One of the very interesting things going on in search is the phenomenon of “predictive search” or personalized recommendations based on a range of user data inputs (i.e., behavior, context, etc). Effectively, these are search results produced without entering a query into a search box.
Much of what is happening in mobile and Google Now is perhaps the best-known example. But, there’s also the related concept of the personal virtual assistant, represented by Apple’s Siri. Expect Labs and Grokr are both doing some version of these things, as are several others.
Trulia Suggests takes a range of data from users’ interactions with the site: their search criteria, home “Likes,” “Hides,” alerts and other inputs and offers personalized home recommendations that the algorithm believes you will like.
It’s not intended to be a replacement for search on Trulia. Instead, it’s a complementary function designed to surface houses that you might not find through explicit search (or browse) activity.
When users sign in, Trulia presents a grid of house images. You’re asked to select five houses (presumably that you would be capable of buying — I was drawn to houses I couldn’t afford) to “tune” the system, indicating your tastes. Trulia also prompts users throughout the site to Hide or Like homes providing further inputs, which are then combined with the other information to generate home “suggestions.”
With all the data Trulia possesses, this functionality was a kind of natural development. However, the purpose of Trulia Suggests is three-fold: to enable users to discover more new properties and get outside their fixed and potentially rigid search criteria, to drive more engagement on Trulia, and of course, to differentiate from competitors such as Zillow.
Currently, Trulia Suggests exists only on the PC site; but, it will be integrated into the company’s mobile apps in the near future. It will also extend to rental properties eventually.
It’s a nice feature and certainly useful to home buyers. However, it’s most interesting to me as part of the larger trend to mash up lots of data and offer personalized search-recommendations and results without requiring users to explicitly query the database.