Experts split over how to take the pulse of the web

By DITA DE BONI

New Zealand's young internet and mature advertising industries face an issue of Olympic proportions: which method of measuring internet traffic is the most reliable?

As the globe is gripped with Olympic Games fever, media outlets have been barraged with lists from internet research companies listing the popularity of sites visited during the event. But no two are the same.

While all research companies can prove the internet's dazzling popularity as an increasingly important information medium for events such as the Olympics, the outstanding question is one of method.

Two methods are battling to become the Holy Grail of web traffic measurement and supply a market crying out for better information on which to base multimillion-dollar web campaigns.

Unfortunately for those new to the measurement game, some industry commentators put differences between the various measurement methods as high as 300 per cent.

The first method, commonly known as panel-based measurement, is used by the likes of international research giants ACNielsen, Media Metrix (which will move into the New Zealand market soon) and NetRatings.

Panels work by recruiting users throughout a region who agree to have their net movements tracked with special software supplied by the research company. The user registers his or her demographic information, but is not individually identified, allowing the method to reflect porn site usage and other popular murky activities.

Log file-based traffic measurement methods from internet service providers work on a totally different premise. These track hits through selected internet service providers or web servers, says Hitwise's Brendan McKeegan.

"Our advantages are that we can see a larger percentage of the market easily, see patterns, and while we are not a traditional research firm we consider we provide real-time web intelligence.

"Of course, we are also affordable for everyone who has a website, rather than our competitors, which are only really accessible to the top 100 companies."

Both methods have their disadvantages. According to Hitwise, the University of Auckland and the BNZ are ranking highly this week yet - while not a direct comparison - these sites did not figure at all in ACNielsen's rankings for last month.

While researchers employing panels must be scrupulous about getting a representative group of users, log file gaugers must be able to extrapolate results from a representative group of ISPs.

Xtra, for example, claims to have 310,000 subscribers but is not a Hitwise client as yet.

Xtra marketing manager Chris Thompson says it uses a variety of research companies and conducts its own research.

He says it is important to do your own research with customers and find out "exactly what you want to find out" before turning to a research company.