Widgets And BT: Charting Changes In Online Behavior
by Phil Leggiere, Jul 18, 2007, 12:30 PM
The simplest, most fundamental rule of advertising has always been to follow consumers where they are. However simple in theory, this rule is becoming ever more complex in online practice, as content usage patterns and preferences increasingly move beyond the conventional metrics and measures of publishers and advertisers alike, as Quantcast CEO Konrad Feldman explains below,
Behavioral Insider: What do you see as the biggest gap or misconception in the way current Web analytics is related to targeting strategy?
Konrad Feldman: Advertisers and publishers have moved decisively beyond thinking only in terms of content targeting to trying to understand who’s consuming the content. That’s been a big and necessary step. However, the more you want to understand consumer behavior, the more important it becomes to look to where and how consumers find and use content. Fragmentation offers a tremendous challenge to traditional schemes of measurement because simply so much of what consumers consume in terms of content is not encompassed in page views or on-site behavior as conventionally understood. It needs to be understood more holistically, in a network sense. What we’ve attempted to do is evolve strategies for measurement -- and thus monetization -- that more accurately reflect the new reality, and a model that contains searches and Web page visits but goes beyond that.
BI: Why are you honing in on video and widgets in particular?
Feldman: The two most dramatic and compelling areas where consumer behavior is not really understood well yet are video and widgets. These are areas where consumers are clearly spending more time, yet from a publisher and advertiser point of view the behavior has been difficult if not impossible to track in a meaningful way.
BI: What’s unique about the approach you’re taking?
Feldman: What we do is offer a free service where Web publishers of any size can utilize a personalized tag which is placed in each of their site pages. Beyond the top traffic sites most measurements and analysis of the so-called long tail is based on projection and speculation. Our mandate has been to ground measurement of the long tail in site-specific data for tens, even hundreds. of thousands of sites. We launched that last September. The challenge currently is to extend what we’ve already begun with Web site measurement into the new applications.
The beta version we’ve just introduced for video and widgets reports on reach, which we define as the total number of downloads of a specific video or widget. We also measure plays, the number of times a widget or video is interacted with over a specific period of time. Next is the content category of each particular media element and the amount of time users spend interacting with them.
The widget and video measurement focuses on usage of all Flash-based applications whether they occur on a publisher’s Web site or are distributed from that site elsewhere. In order to really understand the new environment of video and widget consumption you need to understand the dynamics of distribution, how and where videos and widgets are sent, forwarded or linked. You also need to understand frequency or how often a visitor is exposed to a given media element over a period of time.
BI: What other kinds of behavior are you planning to track?
Feldman: The fundamental difference between traditional content usage and how video, widgets and other emerging phenomena like online games are consumed is that they are far more widely and quickly distributed between users via sharing, messaging, recommendations and other peer-to-peer mechanisms. So it’s not as if content X is found on page Y and is a fixed target. They also are far more dynamic, in that audiences for any given content are likely to form, aggregate and disperse with incredible speed, far faster than traditional measurement schemes can even begin to keep up with. It’s not like you have a TV show where there’s a build-up in promotion and publicity. There’s much more of a rapid spontaneous emergence, and once a video or widget ‘goes viral’ it’s totally unpredictable in scale and duration. In order to monetize that, you need an infrastructure in place that measures in an ongoing way just how a video or widget is being used.
The goal is to associate demographic and psychographic characteristics with particular widgets and videos. There are a few different techniques we can use to do this. In a panel we take known demographic indicators derived from things like Zip codes and IP addresses and correlate them with Internet usage browsing and download activity. We do this through what we call a mass inference engine, which allows us to statistically infer profiles. Another key technique is siteography, which shows other sites an audience frequents or is likely to frequent based on their behavior and interests. If you combine panel-derived profiles with deeper inferences about site preferences, you can begin to make sophisticated lifestyle assessments.
BI: Where do you see video and widget tracking going over the next six to 12 months?
Feldman: The next phase is to be able to target and deliver ads based on where consumer’s interests and attention are gravitating, at truly Internet speed. Audiences are going to keep fragmenting, grouping and regrouping in a myriad of ever-changing ways. So there’s no choice for marketers but to evolve ways of measuring the dynamic behavior of both individuals and audience groups collectively.