Evolving models of online audience measurement: developments since vancouver

At the Worldwide Readership Symposium in Berlin in 1995, I presented a paper that noted the rapid growth of the World Wide Web, especially among consumers in the United States, and speculated on the impact that exposures to magazines online might have upon print media audience estimates (McDonald, 1995). In 1997 at the Vancouver symposium, there were several more papers concerned, in one fashion or another, with the development of online media. Denman Maroney explained how log files from web servers were being used to profile web site activity, and he analyzed the considerable discrepancies that resulted from inconsistent definitions and tabulation protocols (Maroney, 1997). Nigel Jacklin and Peter Highland chronicled their experiences establishing a web site and extracting useful data from their own log files (Jacklin and Highland, 1997). Paul Donato presented evidence suggesting that readers read web pages in much the same way that they read magazine pages, but argued that print media and online media could probably not be measured with a common methdology (Donato, 1997). And I presented a paper that built upon my Berlin paper by making empirical estimates of the degree to which “source confusion� from Web-based magazine exposures contaminated audience estimates for regular magazines (McDonald, 1997). However with the partial exception of Denman Maroney’s paper, there was little attention paid in Vancouver to the mechanisms by which web-based audiences were being measured. To some extent, this oversight was understandable. As of October of 1997, there was still considerable chaos in the field of online media audience measurement. Things have settled down a bit since then, though one could hardly claim that anything about the Internet – including its audience measurement methodologies – is yet stable. However the approaches have taken enough coherent shape to merit summary here.

Download Paper (PDF)