• by Steve Smith , Staff Writer @popeyesm, January 4, 2008

As traffic fragments around the Web, finding a large qualified audience to target becomes an everyday challenge for media buyers. The usual suspects in a given segment, whether it is entertainment, auto or fashion, become the natural locus for ad dollars, inventory squeezes and CPM  inflation. Behavior Match is an intriguing new planning tool from Compete Inc. that segments users into 150 categories and then tracks their usage and concentrations across thousands of large and small sites. As CMO Stephen DiMarco tells us this week, the approach helps media buyers discover hidden gems in the ever-expanding mediaverse fragmented audiences

Behavioral Insider:  Where and how does Compete Inc harvest the clickstream data it uses in its media planning tools?

Stephen DiMarco: We built our own proprietary panels from software people can download and opt into sharing their clicks with us. We also buy data from companies that can resell it to us like ISPs or desktop application developers. We make sure we have permission to look at what people are doing online. We have had a panel of two million folks for about seven years now.

BI: You have been mapping Web traffic for a while, so how does this new product differ from your existing planning tools?

DiMarco: For while what we really focused on was understanding what sites were driving the most traffic, so it's [been] kind of a volume-focused product and measurement. The audience measurement space has focused on audience quantity instead of audience quality. Every time we did the analysis we saw that the top sites were always the top sites: Yahoo, Google, MSN  -- and then, depending on the industry, it might be the larger automotive information sites or the lead generators in the wireless base or Bankrate in financial services. It must be really boring to develop a media plan for the same sites where you are fighting for the same inventory against your competitors all day long

BI: Yet all the statistics show audience diffusion.

DiMarco: Exactly. So we looked at other tools in the market. The media planning tools focused on the largest sites. By the time you did all of your audience selects by demographic or by behavioral group, you are only left with a handful of sites to choose from. We had more data, so we see consumers going to smaller sites without the sample breaking down. We could segment that data and still have a meaningful sample to return to advertisers a bigger universe of sites. They could expand their media buys away from the head and start to move it into the torso. That is what BehaviorMatch does.

BI: Explain the segmentation and how it works.  

DiMarco: We have 151 predefined segments in shopper or lifestyle categories. Shoppers are defined as those in market for specific products or types like SUVs or luxury cars, mortgages or credit cards. Lifestyle segments include early adopters, heavy e-commerce consumers, Millennials or online socialites, Internet addicts.

We separate those consumers out and basically juxtapose the Web sites that they go to most frequently to the average Internet consumers and have this algorithm where we can index individual Web sites based on their composition of that specific customer segment of interest. Sites that have an index over a hundred are more likely to attract that type of consumer to their site. The larger the site the more likely the score will be 100 and takes on the general complexion of the Internet. Yahoo.com tends to score 100 for every single customer segment, but we [also] score the sub domains of Yahoo, which vary greatly.

We started scoring individual blogging subdomains of larger platforms like Typepad or Blogger to see which blogs attract specific customer segments. It's a way for companies to chart their social media strategies and figure out who are the most influential bloggers in their market. That is really hard to do by hand. We thought the Web was hard to navigate. The blogs are even harder.

BI: Are you seeing enough concentrated activity at some of these emerging blog and social media sites to merit advertiser interest?

DiMarco: I am amazed at is the volume of hidden jewels out there. Critical mass used to be defined as a million uniques a month. I think you can redefine it as 50,000 or 100,000. If you are a publisher with that many people at your site you are doing something right.

When we run reports for the segments our clients are looking at, the sites that index the highest I never heard of before. But you look at the site and it is just a perfect fit for what [the buyer] is trying to accomplish. We analyzed sites that attract business decision makers, and the logical places are large properties like Forbes and MarketWatch.

Well, there is a slew of other sites that are investor-related, but not necessarily day-trader-related. They are more about private equity, earnings reports and things like that that have a really good base of users that you never would be able to find. They could drown out in search results and get dwarfed by the larger portal properties -- but when you look at the data this way, the jewels reveal themselves.

BI: Any back-end metrics that demonstrate the strengths of this kind of planning based on behavior matching over typical contextual buys?

DiMarco: We tested with one of the portals where we ran head to head against some contextual advertising that they had around the category of travel shoppers. And then they used our product to run a campaign side by side with the site list that our tool created. Our tool scored four times more effectively, four times higher click-through rates than their own internal contextual network. There are some counterintuitive connections in the data that the human mind can't comprehend. So Behavior Match cuts through that.


BI: With the speed with which blogs become hot and not online, you must see some trends ebb and flow in compressed time frames.

DiMarco:  The political space changes a lot. It is altogether new consumer behavior that creeps up and goes away every four years. A lot of stuff pops up that didn't even exist last year, whether it is blogs or political commentary sites.

Anything that has to do with a consumer-generated product reviews  absolutely are so viral that they can creep up and go away very quickly depending on the product they are focused on. There was a lot of activity when the iPhone launched and some sites indexed high there. But by and large the large sites are stable. The sites that are below 50,000 uniques really are volatile and hard to manage, but there are sites over 50,000 and 100,000 and on their way to being million visitor sites, and those are the ones you want to use the tool to find.

advertisement

advertisement