Lookalike Audience Modeling

• Activity Number: 434 - SPEED: Classification and Data Science
• Type: Contributed
• Date/Time: Tuesday, July 31, 2018 : 2:00 PM to 2:45 PM
• Sponsor: Section on Statistical Learning and Data Science
• Author(s): Sam Hawala*
• Companies: Resonate-Networks
• Keywords: Marketing Analytics; Singular Value Decomposition; Cosine Similarity

Abstract

Lookalike Audience Modeling (LAM) is a marketing analytics methodology used to ultimately generate more leads and increase their conversions. The algorithm discovers similarities between internet users based upon the domain level URLs that they've visited. After a dimensionality reduction step through SVD, we calculate cosine-similarities between users of interest and users at large, based on internet behavior during a fixed period of time. We show that the recall rate is over 90%. We also present clients' key performance indicators.