Download the Executive Summary
Read the Press Release
Online ad effectiveness research is an important tool for marketers seeking to understand how their campaigns perform. However, it is challenged by serious methodological limitations. Questions around recruitment, sample bias and deployment are hampering the validity of this research and undermining the industry as a whole. The growth in online advertising spend requires sound measurement and reliable methodologies to prove effectiveness.
By examining each component of online ad effectiveness research, the IAB hopes to determine best practices along a range of methodologies. To accomplish the goals of developing and articulating best practices, the IAB commissioned Marissa Gluck of radarresearch. This document provides a set of recommendations so that the industry can develop more rigorous research while employing better standard operating procedures during the scientific inquiry period.
While the intent is not to endorse one methodology over another, there are some clear trends emerging as industry dissatisfaction with live intercept recruitment increases. With live intercepts falling increasingly into disfavor due to lowered response rates, greater site clutter, and misalignments between campaign delivery and sample, panels are gaining favor with agencies, publishers and vendors. However, panels are far from an industry panacea today. If properly validated, panels can solve for some of the methodological deficiencies of intercept studies as we know them.
This paper looks across the spectrum of available methodologies to assess best practices in each phase of online effectiveness research, from planning to recruitment, deployment and finally, optimization and analysis. Within each phase, we examine the challenges the industry faces and propose prescriptive remedies to each of these challenges. The issues we look at include organizational hurdles such as planning and staffing, as well logistical and technological impediments such as cookie deletion and declining response rates.