Tuesday, April 08, 2008

No precise measurement for online traffic

Website traffic is measured by several factors: page views, unique visitors, time spent on a page, etc., but David Cohn of Columbia Journalism Review writes that there is no precise way to calculate web readership or traffic.

Cohn asks: Why would advertisers choose a news site over one like Google, which obviously garners thousands of viewers, if there is no accurate way to measure web traffic? And how can they agree on rates?

"This may be the most measurable medium in history, but the measurements all suck," says Steve Yelvington, an Internet strategist for Morris DigitalWorks, which manages websites for more than sixty newspapers and thirty radio stations.

Most publishers know their statistics through census data from web servers that count how many times a site is visited. However, the data includes "bots" and "spiders" that search engines send out to index content for searchers, and they inflate the number of visitors counted by internal servers, mostly because of "cookie deletion." Cookies allow a page to recognize a viewer, so unique visitors can be counted. But with users constantly emptying their cookies, a site's traffic can be increased by 150%, according to Andrew Lipsman, a senior analyst at comScore Inc., a global Internet information provider. Essentially, 10 million viewers is probably closer to 4 million.

Organizations such as comScore and Nielsen/NetRatings will share traffic numbers within a plus-or-minus 2% error margin from panelists of randomly selected sites who agree to be monitored. Cohn writes that these sample-based numbers can underreport, especially since these panels don't account for international traffic or traffic from work computers.

Though the panel and census system aren't perfect, publishers and advertisers can make relative comparisons within their market with the two.

George Ivie, executive director and CEO of the Media Ratings Council (MRC), wants to reconcile the two systems with his not-for-profit trade association. The council began with TV and radio audience size, but turned its attention to the Internet in 2002 and wrote standards for audience census data, including page views, clicks on a page, and time spent on a site, all new measurements with the Internet.

MRC requires news organizations that want their traffic counts to be approved by the ratings council to adopt the council's standards, such as counting only content that is accessed through an end browser.

Only a few organizations have been audited by MRC, such as Yahoo, msn, and AOL and advertisers like Atlas and Doubleclick, giving them an advantage when negotiating advertising rates, according to Cohn.

In 2006, MRC began auditing comScore and Nielsen/ NetRatings. Nielsen/Net-Ratings had decided that time spent on a site was more important than total Web page views because visitors spend more time on a page when watching videos and new programs allow visitors to interact with a page (like voting in a poll) without having to reload an entirely new page. But, having a window open does not necessarily mean that the visitor is engaged the whole time, thus this measurement is also inaccurate.

Because the Internet is always changing, there may be new metrics in the future to retrieve more accurate statistics.


No comments: