Saturday, May 30, 2009

Online marketing from start to finish in 3 hours

Posted by Danny Tarlow
As a favor, I agreed to help some people I know market a rental home in my home town of Lake Oswego, Oregon. It's a very nice, new house, but they don't feel like they're standing out amongst the crowd. Try doing a little search for "lake oswego rental house" on Google:

When I first did this, I was pretty overwhelmed by the number of all-purpose rental sites that showed up. There was no way some puny little listing like our target one would ever see the light of day as is:

There are just so many problems with the listing. No relevant terms appear in the URL, and I don't even see the term "rental" on the listing page. Talk about not caring about search engine traffic!

So instead, I decided to build a new site, making use of the good parts of the listing (great photos and video). In addition, I bought a relevant domain name, installed Google Analytics, put care into the page titles, and chose URLs to match the most important keywords. It took about 2 hours to adapt an old website and css to fit this purpose, then I scrounged up at least a bit of content to fill in the blanks. You can see the result:
the Lake Oswego Rental Home.

Total time so far: 2.5 hours.

Now we already saw how crowded the organic search engine results were, even for this small market, so I decided to throw a little money at the problem. I set up an Adwords campaign targeting keywords related to "rental homes in lake oswego". I came up with four different pitches:
1. Brand new construction
2. Beautiful views of Mt Hood
3. Excellent schools and pet friendly
4. Adjacent to a 40 acre park

Then I wrote ad copy that combined pairs of pitches.

Total time so far: 3 hours

So about 3 hours after I bought the domain name, we were in business. I set some bids, and let the ads loose. It's been running for about twenty minutes now (the time it took to write this post), and we've already even gotten a couple clicks -- how is that for instant gratification?

Of course, it would be great if we could even get a little organic (non-paid) traffic to the site. I'm not sure how long it will take to kick in or how competitive this market is, but I'm working on building a few links and doing some other organic-conscious marketing. As for right now, I can tell you that we are not listed in any search engine results =P. It will be interesting to see how it progresses.

Friday, May 22, 2009

Energy data

Posted by Danny Tarlow is highlighting a data set on residential energy consumption. This is appears to be the same one from the Energy Information Administration (EIA) in the Department of Energy: We made use of the similar EIA commercial building data set when I was working on modeling energy use in Disney parks, but I hadn't seen this one before. The level of detail of the survey is very impressive, covering 12 sections:
  1. Housing Unit Characteristics
  2. Kitchen Appliances
  3. Other Appliances
  4. Space Heating
  5. Water Heating, A/C, and Miscellaneous
  6. Fuels Used and Fuel Payment
  7. Fuel Bills and Non-Residential Uses
  8. Household Characteristics
  9. Energy Assistance and Housing Unit Square Footage
  10. Characteristics of Energy Supplier Data
  11. Energy Consumption
  12. Energy Expenditures
I suspect some progress could be made answering the motivating questions for e.g. the Google PowerMeter project using this data:
How much does it cost to leave your TV on all day? What about turning your air conditioning 1 degree cooler? Which uses more power every month — your dishwasher or your washing machine? Is your household more or less energy efficient than similar homes in your neighborhood?
I'll report back if I get some time and discover anything interesting.

Saturday, May 16, 2009

In Google we trust

Posted by Danny Tarlow
I was at the post office the other day, and the guy in front of me in line was having a heated discussion with the lady behind the counter. Apparently the guy was trying to overnight an envelope with original copies of his passport and birth certificate to some "agency" (I couldn't gather what kind) in Southern California. The lady behind the counter had never heard of the agency and was trying to convince him not to send the package. The guy was very confident that he could trust the agency. His reason: the agency came up very high in Google's search results.

Sunday, May 3, 2009

Kentucky derby

Posted by Danny Tarlow
I watched the Kentucky Derby yesterday, which was quite the show. Not only was the winning horse -- Mine That Bird -- the biggest underdog to win since 1913, the way that the winning jockey (and horse) snuck between the other racers and hugged the rail to get through was pretty amazing. Steven Levitt (the Freakonomics guy) has a model for predicting horse race outcomes, and he's been kind enough to publish his analysis of the Kentucky Derby the last couple years. It's incredibly unfair to make fun of him when we know the outcomes, but I couldn't help but chuckle reading his commentary on a couple recent races:
If I had to pick a last-place finisher (a bet they would never actually offer at the track because people involved with horse racing understand better than most that people respond to incentives), it would be Mine That Bird.
Then from 2006, where Street Sense won:
The two likely favorites are Street Sense and Curlin (both about 4-1). I wouldn’t touch them.
So I think somebody needs to ask: Steven Levitt -- are you sure you don't accidentally have an errant minus sign somewhere in your model?

Friday, May 1, 2009

Horse racing 101

Posted by Danny Tarlow
The San Francisco weather has been pretty nice the last couple weekends, so some friends and I decided to go check out the horse races at Golden Gate Fields over in Berkeley. We were there more to watch and have fun than to bet, but I couldn't help but think about modeling the races. There was certainly no shortage of strategies that people swore by: look for a shiny coat and perked up ears; pick horses with lighter coats in warmer weather; pick the favorite in the short races; and plenty others.

Less anecdotally, there seems to be a bit of agreement on the factors that go into a horse's performance:

I wonder how much of this is relevant if you have some data about the horse's past performance. In the Netflix challenge, it is my informal understanding that information from IMDB about actors, genres, directors, etc. isn't terribly useful in improving recommendations because most of a user's preferences towards these rough categories are already captured in their ratings profile. The argument could probably be made that horses race less than people on Netflix rate movies, so there is less information in the data, but it's hard to say if this would make a difference in model performance without doing some analysis on real data.

I haven't had a chance to go through all of these links, but there does appear to be a lot of data out there. The problem is that none of the sources seem to be centralized, free, and in an easily accessible format:

I think the best bet would probably be to focus on an individual track to start. My first choice would be to find some historical data from Golden Gate Fields, but it looks like other tracks have better data available. For example, the Santa Anita data looks decent:

With a bit of help from the Mechanical Turk, maybe it would be possible to put together a reasonable data set.

It might be fun to play around with this a bit more. It wouldn't be hard to build a model similar to my March Madness predictions, though who knows how well it would work in practice. Regardless, doing something more rigorous is probably better than blindly betting on the horse that is named after my grandfather.