WEB ANALYTICS
Published: November 20, 2007
The counting controversy: a user's guide (page 3 of 5)
 

Audience measurement
It did not take long for companies to pop up that approach the web from an advertising standpoint. First on the scene was I/PRO (Internet Profiles), which performed site audits. This stabilized terminology like visitors, unique visitors, adviews (impressions) and other factors. But it did not provide demographics. The current main contenders for this arena are comScore Media Metrics and Nielsen NetRatings. Both count advertisers, agencies and publishers among their clients.

These two companies both use a sample-based methodology. As with any counting methodology, including those analytics companies mentioned above that "count everything," each methodology comes up with different numbers. ComScore's numbers are generally higher than NetRatings,' but not always.

Both do provide demographics and country by country data, which is very helpful and, they do a good job of parsing out bots and spiders, etc. and trying to show a "real" audience. However, to some degree, the rankings that come out of comScore and NetRatings are vanity numbers for publishers to have bragging rights as they do not measure advertising; they measure sites. With the exception of site takeover buys, the total audience of a site is rarely bought by an advertiser or agency.

There is no question that a higher ranking in comScore or NetRatings can produce more calls from advertisers for a site, so these ratings services are important. But because they cannot measure advertising impressions, they are not the basis for proof of performance or media payment. Both services talk about producing data that can measure advertising, but there remain inherent problems in the sample methodology when used for digital metrics. They both continue to mis-use the media term "reach" and there is an expectation that they should have numbers closer to the web analytics numbers, even though that might not be practical.

Both services are working with the Media Ratings Council (MRC) to be audited. The MRC was first authorized by Congress decades ago to ensure that TV ratings were legitimate, and is now tacking issues in the digital space. There seems to be a general feeling in the industry that this will result in "better numbers." It may result in more consistency, but a sample, especially in the digital world, has its limitations. This is especially true when we want to take the resulting data and slice and dice it for demographics, geography, etc.

The new entry in this space is Quantcast (full disclosure, I have advised this company in the past ...but then I have advised comScore along the way, too). The Quantcast methodology is to do a census rather than a sample. Konrad Feldman, co-founder of Quantcast, recently said, "We need to have data more closely connected to the way the buyers want to evaluate the media. Census level is the only way to bring audience measurement and message delivery together."

To measure on a census basis involves handling an incredibly large volume of data, but with processing speed increasing and the cost of server space rapidly decreasing, this barrier is going away. Adam Gerber, chief marketing officer of Quantcast, expanded on Feldman's statement. "In an addressable advertising world, the process of audience evaluation during the planning process cannot be separated from the actions during the implementation process. Today we do not have these tools. There needs to be a greater depth of information to the media planner. Panels alone cannot accommodate measurement of the distribution model with the way that media flows today. Measurement needs to be more dynamic."

<< Previous page | Next page >>

White Paper Library

View More Research »