Increasing Yield by Dropping Constraints in Ad Serving
Brendan Kitts (Lucid Commerce)
Abstract
A persistent challenge with web-based ad-servers is that the requested advertiser requirements for targeting criteria and delivery may be infeasible. In order to account for this, we pivot the problem to one of error minimization. The system is deployed in one of the largest ad-servers in the United States, where we present results from test ads as well as live advertisers.
