04 September, 2020, Bengaluru: The Behavioural Sciences Lab of IIM Bangalore will host a webinar titled: ‘Lookalike Targeting on Facebook: Seed Quality Versus Match Accuracy’, on September 15 (Tuesday), 2020, from 07.00 pm to 08:30 pm. The webinar is primarily intended for an academic audience.
The webinar will be led by K. Sudhir, James L. Frank ’32 Professor of Marketing, Private Enterprise & Management and Founder-Director of the Yale China India Insights Program, Yale School of Management, and Editor-in-Chief of Marketing Science.
For registration, please visit: Click here
Those interested need to register for this webinar by September 12, 2020. They will be sent a zoom link on September 14, 2020. There is no certificate offered for participation in this webinar.
Contact for information: naureenb@iimb.ac.in /+91-8762666012
About the speaker: Professor K. Sudhir completed his MS and PhD in Marketing from Cornell University. He previously taught at the Stern School, New York University. Prof. Sudhir is currently the Editor-in-Chief of Marketing Science. He has been an Associate Editor at quantitative marketing’s leading journals, including the Journal of Marketing Research, Management Science, Marketing Science, and Quantitative Marketing and Economics. He has been a pioneer in structural empirical methods in marketing to address problems in areas of customer management, salesforce management, organizational buying and marketing channels. His research has won numerous awards, including the Bass and Little Awards at Marketing Science and the Lehmann Award at the Journal of Marketing Research. He has recently begun a research agenda using artificial intelligence and machine learning methods with applications in digital marketing.
Abstract: Lookalike Targeting is a widely used model-based ad targeting approach, that uses a seed database of customers to algorithmically identify matching ‘lookalikes’ for targeted customer acquisition. Typically, the seed database is based on first party customer data from an advertiser. It is then matched and augmented by behaviors/descriptors on the seed customers available with a third party (example: Facebook, Google and LinkedIn), and then used to algorithmically identify ‘lookalikes’ in the much larger third party database for new customer acquisition. Despite its popularity, there is no academic empirical research on Lookalike Targeting and its effectiveness. The research assesses (i) the value of first party data and the role of ‘seed quality’ (ii) the trade-off between seed quality and algorithmic match accuracy on targeting effectiveness (iii) how targeting salience impacts effectiveness. Using the Facebook Lookalike Audience tool, advertising field experiments are conducted for donor acquisition at a non-profit. Seeding based on desired (and closely related) behaviors is surprisingly effective, relative to demographics or behaviors that are weakly related (for example, Facebook engagement or website visits), justifying the overall value of lookalike targeting using first party behavior data. However, conditional on targeting using seeds based on desired behaviors, clicks and donations are not very sensitive to seed quality but highly sensitive to match quality. Specifically, reducing seed quality from top 5% to top 10% of customers had little impact on clickthrough and donations, but reducing matching from the top 1% to 2% significantly reduced clickthrough and donations. But when targeting is made salient, effectiveness improved for low quality matching but had little impact on high quality matching. Given ads to low quality matches are less costly, the increase in effectiveness with salience makes the acquisition cost with high and low match accuracy comparable.
04 September, 2020, Bengaluru: The Behavioural Sciences Lab of IIM Bangalore will host a webinar titled: ‘Lookalike Targeting on Facebook: Seed Quality Versus Match Accuracy’, on September 15 (Tuesday), 2020, from 07.00 pm to 08:30 pm. The webinar is primarily intended for an academic audience.
The webinar will be led by K. Sudhir, James L. Frank ’32 Professor of Marketing, Private Enterprise & Management and Founder-Director of the Yale China India Insights Program, Yale School of Management, and Editor-in-Chief of Marketing Science.
For registration, please visit: Click here
Those interested need to register for this webinar by September 12, 2020. They will be sent a zoom link on September 14, 2020. There is no certificate offered for participation in this webinar.
Contact for information: naureenb@iimb.ac.in /+91-8762666012
About the speaker: Professor K. Sudhir completed his MS and PhD in Marketing from Cornell University. He previously taught at the Stern School, New York University. Prof. Sudhir is currently the Editor-in-Chief of Marketing Science. He has been an Associate Editor at quantitative marketing’s leading journals, including the Journal of Marketing Research, Management Science, Marketing Science, and Quantitative Marketing and Economics. He has been a pioneer in structural empirical methods in marketing to address problems in areas of customer management, salesforce management, organizational buying and marketing channels. His research has won numerous awards, including the Bass and Little Awards at Marketing Science and the Lehmann Award at the Journal of Marketing Research. He has recently begun a research agenda using artificial intelligence and machine learning methods with applications in digital marketing.
Abstract: Lookalike Targeting is a widely used model-based ad targeting approach, that uses a seed database of customers to algorithmically identify matching ‘lookalikes’ for targeted customer acquisition. Typically, the seed database is based on first party customer data from an advertiser. It is then matched and augmented by behaviors/descriptors on the seed customers available with a third party (example: Facebook, Google and LinkedIn), and then used to algorithmically identify ‘lookalikes’ in the much larger third party database for new customer acquisition. Despite its popularity, there is no academic empirical research on Lookalike Targeting and its effectiveness. The research assesses (i) the value of first party data and the role of ‘seed quality’ (ii) the trade-off between seed quality and algorithmic match accuracy on targeting effectiveness (iii) how targeting salience impacts effectiveness. Using the Facebook Lookalike Audience tool, advertising field experiments are conducted for donor acquisition at a non-profit. Seeding based on desired (and closely related) behaviors is surprisingly effective, relative to demographics or behaviors that are weakly related (for example, Facebook engagement or website visits), justifying the overall value of lookalike targeting using first party behavior data. However, conditional on targeting using seeds based on desired behaviors, clicks and donations are not very sensitive to seed quality but highly sensitive to match quality. Specifically, reducing seed quality from top 5% to top 10% of customers had little impact on clickthrough and donations, but reducing matching from the top 1% to 2% significantly reduced clickthrough and donations. But when targeting is made salient, effectiveness improved for low quality matching but had little impact on high quality matching. Given ads to low quality matches are less costly, the increase in effectiveness with salience makes the acquisition cost with high and low match accuracy comparable.