Optimization in Online Content Recommendation Services: from Clicks to Engagement
Coauthor(s): O. Besbes, Y. Gur, and A. Zeevi
A new class of online services allows publishers to direct readers from articles they are currently reading to other web-based content they may be interested in. We study the content recommendation problem and its unique dynamic features from both theoretical as well as practical perspectives. Unlike related classes of online services that focus on instantaneous objectives, optimal content recommendations must consider future clicks of readers, and hence myopic optimization of click-through-rates is strictly suboptimal. Using a large data set of browsing history at major media sites, we develop a representation of content along two key dimensions: clickability, the likelihood to click to an article when it is recommended; and engageability, the likelihood to click from an article when it hosts a recommendation. Based on this representation, that allows to isolate the one-step lookahead value of candidate articles, we propose a class of user path-focused heuristics, whose purpose is to simultaneously address a high instantaneous probability of clicking recommended articles, while also optimizing engagement along the path of readers. We test and validate these heuristics and quantify the potential associated with those. The work is based on a collaboration with a leading provider of content recommendations to online publishers.
O. Besbes, Y. Gur, and A. Zeevi "Optimization in Online Content Recommendation Services: from Clicks to Engagement." , Columbia Business School, (2013).