Commentary

I'll Take My Profile To Go

The thing about a "personal profile" is that it generally isn't very personal. In most behavioral targeting systems, the user never sees or touches her own data and rarely has control over when and where it might be applied across a network. The newly launched service for publishers and consumers, matchmine (www.matchmine.com ) tries to make personal preferences and behavioral data both personal and portable by planting it within a desktop application that users bring with them to apply to member sites. User interaction with movie, music and blog material among these partners builds a profile that another matchmine partner can use to surface the most personalized material. A spinoff of the Kraft Group, matchmine has initial partnerships with Fuzz, FilmCrave and Peerflix. CEO Mike Troiano explains how he plans to attract publishers by helping them raise both their CPMs and their page impressions.

Behavioral Insider: A lot of sites drive niche and personalized content to users. What is matchmine doing differently?

Troiano:
When you interact with one of those systems your profile information becomes the property of that Web site rather than your own. We turn that equation around. In our view your preference data is yours; you should be able to take it with you. We created an object we call a MatchKey that you can think of as a visual representation of your personal preferences and tastes. It knows a lot about what you like but nothing about who you are; we exclude all personally identifiable information for the key, so what it gives you as a user is something you can take around and share with any matchmine partner. You save it to the desktop along with an app called Gatekeeper that identifies all the places you can use MatchKey. When you identify a matchmine certified partner it asks if you want to share your profile.

Behavioral Insider: How is it rendering my preferences from my on-site behaviors?

Troiano:
It is an attribute-based system. We are focused on blog postings, video, movies, and music CDs. We score those items along a set of 'canonical axes.' You can think of 72 attributes for music including genre, rock and blues, observational stuff like popularity, beat, acclaim, etc. We rate each of these components along these 72 dimensions, and as you interact with songs on any application or Web interface that supports the matchmine interface, we can over time develop a sense of what your score is. It is not observing the behaviors on your machine in the way that adware has traditionally done. The technology of the key is within a pretty tight sandbox; it evolves only on partner sites or applications.

Behavioral Insider: How does the personalization occur on the partner site?

Troiano:
The initial implementation will be a kind of recommendation or discovery widget that appears in the partner site. Within that widget we display only content that is available on that partner's site, and if you interact with the content of that widget you acquire recommendations.

Behavioral Insider: How does marketing fit into the model?

Troiano:
A publisher may be using a run of network CPM because he doesn't know a lot about a user visiting the site. If that user has a MatchKey, then the publisher not only knows three pieces anonymous demographic data -- zip code, age, and gender - he also knows this is someone who is interested in hip hop and horror movies and blogs about liberal politics, for example. Our initial value proposition on the publisher side is enabling a publisher to improve their CPM by being able to target promotional content towards a user based on the demographic attributes, but over time we expect to enable advertiser to target content along a set of preference attributes very similar to the way Facebook is going to approach the targeting of advertising based on the preferences expressed in your profile.

Behavioral Insider: What is your revenue model then?

Troiano:
To take a share of that incremental CPM and impressions that are attributable to a partner's use of the key.

Behavioral Insider: Publishers have a lot of targeting and behavioral options out there. How do you break through with yours to get the distribution you need?

Troiano:
The value prop for publishers has three components. First, all publishers recognize the need to better personalize their content. Many of them are adopting some kind of collaborative filtering or third-party method of giving recommendations to users who are just overwhelmed by the magnitude of options. So we start by saying we will score your content and help you make personalized recommendations on an outsourced basis and we will participate in the revenue at no cost. Second, there is pretty clear visibility to incremental revenues to an ad-supported publisher. We are not interfering with the revenue they were getting before, and they are sharing in value we create together. Third, as we get more publishers, and they help more of their users participate in the system, there will be more people showing up on their door ready to tell them some 250 odd specific attributes of their personal preference.

Behavioral Insider: How does this model change the usual behavioral tracking equation?

Troiano:
In some ways we're the opposite of what I think of as conventional server based behavioral targeting where you have your preference captured across sites that becomes someone else's asset. We aren't antithetical to that, but we are an alternative from the perspective of a user. Rather than that profile being represented on server, it is represented on an asset that I control access to. I think there is a sensibility there that will become important to people as concerns over the more traditional form of behavioral targeting become more prevalent.

advertisement

advertisement

>
Next story loading loading..