Interview with Rich Skrenta, CEO of Topix.net

Every hour our crawlers read all the news published online. We crawl over 6,000 different sources and categorize stories geographically as well as by subject. Our system has a news rollup for the 30,000 towns and cities across the United States. There are also over 150,000 subject categories. There are pages for everything from mobile home manufacturing to diabetes to every sports team, etc. Our goal is to have a news page updated about every person place and thing in the world from the broadest variety of sources.

We are trying to find micro targeted content that is being underserved by traditional outlets of news. Some early adopters of RSS readers (which provide news feeds to subscribers, see http://blogs.law.harvard.edu/tech/rss for more information) are finding that once they have about 200 news feeds they are unable to deal with large volume of information.

10% of informaiton available from a given news source is relevant to a specific user. They have to read the entire 100% in order to find the information that pertains to them. Given the massive amount of news that is available online, automation is needed to make sense of it and deliver what they want.

At a local level, I live in the San Francisco Bay Area about midway between the San Francisco Chronicle and the San Hose Mercury news. On any given day, these papers may publish something that pertains to my small little town on the peninsula, so I would have to read each of those papers to see if there was something happening in my town. In addition to that there are quite a number of smaller papers that could also reference my small town as well.

There are also websites for local police departments that have crime alerts, health departments at the city, county and state levels that might be posting advisories about restaurants, etc. To put together a full news picture of everything that might reference everything happening my in my area would require me ot read about 20 or so different sources.

Beyond that, there will be stories written outside of that circle of newspaper and resources about things that happen locally. We saw a story talking about a local town that appeared in the Juneau Alaskan Times. It was a legal opinion about a dispute between two neighbors over the height of a hedge. No one in that town would be reading an Alaskan paper for information about their town. Topix.net caught it and put it on that local towns page.

I haven’t seen them deploy any localization features. Their categorization is standard news with health, business, etc. They have the basic eight categories, whereas we have over 150,000 categories on the subject side. We are tracking to every public company, every new drug, every sports personality, and so on.

The entire system runs off of a massive database, which knows the name of every street in the country, everything with a geographical presence from bridges, lakes, rivers, park, etc. We are tracking the names of just about everything and categorizing them geographically. When a story comes out it tracks references to different locations and associates longitude and latitude coordinates that then provide the local level targeting.

The traffic on our site indicates that the split between local pages and general news is about 50:50. When you aggregate 100% of the news the whole is greater than sum of the parts. We can provide the context around a story.

If a local pizza parlor is looking to advertise online, it is difficult to do so with keywords. If they are located in a place with an ambiguous name then it is hard to identify those individuals that live locally. The second problem is that consumers don’t go online to find a local pizza joint. We think that associating advertising with what people read online with dynamic content is the best way to reach people on the local level.

The first thing we did was put up Google Adsense. Google spiders our site and then match advertisers with our content. That works pretty well. We have local advertisers on about 80% of our pages by virtue of Google. Listing ads from their client list of over 150,000 is the first step. We will also be selling ads directly to local businesses through a turn-key system that we will deploy on our site in the next couple of weeks.

I saw a number from Barry Diller’s Interactive Corp. that said the local advertising market was $90 billion. Google has a lot of advertisers, but there are between 6 and 10 million local businesses in the country. Clearly, only a fraction of businesses are advertising online.

Word of mouth has been pretty strong for our site. Back in 1998 our team created the search engine NewHoo, which we took from zero to millions of hits per day and was eventually acquired by Netscape, and is currently a property of AOL. We are seeing Topix.net grow at a much steeper rate that NewHoo ever did. In part, that is due that there are more net surfers than there were in 1998. Also, we think our dynamic content is resonating with users.

We provide concept level relevance versus keyword level relevance. Our technology attempts to determine what news stories are actually talking about. We are trying to create a very relevant concept driven experience on our news pages that exceeds other solutions.

It depends upon the role. We are in a startup environment and focus on the long haul. We aren’t looking to flip the company, or a bottle rocket type company that shoots up into the air and explodes.