What’s the best way to measure traffic online?

Media managers have long had access to simple statistics about their product’s consumption. Newspaper publishers, for example, have relatively easy access to circulation figures. But more detailed information about consumption habits traditionally has only been available via extensive research, often from outside providers.

That is far from the case online, though. Free blogging platforms track statistics as do most media-sharing sites such as YouTube or Flickr. Rare is the service that even charges to get such information. And free services such as Google Analytics allow anyone with a website insight into how content is being consumed.

But knowing how to interpret those data can be a difficult task. In fact, some argue that all of this data isn’t necessarily a good thing if nobody knows what it means. James Robinson, director of web analytics at The New York Times, is quoted as saying that all of this data is useless if it’s not being used to learn about how people are engaging with content. He said “it’s not about the data, it’s about the insights….it’s not about the page views or click-throughs – it’s about making New York Times customers happy.”

For many news organizations, there is still debate over what the most meaningful metric to follow is. That complicates things for ad buyers just as much as it does for news producers. Knowing what these figures mean will likely be just as important one day as knowing what a newspaper’s circulation is or knowing what a TV station’s viewership is.

The two most often-cited metrics are page views and unique visitors. Page views quantifies the number of times a page was viewed on a site. Unique visitors (often “uniques”) is the number of individuals that visited a site. One person could visit a site and look at three pages; that would be counted as one unique and three page views.

Most web analytics systems are also capable of measuring other things, though,which can be just as useful in analyzing a site’s traffic. The time a user spends on a site is usually measured, and information about each site visitor such as his or her physical location is often recorded.

The reality is that each metric provide by most traffic analysis systems means something different and together paints a picture of how a site’s users interact with it. Because of that, a single metric can be misleading.

A news site may have a huge national story break that draws in users from all over the country. Each one, though, may only view one page and never return to the news site. That person would still be counted as a unique visitor. Therefore, only looking at “uniques”, while accurately portraying the site’s brief wide reach, would overvalue the site’s true audience. That can be an important concept to understand for both content producers and advertisers.

“You really need to know the level of engagement to sell online advertising,” writes Dana Chinn, who writes about news industry web analytics on the News Numbers blog. “And, you really need to know how engaged people are if your business model depends on paid subscribers or content.”

Others rely on metrics that measure progress toward an organization’s specific goals. BusinessWeek divides its total content output and divides it by the number of comments it receives online from readers in addition to looking at more traditional metrics.

Gawker, the popular New York-based gossip blog, focuses on “branded traffic”. They count the number of people who actually type in the site’s address — Gawker.com — or who search for Gawker on a search engine and end up on the site. That allows them to measure the number of people who intentionally end up at the site, rather than those who are led there by a search query.

There are also certain things that systems simply aren’t able to measure. For example, many systems are incapable of determining how long a user spends on the last page he or she views on a given website, which can be problematic for news sites. And some web browser settings can block user-tracking efforts.

Advertisements are also sold using different metrics. Some advertisements are sold online just like they are in other media — for a fixed period of time. But others are sold based on traffic and others are based on performance. Traffic-based methods are often sold on a cost per thousand (commonly CPM or cost per mille) basis. With CPMs, ad buyers purchase ad views. So an advertiser may pay for an advertisement to be viewed 10,000 times. Using performance models, an advertiser often pays for clicks. This is how Google’s lucrative ad system functions.

Regardless of what metric is used, it is becoming more important to understand the data. In some cases, as mentioned in a past post, this data is being used to determine how journalists are paid. And in most cases, the data is being used to determine how much advertisers pay.

With that in mind, will metrics begin to play an increased role for news organizations in shaping actions? Should news organizations look to this data to determine what is newsworthy and what people care about or are these just means for setting ad rates? Is it healthy to expose journalists to their own traffic figures? For advertisers, what statistics can one reasonably expect to be provided with? And what model is the ideal method for both advertisers and content producers for selling advertising?

For those interested in learning more, here is a list of Internet terms that includes many web analytics terms. Google’s Conversion University also has a great deal of information on the topic, but some of the information relates specifically to Google’s Analytics product.