Rachel Fan
Research Areas
Authored Publications
Sort By
Preview abstract
Connected TV (CTV) devices blend characteristics of digital desktop and mobile devices--such as the option to log in and the ability to access a broad range of online content--and linear TV--such as a living room experience that can be shared by multiple members of a household. This blended viewing experience requires the development of measurement methods that are adapted to this novel environment. For other devices, ad measurement and planning have an established history of being guided by the ground truth of panels composed of people who share their device behavior. A CTV panel-only measurement solution for reach is not practical due to the panel size that would be needed to accurately measure smaller digital campaigns. Instead, we generalize the existing approach used to measure reach for other devices that combines panel data with other data sources (e.g., ad server logs, publisher-provided self-reported demographic data, survey data) to account for co-viewing. This paper describes data from a CTV panel and shows how this data can be used to effectively measure the aggregate co-viewing rate and fit demographic models that account for co-viewing behavior. Special considerations include data filtering, weighting at the panelist and household levels to ensure representativeness, and measurement uncertainty.
View details
Preview abstract
Co-viewing refers to the situation in which multiple people share the experience of watching video content and ads in the same room and at the same time. In this paper, we use online surveys to measure the co-viewing rate for YouTube videos that are watched on a TV screen. These simple one-question surveys are designed to optimize response accuracy. Our analysis of survey results identifies variations in co-viewing rate with respect to important factors that include the demographic group (age/gender) of the primary viewer, time of day, and the genre of the video content. Additionally, we fit a model based on these covariates to predict the co-viewing rate for ad impressions that are not directly informed by a survey response. We present results from a case study and show how co-viewing changes across these covariates.
View details
Methods for Measuring Brand Lift of Online Ads
Tim Hesterberg
Ying Liu
Lu Zhang
Proceedings of the 2018 Joint Statistical Meetings, American Statistical Association
Preview abstract
We describe ways to measure ad effectiveness for brand advertisements using online surveys. We estimate the causal effect of ads using randomized experiments. We focus on some technical issues that arise with imperfect A/B experiments--corrections for solicitation and response bias in surveys, discrepancies between intended and actual treatment, and comparing treatment group users who took an action with control users who might have acted. We discuss different methods for estimating lift for different slices of the population, to achieve different goals. We use regression, with a particular form of regularization that is particularly suited to this application. We bootstrap to obtain standard errors, and compare bootstrap methods.
View details
Preview abstract
Many advertisers rely on attribution to make a variety of tactical and strategic marketing decisions, and there is no shortage of attribution models for advertisers to consider. In the end, most advertisers choose an attribution model based on their preconceived notions about how attribution credit should be allocated. A misguided selection can lead an advertiser to use erroneous information in making marketing decisions. In this paper, we address this issue by identifying a well-defined objective for attribution modeling and proposing a systematic approach for evaluating and comparing attribution model performance using simulation. Following this process also leads to a better understanding of the conditions under which attribution models are able to provide useful and reliable information for advertisers.
View details
Brand Attitudes and Search Engine Queries
Jeffrey P. Dotson
Elea McDonnell Feit
Yi-Hsin Yeh
Journal of Interactive Marketing, 37 (2017), pp. 105-116 (to appear)
Preview abstract
Search engines record the queries that users submit, including a large number of queries that include brand names. This data holds promise for assessing brand health. However, before adopting brand search volume as a brand metric, marketers should understand how brand search relates to traditional survey-based measures of brand attitudes, which have been shown to be predictive of sales. We investigate the relationship between brand attitudes and search engine queries using a unique micro-level data set collected from a panel of Google users who agreed to allow us to track their individual brand search behavior over eight weeks and link this search history to their responses to a brand attitude survey. Focusing on the smartphone and automotive markets, we find that users who are actively shopping in a category are more likely to search for any brand. Further, as users move from being aware of a brand to intending to purchase a brand, they are increasingly more likely to search for that brand, with the greatest gains as customers go from recognition to familiarity and from familiarity to consideration. Additionally, users that own and use a particular automotive or smartphone brand are much more likely to search for that brand, even when they are not in market suggesting that a substantial volume of brand search in these categories is not related to shopping or product search. We discuss the implications of these findings for assessing brand health from search data.
View details