Study Questions Quality Of Local Site Traffic

Many visitors to locally oriented news and information Web sites are neither local nor loyal, according to a new study from Borrell Associates.

On average, about 30% of the visitors don’t live in the market and about 25% of the page views are by “fly-bys” who may not return for another year, if ever, according to the study "How Unique is Unique? Gauging the (Actual) Size of Local Web Traffic," which is based on surveys of site visitors conducted over the past two years and some “complicated math.”

The study also found the average local Web site unique visitor count overstates the number of actual local people coming to the site by a factor of five.

“So, a site that tells local advertisers it has 500,000 monthly unique visitors should probably be forthright and start saying it reaches about 100,000 local people,” according to the study’s executive summary.

As long as Web sites sell advertisers on the basis of page views, there’s little concern about not knowing the true size, composition and behaviors of users, the study said.

“However, as many local sites move toward emerging opportunities in charging access fees to individuals, and as CPMs continue to rise because of better-targeted traffic, the need to gauge actual audience size and engagement becomes critical,” the summary said.

The problem of overcounting local visitors plagued washingtonpost.com, the study said.

So, seven years ago, the site instituted a site registration requiring users to reveal gender, year of birth and ZIP code. “Once the site could deliver a purely local audience, sales reps immediately began selling more ads to local businesses.”

Overcounts of unique visitors is a fact of life on the Web, according to the study.

Unique visitors should not be confused with people, it says. In fact, a single person may show up as seven or eight unique visitors in any given month.

That’s because a person may access a site from multiple computers and mobile devices, each time registering as a unique. What’s more, the person may routinely erase the cookies he collects on the computers when he visits a site, which means he will be counted again as a unique visitor the next time he visits the site.

The study divides people who visit news sites into two groups: “fly-bys,” who have been led to the site for a particular story by another site, but have no abiding interest in the site; “core loyalists,” who visit 18-20 times a month, and the “incidental loyalists,” who visit just one to three times per month.

At a newspaper site cited as an example by the study, incidental loyalists accounted for 48% of the persons visiting the site; core loyalists for 27%; and fly-bys for 25%.

The core loyalists drove page views, accounting for 82% of them.

Relying on unique visitors has particular implications for sites trying to decide whether to go behind a paywall, the study said.

A newspaper site with 500,000 monthly unique visitors may assume that 20% or 100,000 will subscribe at $5 a month, yielding $500,000 a month or $6 million a year in revenue.

But, according to Borrell’s analysis of such a site, the actual number of core users -- those who might actually subscribe -- was just 28,100 or 6% of total unique visitors. Assuming a 20% take up rate by them, annual revenue from going to pay would be just $337,200.

Unfortunately for sites trying to bring in local advertisers, 24% of the loyalists (core and incidentals combined) accessed the site from outside the market.