuments.us/reader035/viewer/2022071103/5fdc72c37ed20344002853f9/html5/thumbnails/18.jpg)
Big Data Applications
Copyright © Venkatesh Shankar. All rights reserved.
![Page 19: Big Data and Marketing - CESGcesg.tamu.edu/wp-content/uploads/2015/01/Venky... · 4 What is Big Data? •Large and complex data with challenges to collect, curate, store, search,](https://reader035.fdocuments.us/reader035/viewer/2022071103/5fdc72c37ed20344002853f9/html5/thumbnails/19.jpg)
Applications: Gilt• Within 1 minute at noon, 3,000 versions of message
sent to customers based on analysis of 5-year history
• 65% sales in a 90-minute period = Amazon’s daily sales
• 30% sales from mobile device
• Cloud based data capture
• Big data analysis helped add 1M customers each year
• Helped launch gourmet food, wine
Copyright © Venkatesh Shankar. All rights reserved. 19
![Page 20: Big Data and Marketing - CESGcesg.tamu.edu/wp-content/uploads/2015/01/Venky... · 4 What is Big Data? •Large and complex data with challenges to collect, curate, store, search,](https://reader035.fdocuments.us/reader035/viewer/2022071103/5fdc72c37ed20344002853f9/html5/thumbnails/20.jpg)
Applications: Linkedin
• Successful data products
– People you may know
– Jobs you may be interested in
– Groups you may like
– Inmaps
– Annual network summary
• Marketing decisions
– Decision-maker targeting using Inmaps
– Linkedin Services using propensity to buy models
– Rapid testing of multiple website changes
Copyright © Venkatesh Shankar. All rights reserved. 20
![Page 21: Big Data and Marketing - CESGcesg.tamu.edu/wp-content/uploads/2015/01/Venky... · 4 What is Big Data? •Large and complex data with challenges to collect, curate, store, search,](https://reader035.fdocuments.us/reader035/viewer/2022071103/5fdc72c37ed20344002853f9/html5/thumbnails/21.jpg)
Other Applications• Chemical company: Sales force territory alignment
• Business equipment firm: Third party sales force productivity
• GE: Intelligent aircraft engines, turbines, BOPs, MRIs
• Wells Fargo: Path to customer attrition
• Vodafone: Customer churn analysis
• Nationwide: Demand generation: 15-20% for 4 years; 11% fewer headcount; $5M savings; National campaign over high spending in sales agent concentrated MSAs
• Tesco: Targeted promotions. Net profits grew 7X (1991-2014)
• PayPal: Identified an overlooked acquisition channel
Copyright © Venkatesh Shankar. All rights reserved. 20
![Page 22: Big Data and Marketing - CESGcesg.tamu.edu/wp-content/uploads/2015/01/Venky... · 4 What is Big Data? •Large and complex data with challenges to collect, curate, store, search,](https://reader035.fdocuments.us/reader035/viewer/2022071103/5fdc72c37ed20344002853f9/html5/thumbnails/22.jpg)
Amazon: Anticipatory Shipping
Copyright © Venkatesh Shankar. All rights reserved.
![Page 23: Big Data and Marketing - CESGcesg.tamu.edu/wp-content/uploads/2015/01/Venky... · 4 What is Big Data? •Large and complex data with challenges to collect, curate, store, search,](https://reader035.fdocuments.us/reader035/viewer/2022071103/5fdc72c37ed20344002853f9/html5/thumbnails/23.jpg)
My Research in Big Data• Aisle and display adjacency through spatial modeling (with
Bezawada, Balachander, Kannan)
• Third party sales force productivity using MVDP models (with Voleti)
• Multichannel marketing allocation using Bayesian HMM andoptimization (with Kushwaha)
• Multichannel targeting using HB conditional probability models
• Market evolution through semi-parametric market responsemodel (with Lin)
• Business model evaluation using Bayesian clustering andgraphical models (with Mallick)
• Market basket size forecasting through Bayesian variableselection model (with Johnson)
Copyright © Venkatesh Shankar. All rights reserved. 22
![Page 24: Big Data and Marketing - CESGcesg.tamu.edu/wp-content/uploads/2015/01/Venky... · 4 What is Big Data? •Large and complex data with challenges to collect, curate, store, search,](https://reader035.fdocuments.us/reader035/viewer/2022071103/5fdc72c37ed20344002853f9/html5/thumbnails/24.jpg)
Future Outlook for Big Data
Copyright © Venkatesh Shankar. All rights reserved.
![Page 25: Big Data and Marketing - CESGcesg.tamu.edu/wp-content/uploads/2015/01/Venky... · 4 What is Big Data? •Large and complex data with challenges to collect, curate, store, search,](https://reader035.fdocuments.us/reader035/viewer/2022071103/5fdc72c37ed20344002853f9/html5/thumbnails/25.jpg)
Future Outlook• Will big data lead to better/newer models?
• Will big data lead to bigger insights or better judgment?
• Successes: Google translate, Google trends
• Failures: Google flu trends, Award predictions based on tweets
• Machine learning: Science of getting computers to learnwithout being explicitly programmed. Examples:
– Google’s page ranking
– Apple’s iPhoto
– Spam recognition
• Reducing big data to small relevant data for decision-making
• Marketing expert systems, Marketing machines
• Test and learning built-in models
• Blend of marketing and engineering research cultures
Copyright © Venkatesh Shankar. All rights reserved. 25