2016 William F. O'Dell Award
From Generic to Branded: A Model of Spillover in Paid Search Advertising
Oliver J. Rutz and Randolph E. Bucklin
Journal of Marketing Research: February 2011, Vol. 48, No. 1, pp. 87-102
What motivated you to research this particular area?
Our research problem originated with a question posed to us by a major advertising agency. In the course of managing paid search ad campaigns on the Google and Yahoo! search engines, they had noticed that “generic” keywords (i.e., search terms that do not include a brand name or trademark like “hotels Los Angeles”) significantly underperformed “branded” keywords (e.g., “Hilton Los Angeles”) based on the easy-to-compute metrics of cost per click and cost per conversion. They wondered, along with their clients, whether or not to curtail their spending on generic keywords. On the other hand, if search based on generic terms introduced consumers to the brand’s availability -- and that was subsequently followed up with search using branded terms -- it seemed reasonable that generic terms should be credited for at least some of the conversions tied to subsequent branded search. Our research question thus became how much, if any, credit could be tied back to generic search and how could that be estimated using the data available to managers?
Describe any challenges related to research and writing of your article?
Though we found the anticipated spillover effects from the start of our modeling work with the data, the paper itself did not have an easy time in the review process. Perhaps because it was one of the early papers in marketing on paid search advertising, we found that some reviewers were not acquainted with the limitations of the paid search data made available to managers by the search engines. This led to a few requests to conduct analyses that we simply could not do and many questions about our modeling approach. After several rounds, we were ultimately able to convince the reviewers that our findings regarding the spillover effects were sound.
How might marketing managers apply your findings in the field?
We believe that our work has contributed to marketing practice by highlighting the different role played by generic and branded keywords in paid search campaigns and by empirically demonstrating the importance of not relying solely upon easy-to-use “last-click” metrics. Our conversations with practitioners suggest that the issue is now far better recognized than in the past, but the lure of seemingly less expensive branded keywords still continues to affect some advertiser judgments. For example, a marketing executive from a major international company recently proclaimed to an MBA class -- taught by one of our colleagues -- that they use only branded keywords in their search campaigns due to their much lower cost-per-click. We hope that the modeling approach we developed will aid advertisers in solving this important piece of the attribution puzzle and raise the productivity of their search ad spending.
Were there any interesting questions stemming from your article that you hope will be explored in future research?
One important question is whether our findings generalize beyond the data we analyzed from the hotel industry. We are pleased that our approach has been replicated by other academics using data across several industries and that the general phenomenon of generic to branded spillover holds – though, as one might expect, the magnitude of the spillover effect varies.
Our research is also part of the broader problem in digital marketing known as “attribution” – what advertising activities should get the credit for driving customers to buy? A number of other scholars have already made contributions to this problem and attribution modeling has become an important topic for both marketing academics and practitioners.
Questions for the Classroom
Describe the difference between branded and generic keywords in paid search advertising and why they might appear to perform differently in a paid search campaign. Note that branded terms are often referred to as “trademarked” and generic terms as “non-trademarked.”
Should advertisers include both generic and branded keywords in their paid search campaigns?
How can the concept of the purchase funnel be used to think about this problem?