Hongshuang (Alice) Li Robert H. Smith School of Business College Park, MD 20742 Phone: (301) 275-5824 Fax: (301) 405-0146 E-mail: hli2@rhsmith.umd.edu Education Ph. D. in Marketing, May 2014 (Expected), College Park, MD (Minor: Economics) M. S. in Resource and Environmental Economics, August 2009 University of Illinois, Urbana-Champaign B.A. in Agricultural Economics, July 2007 Renmin University of China, Beijing, China (Minor: Applied Mathematics) Publication Li, Hongshuang (Alice) and P.K. Kannan, Attribution Modeling: Understanding the Influence of Channels in the Online Purchase Funnel, MSI Report, No. 12-115. Featured in Insights from MSI (Issue 1, 2013). Manuscript under Review Li, Hongshuang (Alice) and P.K. Kannan, Attributing Conversions in Online Multi- Channel Environment in the Presence of Carryovers and Spillovers, under 2 nd round of review at Journal of Marketing Research. Work in Progress Optimal Design of Content Samples for Digital Products and Services (with Sanjay Jain and P.K. Kannan), Manuscript in preparation for submission to Marketing Science. Bayesian Inference of a Cognitive Process Model of Visual Search (with Michel Wedel, Eliot Siegel, Elizabeth Krupinski and Jin Yan), Manuscript in preparation for submission to Radiology. Understanding the Impact of Attribution Metrics on the Realized Effectiveness of Keywords in Paid Search Advertising (with Siva Viswanathan and P.K. Kannan), Data analysis stage. 1
Advertising Frontiers in the Digital Age: Micro-targeting in Mass Marketing (with Michael Trusov and P.K. Kannan), Data analysis stage. Modeling Online Word-of-Mouth: Are Consumers Good Shoppers or Nice Reviewers? (with David Godes), Model formulation stage. Research Interests Attribution Models, Marketing Resource Allocation Strategies, Multi-channel Marketing, Customer Relationship Management, Social Media, E-Commerce Honors and Awards Best Paper/Presentation at the Haring Symposium, Kelley School of Business, Indiana University, March, 2013 MSI Research Grant Award for proposal titled Understanding the Path to Conversion in E-Commerce Sites with P.K. Kannan, December, 2010 Marvin A. Jolson Outstanding Marketing Doctoral Student Award, University of Maryland, May, 2013 INFORMS Doctoral Consortium Fellow, 2011-12 Dean s Research Fellowship,, 2010-13 Brockson Fellowship, University of Illinois, 2008 09 Outstanding Student in Beijing City, 2007 Special National Scholarship of Excellent Study Performance, 2006 Outstanding Student of Renmin University of China, 2004 Australian Pure Land Learning College Scholarship, 2003-2005 Tsang Hin-chi Excellent Student Scholarship, 2003-2006 Professional Experience Adobe Systems, San Jose, CA June - August 2012 Developed economic models incorporating both demand and supply sides of the online advertising market; developed a validation model to compare the marginal ROI across various attribution models. Marriott International, Bethesda, MD June 2010 - May 2012 Provided attribution analysis across multiple online marketing channels to the ecommerce Measurement & Analysis group Dalian Commodity Exchange, Dalian, Liaoning, China January - July 2007 Provided field survey and consulting on issues in market penetration in Northeastern China. 2
Research Skills Econometrics Analysis in R, Matlab, SAS, Stata Web Crawling by Programming in PHP Bayesian Statistics in R and WinBUGS Spatial Analysis in ArcGIS and GeoDa Teaching Experience Instructor, BMGT452, Marketing Research Methods (undergraduate), Robert H. Smith School of Business, Fall 2011. Teaching Assistant for Prof. David Godes, BUSI650, Marketing Management (MBA),, Fall 2012. Teaching Interest Marketing Analytics, E-Commerce, Marketing Strategy, Marketing Research Method, Marketing Management Conference Presentations Attribution Modeling: Understanding the Influence of Channels in the Online Purchase Funnel, Mid-Atlantic Marketing Doctoral Symposium, Philadelphia, PA, April 2013. Attribution Modeling: Understanding the Influence of Channels in the Online Purchase Funnel, Haring Symposium, Bloomington, IN, March 2013. The Long and Winding Road: Modeling the Influence of Channels in the Online Purchase Funnel, INFORMS Marketing Science Conference, Boston, MA, June 2012. Service President, Association of Doctoral Students at R.H. Smith School of Business, 2012-13 Social Chair, Association of Doctoral Students at R.H. Smith School of Business, 2011-12 Event Committee, Chinese Student and Scholar Association at, College Park, 2007-08 Related Coursework (GPA: 4.0) Marketing courses: Marketing Models in MCMC Marketing Models in R Marketing Models Marketing Strategy Survey of Consumer Behavior Instructor Michel Wedel Michel Wedel P.K. Kannan Wendy Moe Amna Kirmani 3
Experimental Research in Marketing Mathematical Models in Marketing Rebecca Hamilton Yogesh Joshi Economics courses: Instructor Microeconomics Analysis I Daniel Vincent, Lawrence Ausubel Microeconomics Analysis II Erkut Ozbay, Rachel Kranton Risk and Information Charles Nelson Applied Microeconomics Kislaya Prasad Spatial Econometrics Kathy Baylis Applied Econometrics Roger Koenker Applied Econometrics I Richard Just, Anna Alberini, Marc Nerlove Applied Econometrics II Richard Just, Anna Alberini, Barrett Kirwan Bayesian Inference & Measurement Models Robert Mislevy Empirical Microeconomics Raymond Guiteras Empirical Industrial Organization Ginger Z. Jin Computational Economics John Rust Reference P.K. Kannan Ralph J. Tyser Professor of Marketing Science, Chair of Department of Marketing Phone: (301) 405-2188 E-mail: pkannan@rhsmith.umd.edu Michael Trusov Associate Professor of Marketing Phone: (301) 405-5878 E-mail: mtrusov@rhsmith.umd.edu Michel Wedel PepsiCo Professor of Consumer Science Phone: (301) 405-2162 E-mail: mwedel@rhsmith.umd.edu Siva Viswanathan Associate Professor of Information Systems Phone: (301) 405-8587 Email: sviswana@rhsmith.umd.edu 4
