Every publisher or network who deals with Comscore or Neilsen has probably complained at some point about undercounting. Bob Bowman at MLBAM has declared war, and there’s a general consensus that their methodology produces inaccurate results, but that “you have to play the game.” I think Comscore’s underassessment of Yardbarker’s traffic is an institutional flaw that is preventing many brand advertisers from recognizing that small, high quality blogs can be powerful marketing vehicles.
The Yardbarker Network broke the 400 publisher mark this week. According to our internal numbers, the sites in the network collectively reached over three million unique users last month, so while most of the sites in our network are small blogs, when you combine them, there’s some real reach there. The number of sites in our network has doubled since January, and we’re starting to scale our ad sales team to fill the inventory with more and better campaigns. Our story about reaching the new generation of sports fans, who spend more time on blogs than on big sites like ESPN.com, is starting to resonate with brand advertisers like Dos Equis, 2K Sports, and Levi’s.
Our INTERNAL numbers had us at three million last month. According to Comscore’s new survey of the traffic that we paid them to “roll up” across our network – and that every advertiser and agency looks at to evaluate our traffic – we’re at 565,000 uniques. Ugh.
Comscore measures traffic by surveying a sample of 120,000 web users. The smaller the site, the less accurately Comscore can count its traffic. That's because there are 190 million web users in the US, meaning Comscore surveys a little more than half a percent of the total web audience. So, a site needs to have over 1,500 unique users per month in order for it to be more than random for even a single panelist to report visiting that site. On average, our blogs have 7,500 unique users per month, but the smaller sites really need to get lucky to have Comscore panelists recall visiting them. Because Comscore counts each site in our network individually to derive our network’s reach, the problem is compounded for our network, and we need all of our sites to get lucky in the same month in order for our network’s traffic to be represented properly.
For us, this (along with some other factors) means that Comscore undercounts our traffic by 5x, since most of the 400 sites in our network are too small for Comscore’s very 1.0 system to count correctly. We could represent 4000 or 40,000 or 4 million of these small sites, and Comscore wouldn’t be any better at counting our overall traffic, though of course our REAL reach would be much more significant.
Newer companies like Quantcast and Compete.com are addressing this problem, but as long as the dominant audience measurement services insist on pursuing a methodology that ignores small sites and promotes big ones (who write them the biggest checks), advertisers will miss opportunities to reach the audiences that populate those niche pockets of the web. But in a good sign of things to come, we’re starting to find advertisers who are savvy enough to acknowledge the issue, and who are interested in planning creative campaigns to reach sports fans wherever they happen to hang out on the web.
