• by Phil Leggiere , September 19, 2007

In their quest for data, marketers often lock in on browsing and surfing behavior (gathered non-consensually) as the royal road to unlocking consumer value. Understandably -- but short-sightedly, as Chase Norlin, CEO of image and video search firm Pixsy, explains below. Browsing data may give an interesting snapshot of what consumers are sampling online, Norlin says, but real understanding of what consumers want requires a consensual opt-in model of consumer cooperation and participation in behavioral targeting.

Behavioral Insider: How did Pixsy get involved in image and video search?

Chase Norlin: Image and video search are the fastest growing consumer search verticals, and of the two, image search is actually the more popular, even though people are more obsessed with video. So our premise all along has been: if these search verticals are so wildly popular, why doesn't every site have it?  

The idea we started with was to make image and video search easily accessible to any site, from a huge commercial site like Travelocity to a small blog site or MySpace user.  We did this by creating a media platform that aggregates content from across the Web, slices and dices it into an organized database, and then enables any site to get access to that material via image or video search.   Publishers benefit from new search activity, content, and monetization for their sites; content owners benefit from traffic being driven to their material.

BI: What are the challenges of using behavioral data to target this kind of search activity?

Norlin: One unique thing about search, the reason it's the best targeting mechanism ever known, is that there's a real window into consumer intent from keywords used. Monetization is the next challenge and big opportunity. Monetization of image and video search is in its infancy. The reason is, you don't have the same volumes of commercial keyword content as you do in algorithmic text search. People don't search for ‘mortgage videos,' for example. On the display side, graphical networks are reluctant to run branded ads on images and video search due to the potential for ‘mixed' results.  This is why Pixsy is building a new advertising system to address the opportunity for monetization of image and video search results across our large network of publishers.

BI: What will that entail? Could you describe the road map?

Norlin: The road map to using behavioral targeting is for advertisers to understand that that's where their users are going. There are really two ways, fundamentally different ways, of looking at how to optimize the behavioral potential in multimedia, and they pivot around whether there's consent by users or not. On the one hand, there's a non-consent model, which is what behavioral targeting has been for the most part thus far. That's entailed collecting as much browsing data as possible from cookies and predicting interests based on that. I think as consumers and marketers become more savvy about what they want out of the Web, there's a new paradigm emerging, which is consent-based.

BI: Could you elaborate on that?

Norlin: The reason is that over time consumers have moved from having their Web experience being primarily about surfing to being about utility, about actually being able to find what's important to them personally in a short period of time. As this happens, the Web experience becomes about doing less, not more. For Web site publishers looking to keep visitors engaged, and for advertisers, that implies a fundamental shift from a model based on non-consent to a consent model, where you provide consumers with specific value in exchange for information they provide about their interests and the type of content they want to see. One example might be a frequent visitor to travel sites [who] could opt in to see ads for special discounted fares in flight lanes they're particularly interested in.

BI: So in a way this involves a rejection of the cookie or pull model of web targeting that's been predominate?

Norlin: The irony is that we may well find that the next phase of behavioral targeting will actually revive the long-discarded Pointcast model from the earliest days of the Web.  In this model, instead of having to go out and surf the Web for what they wanted, consumers could define what kinds of information they were most interested in and have that information ‘pushed' to them. We've gotten to a point where there's so much information out there, so much extraneous noise, that finding what we want, when we want it, is becoming increasingly more difficult.  As the Web gets deeper and more complex, its intrinsic value becomes derived from simplicity and immediate gratification.

BI: So, similarly to how email has evolved, you see opt-in as the lynchpin of next-generation behavioral targeting?

Norlin: Another irony I think we may be looking at in the next year or two is the resurgence of consent-based adware style technologies. While they've received a bad reputation in the past for deceptive practices and consumer privacy issues, ultimately their consent-based model represents some of the most innovative thinking in the ad targeting world. Opting in is going to become even more important to the Web of tomorrow.  For example, Google giving away cell phones in exchange for targeting and monetization.  Not much different than the value exchange between opt-in ad providers and consumers. 

As the Web becomes more and more fraught with issues like identify theft, credit card fraud, etc.... consumers will end up trying to protect their personal information and behaviors even more.  The way for marketers to get at that is not through deception, but by having consumers willingly share that information because of the value being offered in exchange.