Talking about Zillow’s neighborhood aggregation got me thinking…

…about how I tried to solve the problem at back when I was on the product team at Move (May ’06 to Dec ’06).

Back then, I took over a mapping initiative because I wanted to create an interface to that could aggregate and display the vast amount of data within the grasp of,,, etc. within a simple mapping overlay.  My first (and only) pass at the project involved working with the one engineer/architect at my disposal (a darn talented guy!) to develop a widget that was eventually implemented on each neighborhood detail page of  (To see an example, scroll to the bottom of the Calabasas page).

LA Heat Map

The options boxes on this implementation are pretty limited in that they only include a few datasets (i.e. average home and rental price, number of listings and rentals and medium household income) at a few listing levels (i.e neighborhood, zip code, and school district), but you can imagine that if the data is batched in at a fine enough level (census blocks for example), then it is somewhat trivial to add any new dataset and aggregate the data up to any level (neighborhood/zipcode/city/school district/county/state) using this basic design.

The discussion about Zillow’s neighborhood aggregation got me thinking about this because I can tell that a ton of work went into developing their overlays, and yet they are just barely touching the surface of the data they could be providing… It seems to me they’ve over-emphasized the importance of their zindex-type data when, my educated guess, is that relevant consumers (i.e. potential buyers) are interested in so much more.

I’m not going to say that the overlays on the neighborhood project are better implemented (they’re not), but I think the concept I put together was closer to providing information that would be relevant to home buyers, which, after all, is the group most likely to be looking for neighborhood information.  In writing this, I feel a bit sad that it didn’t make sense for me to continue down the path of improving the overlay widget because until someone provides a much more slick way for consumers to filter and prioritize neighborhoods based on information that would matter most to them, I’m convinced some major innovation will come into this space.

12 responses

  1. What I was surprised about with Move’s neighborhood stuff was that the listing integration was not very strong. It seemed like the dev teams for the new stuff were very different and divorced from the dev teams for the listing side.

  2. Attempts like this are moving in the right direction –toward what buyer consumers want when they go house hunting on the web.

    Someday in the coming year, an outfit like Zillow,, or an as-of-now unknown (ESRI Breakaway?) will assemble a huge data warehouse running on an enormous geographic information systems server.

    GIS data layering is the best way for the next great search to come together — putting incredible visual data at the potential home buyer’s fingertips.

  3. Michael: Good call…

    I’ll keep my opinions on the internal team structure stuff to myself, but you make a very valid point that the listings were not better integrated into the project. Granted there is an option to show “home for sale” on the widget, but the bulk of the listing integration of neighborhood data happens around the Showcase product. (see the neighborhood info tab on any showcase listing for an example.)

  4. Joe:

    Because I ran GIS databases for years (as a transportation engineer) before jumping into online real estate, I also have a GIS bias… And as a matter of fact, the Neighborhood widget I mentioned above has a GIS backend. Getting a new dataset to show up on the map is as simple as importing the data into the backend and maping it to a new option to the option box. The overlay tiles are actually generated on the fly, although the tiles are cached so areas that get hit often can display quickly.

    The weakness of this approach is that you loose interactivity with the map once the maps are rendered. That’s why I think the work that Zillow put in to get each area to get highlighted on mouseover is really quite impressive. The drawback to that approach is that it appears to be quite front-end heavy and my guess is that it will be a lot more work to get useful heat maps out of that system (especially heat maps that render well cross-browsers).

  5. Hello Dustin,

    Although I love these maps and nice tools (they are all great link bait for a tech point of view), I sometimes question their real usage by actual buyers. Obiviously for relo it’s a great tool.

    If memory serves me correctly from the NAR Homebuyers report, the normal buyer purchases a home witin 8-11 miles from were they live currently. If that is the case, then how much usage do these tools really see, as I suspect most are famaliar with the areas they are interested in.

    My personal experiance, is that most seem to be interested in finding out exactly what school their children will be going to more than anything else.


  6. […] been reading about Zillow’s new smart search, and the discussions that are springing up about map based real estate navigation, and think, what Zillow really needs is a Google My Maps type […]

  7. Jessie: I’d agree that this type of neighborhood data will only be useful to a small subset of users. I happen to think it wouldn’t be too difficult to make graphical data relevant to people moving even only a short distance, but that is another project for another day.

    Overall, I agree completely that neighborhood map data is only valuable to small set of people who are looking to relocate.

  8. I had totally forgotten that there is a public view into more data using the same overlay widget (but on a smaller map) to give an idea of what I mean by adding more data options. If you look at this page on the Calabasas neighborhood page, you can see an area where the team opened up even more GIS-based data fields.

  9. I agree with Michael; the genius in Zillow’s implementation of neighborhood search is that it’s well integrated with listings. I do however really like having access to a map view of the neighborhood on the neighborhood overview page.

    It’s our hope that neighborhoods on the map drive traffic to neighborhood detail pages like homes on the map are an entry point to home detail pages. Time will tell if the Zindex is sufficient to attract that interest — but I’d love to hear suggestions for more effective hooks to Zillow’s local pages.

  10. I haven’t taken the time to comment on any of this, but I’m at a loss to understand why Zillow invested even more money on really amazingly badly defined “neighborhoods.”

  11. David: You’re awesome… Despite your implication, I don’t think Michael would agree with you that Zillow’s implementation was genius. Clearly Zillow is not going to have a great neighborhood experience if your existing listing content is the “genius” that makes it work. Zillow’s listing content is clearly not nearly comprehensive enough.

    Thinking of neighborhood definitions… One of the other mapping projects I spearheaded at Move was to use the neighborhood definitions given by agents on their listings (many MLS provide this as a field) to develop a unique set of “agent-defined” neighborhood definitions that was inspired by the work done by the neighborhood project. Besides, I think only craigstlist is sitting on enough user-defined points to put together a truly unique nationwide set of neighborhoods.

  12. […] as this one for Calabasas. In a timely coincidence, there was a good discussion earlier today on the post about neighborhood mapping , where Jessie B pointed out that maps are really valuable for only a small subset of people moving […]

Leave a Reply