Big Data, Not Big Brother: Best Practices for Data Analytics · Big Data, Not Big Brother: Best...
Transcript of Big Data, Not Big Brother: Best Practices for Data Analytics · Big Data, Not Big Brother: Best...
Big Data, Not Big Brother:
Best Practices for Data
Analytics
Jennifer Glasgow
Acxiom Corporation
March 2013
Client Applications of Analytics
• Understanding Market/Product Trends
• Understanding Promotion
Effectiveness
• Targeting Communications
to Prospects and Customers
• Mitigating Fraud with Identity
Verification/Authentication
2
Acxiom Applications of Analytics
• Understand Data Quality
• Creating New Data
– At Individual and Geographic Levels
– Demographic/Lifestyle/Interest Modeling
– Look-a-Like Modeling (Propensities)
3
Phases of Analytics
• Discovery Phase
– Data Collection and Integration of Historical Data
– Data Analysis
– Conclusions/Predictions Reports
Models
– Acceptability of Results
• Application Phase
– Apply Model to Current/Future Data
Privacy by Design Data Considerations
• Data Origination
– Notice and Choice
• Consumer Perspective
– Expectations/Understandings
– Benefits versus Risks
• Sensitive Data Considerations
• Applying the ‘Data Minimization Principle’
• Use of the Analytics – First Party versus Third Party
– PII versus Anonymous
Privacy by Design Anonymization Spectrum
100% 0%
Device Identifiable Information
Anonymous
Choice
Notice
X
De-Identified Information
X Personally
Identifiable Information
Aggregate Information
Pseudo- anonymous
/ /
Personal Pseudo-
anonymous
PII DII AGI De-ID
SANI
ANI PII SANI
Covered Information
Ease of Technical Re-identification
Privacy by Design Security Considerations
• More Data = More Risk
• Anonymization Helps
– But doesn’t solve the problem
Take-Aways
• Many Factors Influence Right Approach
– Consider Value to Consumer
• Data and Use Define Most Parameters
• Anonymize as Much as Possible
• Don’t Overlook Security