Specifically, David notes that most publishers running Quantcast tags "find their Quantcast audience to be higher than that reported by either comScore or NetRatings and closer to their log file reports." There is a good reason for this: the tags essentially mirror the counting techniques employed by site-centric, census-based Web Analytics solution providers.
In this and other spaces over the past 18 months, we have collectively vetted the differences between internal, site-centric counts of machine intercourse, and panel-based, user-centric people counts. I think we all generally understand that the panel-based Unique Visitor, for example, is fundamentally different from the site-centric Unique, which typically counts unduplicated cookies (I would refer you to the comment about this on David's post, from the always insightful John Grono of Australia's GAP Research. But then I'd have to link to the post.)
Recently I was working with a publisher who wanted to understand why the comScore counts for Unique Visitors were so divergent from their other sources -- Quantcast, Google Analytics, and Hitbox, all of which were clustered closely together. Now, I'm about to describe a personal epiphany that will strike about half of you as painfully obvious, but bear with me; I'm an audience measurement guy -- a Media Research guy -- not a Web Analytics guy. So this publisher has me navigate to their home page and right-click to view the source code, then scroll down. There at the bottom, one after the other, were the tags for Google Analytics, Hitbox, and Quantcast.
See, there's knowing, and then there's KNOWING. In that moment I got a visceral understanding of this whole tagging thing that I hadn't had before. When you put essentially the same tags in the same place on the same pages of the same site, then yeah, chances are the resulting counts will be pretty darned close. Sure, I knew this. But now I KNOW it.
Here's another anecdote from the real-life trenches of online metrics. So I'm working with a different publisher, whose internal data says they have about a million and a half U.S. Uniques monthly. ComScore shows somewhat fewer. In the course of our conversations, I got curious about what Quantcast showed, so I went and checked. About 200,000 monthly Uniques (the publisher was not tagged by Quantcast.) Presumably, if that publisher added the Quantcast tags, their reported audience would line up with their internal data. But David's point about Quantcast is not about that 200,000 UV estimate; rather, it was about the site-centric 1.5 million Uniques they'd presumably show if this publisher was tagged. And that, pretty clearly, is Web Analytic data. Indeed, if I were to predict, I would speculate that the most direct competitor to Quantcast ends up being Google Analytics, because I think they may also end up going the open source route.
This is in no way meant to cast aspersions on Quantcast (the pithy email I'm about to get from Adam Gerber notwithstanding.) Rather, it is to suggest that Quantcast represents the next generation, not in Media Research, but in Web Analytics. I grew up in Media Research, and am still learning the basics of Web Analytics. But I would argue that the potential revolution Quantcast represents is in the Web Analytic space. Historically, site-centric tagged data was internal data, available to the publisher but not the world at large. Quantcast provides open source web analytic data; your site-centric data published in one place by a third party. Web Analytics 2.0. This is a good thing, and all kidding aside, I hope my friends at Quantcast are OK with that characterization. As I've been saying all along, Audience Measurement (or Media Research) and Web Analytics are two overlapping halves of the Online Metrics space. I just like to keep clear who falls into which half.