quantcastlogo.png Quantcast this week launches a service which enables searching of websites according to key readership demographics. The new facility enables marketers to enter a list of demographic parameters, and returns a list of websites whose readership matches those criteria. As an example, if a marketer is targeting black male readers, over 40, with an income over $60,000, then Quantcast’s new service serves up a list of websites which attracts that audience, and also tells you whether or not the site accepts advertising. Better yet, the service is free. Just sign up and target away. There are two things everyone with a passing interest in Web marketing needs to know about Quantcast. The first is plain awesome: unlike other public Web metrics trackers like Alexa, these guys gather directly-measured (real) traffic figures directly from the website server (similar to Google Analytics). The other thing you need to know is equally great, if it works. Quantcast profiles a website’s audience along demographic lines like those mentioned above. The proviso of course is: are their inference methods strong enough to return reliable results on audience composition? The Quantcast model is built along the following lines. There are two types of sites: ‘Quantified’ sites and non-quantified sites. Quantified sites are those whose raw metrics are measured directly. You sign up, embed a snippet of code, and allow Quantcast to track all media consumption events on your site. You can control which data is surfaced publicly, and which data is maintained on a private basis. And there is significant additional upside, which we will look at in a moment, and these advantages have moved organizations like Wordpress.com Hosted Network ( Scobleizer, Lolcats etc.) and Gawker Network (Gizmodo, Valleywag etc.) to sign up to be quantified. Given the direct method employed, the raw traffic figures you get from quantified sites should be highly accurate representations of their size. ‘They are as good as a cookie-based metrics gathering system can be,’ in the words of CMO Adam Gerber. And so if you go look at Drude Report’s traffic page, ignoring the demographic breakdown for the moment and focusing on traffic stats alone, you find trustworthy figures which return some insightful results. Like … now we know that Drudge’s traffic is almost exclusively coming from the U.S.A. And that repeat business is huge for Drudge: ‘Addicts’ are those who visit 30 times or more in a month, and while 14% of his readership are addicts, they comprise 75% of his total visits. And so on, and so forth. Obviously, there are a majority of significant websites which are not quantified, and these do not have their raw metrics measured directly by Quantcast. But they still run panel-based profiles on traffic, audience, demographics etc. (data is only US-based today). We had the pleasure of speaking to Adam Gerber, CMO, and he talked a good game about the reliability of the company’s figures for non-quantified sites. To do with panels, and profiling, and a bunch of other stuff that’s doubtless very interesting to a statistician but which whistles straight through the ears of your average Web geek. Nevertheless, we maintain a healthy skepticism regarding the published metrics of sites which have not yet signed up for direct measurement.

Demographic Profiling

When it comes to Quantcast’s demographic figures, it is important to note first of all that all such data is based solely on data inputs from the USA. Quantcast hasn’t yet extended its profiling algorithm to Rest-of-the-World data. Maybe when they get their hands on that second-round funding… There's all kinds of great things you can find out about, say, the average Digg reader from this data. But how do they know all this? It goes back to census level visibility of the internet, more panels, and profiling, and means and medians, and a bunch of other high-falutin’, greek-letter writin’, snore inducin’ math. One specific way Quantcast develops audience profiles is that they track back IP addresses to postcodes, and tally this information against Census data to establish key demographic benchmarks. Given a sufficient sample size, not a bad place to start profiling visitors. And as already mentioned, this method is just part of the algorithm. After all, the people behind Quantcast are serious statistical academic bigwigs from Stanford and NASA, not a bunch of dingbats throwing darts at demographic etchings on the office wall. We think. In any case, consulting our much-abused CMSWire crystal-ball, we can see how a certain non-Evil search company might be tempted to come charging into the Quantcast boardroom waving fistfuls of dollars to get their hands on some of these capabilities. If Quantcast’s demographics modelling is mustard, think what Google Analytics might do with it? The new 'Advanced Search' facility which enables searching according to demographic parameters is currently in Beta. For more info check out Quantcast's blog entry on the subject.