In this paper the authors outline the underlying theory of Internet advertising exposure in today's ad server world. That theory is reduced to a formula called the Probability Model by Exposure Class (PMEC).While helpful in informing thinking about advertising campaigns, the PMEC formula is not practical in day-to-day use. The authors then outline a simulation system based on consumer panel data that was used to build a pragmatic model which produces nearly identical estimates as that produced by the PMEC, but is easier to use in a fully developed Internet reach and frequency planning system. Results from the new model, Coffey-Mazumdar Internet Reach &Frequency model (CMIRF), are compared to those of the PMEC model. Last, a summary of learnings and an outline of potential applications are provided.
The population of the Internetâs World Wide Web users has grown at astonishing rates during the past three years in the United States. As of mid-1997 over one in five American households surfs the web. In order to realize the potential of this market businesses advertisers and researchers need an accurate and reliable measurement system that competently measures the web and other interactive behavior not only of established users but of infrequent users and of new users as they first come onto the web. The authors demonstrate the importance and relevance of new and infrequent web users and offer a sample design that successfully captures these key users.
This paper is in three parts, to be delivered in turn by the three speakers. First, we comment on the fact that the development of computing and of scanners has revolutionized both market and media research. We point out that the boundary between the two fields is becoming increasingly indistinct: market researchers are finding themselves more and more involved with media matters, and media researchers with marketing matters. To illustrate this, we describe two services of Nielsen Marketing Research (U.S.): the Scantrack store movement service based on retail scanning, and the Nielsen Household Panel which employs in-home scanners to record product purchasing: and we present two new practical applications of these research services which can contribute to the assessment of television advertising. The first application is experimental multimarket testing to assess the relative effectiveness of television plans, in terms of campaign testing, weight testing, or interactions with other media. The second application is a syndicated media/product service, newly introduced in the U.S., based on the Household Panel. This service, called HOME*SCAN, links product purchasing data on an aggregated (12 month) basis to media data collected from individual adult members of the Panel. Prospective television applications of the HOME*SCAN data are discussed.