Dan Farber: David thanks for joining me.
David Beitel: Thanks for having me Dan.
Dan Farber: Zillow is one of those applications that could only exist on the internet, could you explain what Zillow does?
David Beitel: Sure. Zillow is a free consumer website that provides home property and transaction information on over 80 million homes in the US. On 67 million of thosehomes, we have what's called a "zestimate", or Zillow's estimate for what that home is worth. We also provide a 12 year trend of what we think your home is worth over that period of time. It's a really great way to come on and do a lot of research about home valuations
Dan Farber: How did the idea for Zillow come about?
David Beitel: Well we started around over 3 years ago when we were batting around a number of ideas. We were excited about the real estate space and there was really not much information available. When you go back just a couple of years there really wasn't a resource where you could go on and find information about your home. There might have been a few listing sites where you can go on and see what's on sale. But what we have built is really a database of all homes in the US and we think that's a much more powerful asset. We thought about "do we want to be your agent or real estate agent", and we decided that we wanted to take a different tact, we wanted to provide a lot of great information free to consumers and then make money by advertising around that information.
Dan Farber: Now your providing a huge amount of information and I know that you've just upgraded your database to include a lot more homes. How are you able to keep up on the backend with the rate of growth you're having?
David Beitel: It's a real challenge; we're processing terabytes and terabytes of data each night, we're processing millions of new property records, new transactions, new property attributes, we get that information from a variety of sources. As I said, there's over 3 thousand counties in the US and some of those have electronic records and some of them don't and so it's a real challenge to pull that information together into a common database and present that on the web in a real easy to use format.
Dan Farber: So what are you using on the backend to ensure that you have something that's reliable and allows you the flexibility you need to really grow and change and keep that database from getting stale.
David Beitel: We've created a processing pipeline that allows us to import data, new records, to get them cleaned in a common format. We store all that in a variety of databases, we have Microsoft SEQL server databases as well as MySQL databases that power our site. On top of that we've built a proprietary search engine in C++ running on Linux that allows us to serve up that information at lightning speeds.
Dan Farber: Now I go to a lot of sites and one of the issues around sites that aggregate lots of data like this is the accuracy of the data. You talked about cleaning the data I'm assuming that's just the process of normalizing it in terms of the different fields, but how do you make sure that the accuracy is something you can count on.
David Beitel: Well we run it through a number of different systems. We want to make sure the address is accurate so there's some services that we've been able to bring into our pipeline that allows us to make sure the address is clean and accurate. And the other thing we've done is really position Zillow as a starting point that people can come on to our site, claim their home and update information. So we have over 2 million people that have come on to our site, claimed their home and updated property information so that we can start with the information that we found, there's some information that doesn't make its way to the county so maybe an additional bedroom that was added through a remodeling project, and we have our users coming on and uploading photos, so we really feel like we built a real vibrant database that is available to our users.
Dan Farber: Sounds like you're building more than a database, it's almost a community around housing and I kind of liken it with all the people you talk about coming to it and adding their own data almost like a kind of fantasy league, check your process and check your neighbor's prices. Is that something that you're investing more in?
David Beitel: Absolutely, we are thrilled about our community. It's kind of the new trend as we're seeing in the social networks that are forming up all over the web and we're seeing it in the real estate vertical as well. We have homeowners and home enthusiasts who are coming on; we have real estate professionals who are really seeing how they can use a free service like Zillow to promote themselves by creating a profile and uploading their listings, by asking and answering information. Neighborhoods are really a vibrant kind of thing in real estate, where people are really concerned about the neighborhood they're living in, or maybe they're moving to a new city and they don't know much about the different neighborhoods and the best place to go is to go online and ask those real estate professionals and experts to tell them about their neighborhoods and answer some questions. So this is absolutely much more then just a database, but a really vibrant online commulity.
Dan Farber: Now as part of your role as CTO you also have a developer community and open APIs. Could you talk a little bit about how you're using those APIs to kind of expand the footprint of Zillow?
David Beitel: Sure, we early on decided to set out data free and it was a little daunting but we decided that the best thing for us to do to really get the "zestimates" and the Zillow brand out there was to open the data up through an API and we've been really excited about some or our partners that have used those APIs. We have Yahoo Real Estate, Zip Realty, Redfin just to name a few and they're all using "zestimates" on their website. We see that one thing that does is help our Zillow brand and secondarily it drives people back to our site. We don't provide all the information through the API so if people want to get a little more detailed and see more of a history of a home, they're going to come back to out site. Another thing we've shipped is what we call "neighborhood shape files", these are neighborhood boundary points that we've opened up to the open source community. It's our belief that by opening that up to the community, we're going to end up with the most accurate and comprehensive set of neighborhood information online.
Dan Farber: Now you're collecting a lot of data, and I assume that some of that data is going to business intelligence applications, dashboards. What are you doing on that end? I know you published some reports such as a report on housing starts, housing sales. Is that something where you have other designs in terms of creating new products?
David Beitel: Well we've built a real exciting data warehouse of information, some of that's just based on the information that we're able to process, kind of home facts. And certainly there's analytics that we can do on top of that as we learn more about what users are doing. And some of what we've done is we've built a real powerful ad targeting system for our advertisers so we can do very specific targeting so that the advertising that people are seeing on our site is really relevant to them. It's very often that someone who's coming on and looking to buy or sell a property will ultimately need a mortgage, so we can provide great targeted advertising for them through that data warehouse that we've built.
Dan Farber: What made you decide to build your own ad serving technology as opposed to going with someone else that can provide that technology? Is there some competitive advantage in that?
David Beitel: We believe so. When you're primary way of making money is through ad revenue, we want to make sure that we can do some real exciting things in that space. The targeting that we've done is real revolutionary, patent pending, and we're able to do a lot of neat analysis specifically on our users and what they're doing and how they're finding their information and then we infer really interesting specific properties about those users that put them into certain categories. That's something that we wouldn't really be able to do through some of the more generic ad serving technologies out there.
Dan Farber: Now you mentioned the word "revolutionary" in regards to the ad system, what specifically would you say is revolutionary and how did you develop it?
David Beitel: Well the targeting is really quite unique, we're processing all kinds of data. When we look at the kinds of page views that users are doing on the site we can infer some really interesting facts. For example we can tell when a home is likely to be put on the market or a particular user is likely to be a "mover". These "movers" are really attractive users for our advertisers as they look to get further up the pipeline as opposed to in many cases to day, when an advertiser is able to market to someone they've already moved, and a lot of the services that they potentially want to offer our customers it's maybe too late. So we've done a lot of neat statistical models that we've built using the data that we're processing to be able to do that kind of targeting.
Dan Farber: Now it sounds like you have a lot of innovation going on at Zillow. How do you maintain that pace and especially in this competitive market, keep people who are really talented?
David Beitel: Well it's a challenge. You know it starts with hiring great people and we've been very fortunate over the last few years to hire a great group of engineers from all over the US. Especially here in Seattle we have a real great kind of community of development. We've paired those engineers up with really smart business people and work collaboratively in these feature teams who own specific areas of the sites. And my job as CTO is to give them a playground to experiment, to take risks, and I tell them to take risks and don't be afraid to make mistakes. The great thing about the web is being able to measure the effectiveness of our features and if we get something not quite right, we'll iterate on it and get it right next week. It's really a powerful model and one that we've been excited to use at Zillow.
Dan Farber: I was thinking about Zillow in terms of the usage model and do you have this concept of a recommendation engine? You come to the site, you give them some data, and then it recommends housing, neighborhoods, anything like that for you.
David Beitel: We're working on some things like that. I think what we're trying to do is really make the site personal. Whether you're a homeowner, or a homebuyer, or just someone who's just interested in housing, whether you're a real estate professional, we want to make sure that your experience on the site is different depending on what you're doing or what you're trying to accomplish. So that's really kind of a big challenge and one that we're going to build personalization systems, recommendation engines as well as just making sure that we surface the right information to help you find what you're looking for.
Dan Farber: And finally, what are the technologies that you see on the horizon that could really change the way you do things or have a major impact on your business.
David Beitel: Well we're continuing to see the exciting things happening in social networks and in our communities. So I think you'll see a lot of advancements as we look to continue to provide our users with great ways to share information with each other, for homeowners to publish more information about their homes and their neighborhoods, for real estate professionals to continue to promote themselves in a free way, that's very exciting. I think you'll also see a lot of advancements in mobile. The neat thing about real estate is it's local, but a lot of it happens when you're out and about. So you're out looking at an open house or you're out as a mobile professional and we want to make sure we're providing you with useful tools that let you find the information you're looking for. We recently, a little while ago launched what is called "Zillow mobile", it's a way to text message us an address and we'll quickly return within seconds information about that home, our valuation of course, but also other really interesting facts as you're looking to understand more about that property.
Dan Farber: Well David thanks so much for speaking with me.
David Beitel: Thanks for having me Dan.
Dan Farber: I've been speaking with David Beitel who is the CTO of Zillow.com, for CIO Sessions I'm Dan Farber, thanks for watching.
