• by Steve Smith , Staff Writer @popeyesm, October 12, 2007

Marketers continue to seek ways of scraping and mining the most common interactive behavior of them all -- talk. Discussions about products should be some of the most valuable activity for both marketers and consumers, but conversation is difficult to surface in an organized way. RelevantMind tries to tap the collaborative nature of purchase decisions by connecting products and services with the online conversations occurring around them. CEO Aaron Mann tells us how the newly launched site plans to reign in conversations both for consumers and brands.

Behavioral Insider: How are you harvesting opinions and conversations from around the Web into coherent content that surrounds a product listing?

Aaron Mann: We organize into verticals around specific interests. For a hobby like photography there are online user communities, forums and things like that. But a challenge is that you might have a question about a camera body or lens and you know that the question has been answered in dozens of places but finding it in the forums is difficult. We first curate the collection of sources, so we know that photography has maybe a dozen user communities that are passionately talking about this. We don't ask the entire Web, but curate specific ones that give you good information about photography.

Second, we have a subject matter expert that is responsible for each interest area and would know the talk about photography and its shorthand. We build that into the keyword index.

It is not just about product model names but what is in conversations. For example, if you were to bring up a particular camera, we give you subset of all the conversations that are discussing that exact camera.... If you are doing black and white you would refine the conversation to where people are talking only about those attributes.

Behavioral Insider: The results are product-centric? You need to work from a specific model? I can't just ask about good photo printers for black and white?

Mann: The results are product pages, and you have a series of links out to the actual source documents. We've got plans to do more of a buying engine [that can search for products by attributes] but our first release is the ability to bring all the conversations about that product into one place. The typical search on our site involves a model you think you want and you can put in competitive models. Compare and contrast is a really difficult search to do. You call up product A and then product B to see when both are being talked about.

Behavioral Insider: You say a lot of this is being grounded in an understanding of the behaviors people have demonstrated in shopping online. Could you give us some background on what sort of research you did?

Mann: We started by looking at the online user communities and what were the typical behaviors there. What kinds of things were being posted in terms of questions. You don't have to spend a lot of time in something like a photography forum to understand that a lot of what is being discussed is technique, and a lot is product-related. The problem was if you don't know to go to a forum or to this particular wiki or blog, finding that information is really hard. Most people start typing in search terms but don't get off the first page. So when we looked for discussions about what to buy, they weren't coming up until the fourth or fifth results page.

Behavioral Insider: Were product research behaviors the same across product segments?

Mann: That gets to the heart of the types of product that lend themselves well to what we do -- high involvement purchases. I am making a value decision to buy this and not that. Our two launch categories are road bikes and golf, because I do a lot of biking and my partner is a golfer, so we could see it was working. We will be opening others, from cooking to musical instruments. We've got a list of about forty vertical categories. But what is common across those are a lot of high-involvement searches. People make a value decision -- if I buy this printer, I made a decision not to buy these three other printers.

Behavioral Insider: This sounds work-intensive. What staffing is required just to do two categories?

Mann: Each category requires a 'vertical guru,' the gatekeeper on what are the right search sources. The bigger job is creating keyword combinations for each product.

Behavioral Insider: What is the revenue model?

Mann: There is a tab that says Buy Now which features that product from multiple retailers that we will have affiliate deals with. It also shows current eBay auctions.

Behavioral Insider: So you are using the same revenue model as a comparison engine, but with a different content model -- what to buy rather than where to buy.

Mann: Instead of focusing on price we're focusing on information.

Behavioral Insider: How will you promote it?

Mann: Our marketing is aimed at each specific vertical, so we're not trying to drive traffic to RelevantMind.com but within road bikes and climbing. It is to advertising in the user communities and search engine marketing. There is a lot of depth in our product page so it optimizes pretty well for organic search.

Behavioral Insider: Any advertising on the product pages?

Mann: Not planning to right now. There is a reason for us to do content sponsorship at the vertical level. We are being very cautious. It is important to keep the user at the center of this process. We are talking to people who want to use this content in interesting ways, especially retailers. If you think about it, we read a number of reviews at a retail site to get into the conversation. If retailers can bring that conversation into their site, it can be a powerful information source.

Behavioral Insider: How about mining the market intelligence?

Mann: We know what people are saying and how many conversations out of the total pool are mentioning a particular product, and how many times they are mentioning it in conjunction with a competitive product. We have already found some interesting things where retailers thought their competitor was product A or B, and it turned out to be C or D. Or the attributes they thought were important weren't, because people were talking about the more non-obvious things. Users find very creative ways to use products.