Matt Colebourne: Phase one is a simple tool or service for users which allows them to track every conversation that they have [online,] whether it be in a blog or commenting on a piece of content or in a forum or a thread. Highly trafficked sites get thousands of comments and you get lost. From a single location you can track every single conversation. If anyone responds, you get alerted and see what they have said. Then what you really want to do is follow what well-known individuals online are saying across a range of sites. So we started building that functionality. It is anonymized; people don’t have to give their real name and ID. To track a specific individual you would have to know what they are calling themselves.

BI: This is a free service. What is the business model?

Colebourne: It is advertising-funded, based on display advertising. We aim to be the enabling layer between people and finding a good conversation. With the volume of conversations, it doesn’t need that many people to go through to become highly profitable. It also allows us to research attitudes toward products, services and the like. Again, on an anonymized basis.

BI: Where do behavioral models fit into this?

Colebourne: That folds into where we are going. What I have just explained is version one. But what that kind of thing doesn’t solve is the issue of finding a good conversation. If you want to discuss politics, religion, cars, then people who actually know something about it are far more interesting than people who know nothing about it. We are looking at how we can help a community find its own experts.

Version two of the product becomes more about individuals not just keeping track of, but making sense of, the world in terms of how they want to interact with conversations. So they can look at a conversation stream of tens of thousands of comments and say, I only want to see those answered by people that I know. But then it is useful if you can start to pick up on the individuals who have natural authority on the topic. So we are building a ranking system or a behavioral system where people rate other people. But after a while, once the experts start appearing, they in turn should be able to bring other people up quickly. So an expert on medieval history can see I actually do have something to say on it and give me a positive ranking. That will give me a much higher rank than someone else who knows nothing about it. So it is like a peer review and commenter ranking system, but against the taxonomy of topics that allows the natural experts to appear.

BI: So people are judging one another’s behavior in the threads. Will behaviors like frequency and recency of posts identify expertise?

Colebourne: We are working on other behaviors as well, but we want to be careful about doing too much behavioral tracking in a very blogger-heavy audience. We have to play by the rules of the market, so we’re shying away from doing too much that wouldn’t be voluntary. We are looking at some very interesting technology that allows us to do a mind map representation of people’s views on particular topics and incorporate that into ranking. How about if we rank on like-mindedness? You can say I want to see what people who normally think like me say about a topic and how people who think opposite of me think.

BI: But you are also amassing an index of conversations, which are a kind of behavior.

Colebourne: Obviously we are building a considerable behavioral database. Comments are being tracked. What we end up with is how often people are commenting, where they are commenting and even what they are commenting on to a defined taxonomy. You do get a buzz index. We are trying to put together behavioral — i.e. recency, frequency, and a value measure from the community — and put all that together with potentially psychographic information.

BI: What value does it offer marketers?

Colebourne: The behavioral and demographic information gets interesting in the research space, where it really adds more value we think. If we are tracking 20% of the world’s conversations, this is a huge amount of information, and you don’t need to be in any way invading privacy because you can do it anonymously. People can understand how groups of people think. Anyone asking what people think about my company or product would probably want to find out what the blogosphere thinks, but they wouldn’t base their entire strategy on that group because it is a small percentage. They want to look at different demographic groups, activity groups, and behavioral groups in terms of how they interact and converse. Otherwise you could get a very misleading impression.

If I am Coke and I want to know I am being talked about, that is an easy measure in a database. But how do I find out if I am being talked about positively? That is exactly where we can offer something that at the moment we don’t believe anyone else can. You need to have all of those stored conversations.

There is plenty of tech that will help you pull out what is essentially a semantic query. The issue is, do you have the data to run them against? In storing a copy of every conversation we then provide a searchable database so we can say what is appearing.

Companies already are interacting and having conversations. But the problem is that is you can’t cover the entire commentosphere.

BI: Where would this knowledge make a difference?

Colebourne: You can imagine Coke in a given country wanting to understand how the branding is moving in relation to certain strategies. It is an old example, but look at Kryptonite locks and the posting of the article that showed you could break into [the locks]. That company did not realize how big an issue this had become, because they were only monitoring a small number of areas where it [had] only started appearing –when suddenly it exploded… and it cost them a lot of money. But they were unable to [discover the extent of the problem] because the initial conversations were separate, and then suddenly they reached critical mass.

y>