Dissertation Abstracts: Dissertation: Attribution Modeling and Optimal Resource Allocation in Online Environments Dissertation Committee: P.K. Kannan (Chair) Michel Wedel, Michael Trusov, Siva Viswanathan, Shapour Azarm Essay 1: Attributing Conversions in Online Multi-Channel Environment in the Presence of Carryovers and Spillovers Current technology allows firms to produce a granular record of every touch point a consumer makes in their online purchase journey before they convert at a firm s website. However, firms still depend on aggregate measures to guide their marketing investments in multiple marketing channels (e.g. display ads, search, referral, e-mail, etc.). In practice, the widely used last-click attribution metric assigns purchase credit to the last touched channel and entirely ignores all the other channels a customer might have touched prior to the purchase. Such aggregate and incomplete measurements, in turn, bias the investment decisions for future marketing campaigns. The purpose of this research is to provide a method to attribute the incremental value of each individual marketing channel (e.g. display ads, search, referral, e-mail, etc.) in a multi-channel environment using individual-level purchase funnel data of customers. This essay proposes a three-level conceptual framework to analyze (1) customers consideration of online channels, (2) their visits through these channels over time, and (3) their subsequent purchases at the website. The nature of carryover and spillover effects of prior visits is explicitly estimated at both visit and purchase stages. Using a rich panel dataset of customer-level visits and purchases at a hospitality firm s website the analysis result shows significant carryover and spillover effects of prior visits to subsequent visits and purchases. For example, prior e-mail clicks trigger visits through Paid Search and Referral channels and lead to more purchases in Paid Search channel. Based on the estimated carryover and spillover effects, the essay attributes the purchase credit to different channels and finds that the relative contributions of these channels are significantly different as compared to the contribution based on the widely-used last-click 5
metric. Specifically, the contribution of E-Mail, Display and Referral channels is underestimated by the last-click metric, while search channels contribute less than what the last-click metric suggests. To validate our model, we conduct a field study at the firm s website by pausing paid search for a week. The result clearly confirms the ability of the proposed approach in estimating the incremental effect of a channel on conversions. In another simulated application, we target customers with different touches in their purchase journey and illustrate how to use the model estimates to identify cases where e- mail retargeting may actually decrease the conversion probabilities. Essay 2: Understanding the Impact of Attribution Metrics on the Realized Effectiveness of Keywords in Paid Search Advertising Essay 2 analyzes the impact of attribution metric (used for imputing conversion credit to search keywords) on the realized effectiveness of these keywords in search campaigns. Different attribution methods assign conversion credits across keywords based on different weights, which affect future budget allocation for keywords, and which in turn determine the effectiveness of future search campaigns. The essay analyzes the impact of attribution metrics not only on the purchase conversions, but also on new customer acquisition, word-of-mouth generation and consideration. Using a six-month panel data of several hundred keywords from an online jewelry retailer, the essay empirically models the relationship among the firm s bidding decision, the search engine s ranking decision, and the effectiveness of individual keywords in leading to various marketing outcomes, accounting for the simultaneity and endogeneity in the system. The key aspect of this analysis is the change of attribution metrics (from lastclick attribution to first-click attribution) executed by the firm half-way through the data window. This allows the estimation of the impact of different attribution metrics on budget allocation, which in turn influences conversions under different attribution regimes. The results show first-touch in general influence all marketing outcomes negatively for the focal firm. However, in the analyses of individual keywords, first-click attribution is found to be a better metric to capture the true impact of broad and generic keywords. 6
Essay 3: Advertising Frontiers in the Digital Age: Micro-targeting in Mass Marketing Essay 3 studies dynamic probabilistic profile-based targeting in the multi-channel marketing context. Many companies use multiple online marketing channels, such as paid search, display ads, referral sites, etc., to reach existing and potential customers. However, in practice, marketing resource allocation is still made within each channel individually, without considering the potential synergy across channels. Additionally, in such a multi-channel environment, there is a probability associated with making a marketing contact with specific customers in a specific marketing channel for example, a customer can be reached by a paid search ad only if she uses specific keywords, or a customer can get a display ad impression only when he visits a specific website. All of these marketing contacts are stochastic events. The manager should jointly consider all these behavior profiles and their associated probabilities in order to determine when and where to reach specific customers. Additionally, aforementioned paid search and display advertising accounts for more than 80% of online marketing spending. Such stochastic profile-based targeting of potential customers leads to an intermediary-assisted targeting practice that is between mass marketing and one-to-one marketing. The essay proposes a model for dynamic profile-based probabilistic targeting in the multi-channel environment. The analyses use two sources data: a company s internal data on customer online behavior related to the firm s website and ACNielsen click stream data that includes customer browsing behaviors at all websites. By fusing the two datasets together, the essay proposes a methodology to plan marketing resource allocation across multiple online marketing channels and improve the overall marketing efficiency. 7