Property Site tells users where to find Property!

Your Property woes are gone. Finally, an initiative, maybe even a world first from the owners of Gravity.co.za, South Africa's first property search engine.

An article from the owners of Gravity.co.za...




"For quite a while we've wanted to provide our users with an area-type search feature where they could see, at a glance, all the houses in a particular area/within a given radius.

Most real estate property sites overseas would probably implement it as a map, but since South African agents don't provide street addresses for their listings and reliable street/area information biased towards real estate (e.g. that include estates, brand new road and suburbs that are continually appearing) is not really available, map searches are not an option.

There just seemed no way we could get something useful out until we suddenly realised that we had hundreds of thousands of users who had been telling us exactly what suburbs they thought were relevant to them. South African users are patient and diligent when looking for property due to the lack of user-friendly search capabilities like Google Mashups etc. They had given us really great data to analyse for trends. We ran a few basic algorithms to answer the question "if a user is interested in property in suburb X, what other suburbs will he/she most likely search" and the answers/correlations we got were just incredible. We optimised the algorithm and extended it to "if a user is interested in property in suburb X, Y and Z what other suburbs will he/she most likely search" which was launched on 1 August, try it out for yourself:

go to the Gravity Property Index homepage - www.g.co.za and enter the suburbs you're interested (comma separated) in in the Quick Search Text box. Select your price and hit enter - you'll see at the top of the results page that is returned, a list of recommended suburbs.

The powerful thing about this feature is that it's changed the search results from 'nearby' to 'most relevant'. GPS/map-based solutions have 2 shortcomings which 'most relevant' solves:

1. "Other side of the road". Often more expensive ('X') areas border on lower cost areas ('Y') and on map-based/radius searches both will be included in search results. It was quite interesting to us that although people looking for houses in X would never specify Y in their search (as you'd expect), the converse was not - people searching in Y would search a house in X. Intuitive I guess but satisfying to see 2 completely different result sets. If you're familiar with Cape Town, try a search for Rondebosch (X) and compare it Athlone/Rondebosch East(Y) - they border one another and even sound the same, but the relevant suburbs are not at all similar

2. "Similar Suburbs across town". It seemed buyers were interested not only in the location of a suburb, but its overall profile e.g. property values, great views / location next to the sea, near schools A OR B (across town from one another) etc. Area searches don't correlate 'LIKE' suburbs, just ones close together. Just as important we found was that, in a region with say 10 suburbs, often 3 or 4 would be LIKE, so the user would not have to search all 10 to get the houses in the most relevant 3 or 4.

And this was only possible thanks to the information contained in the minds of our visitors - keep searching!