How online audiences are measured is very relevant to magazine publishers, as online becomes an increasingly important platform. Unfortunately the closer one looks at the two principal methods of measurement (site centric and user centric), the bigger the problems seem to be.
As reported by FIPP, two of the world's leading exponents of online audience measurement, Nielsen and ComScore, gave significant papers at the recent Worldwide Readership Research Symposium. Pat Pellegrini spoke for ComScore and Mainak Mazumdar spoke for Nielsen. Both agreed that ‘hybrid' methods of audience measurement represent a big step forward.
Site-centric methods, which place meters on servers, have severe problems with cookie deletion - for instance, some visitors to a site are counted more than once. Multiple users of a given computer might not be counted. Use of multiple computers by the same person is not included. Use of multiple browsers by the same person on a single machine might lead to underestimation. Visitors from outside the home country might not necessarily be identified. A large problem is separating non-human visits to a site, i.e. machine traffic, which is considerable. Site-centric methods measure machines, not people.
User-centric methods, which install metering software on the panel members' computers, have great difficulty in sampling people's use of the internet at work, especially as many firms will not allow external meters/software into the office. Mobiles are hard to monitor. The sample sizes are too small to accurately measure small specialised websites, such as specialist magazine sites. But user-centric methods have the great advantage that they measure people, not machines, and therefore can be integrated with other currencies for media planning, and be built into the big multi-platform databases.
Nielsen and ComScore presented constructive papers on the hybrid methods each is developing. They take their user-centric panels as a base and use fusion or other techniques to incorporate findings from a variety of other sources, to correct errors and plug gaps. For example, site-centric data could be used for small niche websites or usage at work; data from a separate study of mobile users might be taken in. The development of hybrid methods was hailed as a big step forward.
Nielsen and ComScore use different sources for complementing their main user-centric panel to create their hybrids. The methods for both companies are still evolving, and it is difficult to say that one company's method is better than the other, when enhancements are being made so rapidly.
OPINION/FEEDBACK TO THE EDITOR