figshare
Browse

Optimizing Ads Marketplace Diversity

Download (1.49 MB)
preprint
posted on 2020-06-09, 00:39 authored by Ashim DattaAshim Datta, Kevin McLaughlin, Javier Llaca Ojinaga

Advertisements have become nearly ubiquitous on web pages and mobile applications as monetizing users via advertising has become a common practice for generating revenue in otherwise free-to-use services. In these marketplaces, advertiser value is generally expressed in the form of bid (with advertiser surplus expressed as bid-minus-price of the ad), the value provided to users is often determined based on less tangible metrics such as the relevance of the ad in the context or with outcome variables such as click through rate (CTR), with platform value being measured by revenue capture in the marketplace. In this paper we discuss how we have developed a novel procedure which incorporates users’ ad history in order to make their ad experience more diverse. We then show how this procedure increases the CTR of ad campaigns in the short term, while in the long term results in increases in the number of users generating revenue, increases in total impressions (organic and promoted content combined), and increases in engagement with both organic and advertising content. We are able to achieve this all without a loss in revenue capture for the platform, thus providing a Pareto improvement for the marketplace.

History

Usage metrics

    Licence

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC