Yelp meets Google PageRank, has baby: Grayboxx
Grayboxx is a local recommendations service that's been quietly humming along since 2005. This morning they added 100 cities to the network, bringing the grand total up to 175. Grayboxx takes aggregate customer reviews from all over, and combines them by neighborhood to serve up business recommendations, kind of like what Google has done with its search results. Grayboxx will scour the internet for references to a business (be it tagged photos, or mentions in a blog post), and give that business a certain rank based on its pervasion. However unlike Yelp and Yahoo Local, which are designed and organized to feed off user reviews, Grayboxx's algorithm is completely automatic.
What makes the service particularly interesting is that it's largely unavailable in major U.S. cities right now. For instance, New York, San Francisco, Portland, and Seattle won't be getting the Grayboxx treatment until December, while many smaller towns are fully listed. Grayboxx's CEO has previously mentioned that the reason for this was to avoid head-to-head competition with other services like Yelp, while building up their technologies in smaller markets.
So what kind of stuff do you find doing a search on Grayboxx? For the most part, results are similar to what you'd see on other local search sites. There are addresses, hours of operation, phone numbers and any related Web sites. You also get neighborhood recommendations on the side of every listing, which will tell you if the service has a buzz. What was sharply missing in my testing though, were user reviews of any sort. Grayboxx claims to pull in reviews from third-party sites (like the Yelps and Yahoo Locals of the world), although I couldn't find a single one in my two test cities. While there's space for them on each listing, you can't add your own two cents about the service directly.
I find more often than not that user reviews can be the most helpful part of a business listing when it comes to looking for a recommendation. While services like Yelp and Yahoo Local offer mostly subjective reviews--and widely about food, it's the little things like which food dishes are the best, or important information like times to avoid a place when it's too busy or too quiet. While I don't doubt the interesting new direction Grayboxx is moving towards, I think the user-generated quotient is critical and will remain king.
Don't let the Digg-like counters fool you, those numbers are automatically generated by Grayboxx, the local listing recommendation service.
(Credit: CNET Networks)
Josh Lowensohn writes for Webware.com, CNET's blog about Web applications and services. E-mail Josh, or follow him on Twitter at http://www.twitter.com/Josh. 
An alternative approach is what is being done at MyFriendSuggests.com , they are using collaborative filtering (like what Amazon and Netflix due to recommend books and movies) to recommend places based on your personal tastes and the tastes of people like you. The site is still in its infancy and has very little user base making it less practical right now but the approach of using your actual tastes to find new places is something I think is worth keeping an eye on.
For me, grayboxx is a great, natural compliment to user reviews, because it picks up the slack in situations where user-review websites, like Yelp and Citysearch, become unreliable.
There's a basic point here that most folks aren't quite getting about grayboxx; it provides reliable info on subjects where all the other local engines provide no recommendations at all, because of their complete reliance on user reviews. Here's what I mean:
1) There are over 7,000 different types of businesses in the U.S. User review websites give feedback on just over 200 categories. Grayboxx gives feedback on over 6,000 types of businesses.
2) Almost all user reviews are written for the largest 15 or so cities in the country, but over 70% of the internet-savvy population of the U.S. live in smaller cities and towns. This means that user review-based web searching doesn?t give reliable results for most of us. Grayboxx works equally well in small cities and towns as it does in large cities.
Fapap brings up a good question about user reviews: Are they really statistically significant? We?ve been told for awhile now that user reviews are important, but honestly, isn?t a review just one person?s opinion?
On the other hand, even if several hundred reviews have been written about a business, do you have the time to read through enough of them before you feel that you can trust the overall result?
User reviews are important, but really they're just the icing on the cake.
Hey, don?t get me wrong. I *like* my cake with icing! But icing without the cake . . . the sugar high isn?t worth it.
Ben English
grayboxx