User ratings and reviews are among the hottest areas of local search (video and mobile being the others) in terms of media attention and funding. But there are still significant barriers to cross and challenges inherent in corralling a user base to participate in a review program.
One of those challenges, pointed out in UGC: To Be or Not to Be, is creating a destination that has consistent volumes of reviews across categories and locales. Generally, getting more than one review per listing is also challenging, yet important to get a true sense of a business — given false positives, false negatives and fraudulent reviews (think businesses writing about themselves).
Yesterday I had the chance to talk with Bob Chandra, CEO of Grayboxx, a start-up with a beta product that aims to solve this problem using algorithms instead of humans. The company is very secretive about what’s under the hood, but Chandra was able to explain in conceptual terms what it is trying to do.
The gist of it is that behavioral patterns can be recognized by tracking what users do both online and off (as we already know). But from that, overall levels of quality or popularity of certain products or businesses can be gleaned. For instance (using yesterday’s restaurant subject as an example), if a user checks out a restaurant online and makes a reservation (perhaps on OpenTable or a similar site) and then a week later comes back and makes another reservation for two, this can be a positive sign that he was happy with the experience (and possibly scouting date locations).
“Conversely, if I go to a hardware store and make a large purchase, that says something,” says Chandra. “If I go back the next day and make a large return, that says something else.”
Taking this a step further and examines more granular minutiae of online and offline activities on a much larger scale, Grayboxx is able to pick up patterns and devise a system of what it calls preference scoring (it seems this could also have future implications for contextual ad placement based on correlations detected between different products). Compared with traditional reviews, this can be more easily scaled across categories and locales.
“It’s an algorithmic and scalable way to rank businesses, not just restaurants and night life but every category,” says Chandra. “And the nice thing about that is that we have results just as well in Moscow, ID, and Bridgeport, CT, as we do in Chicago.”
This is key because over half the Internet population resides outside the top 30 metros, according to Chandra, which he believes are currently underserved with adequate local content and reviews. Grayboxx will also likely layer in traditional reviews on top of this content in the future, in addition to other forms of information.
Sharing the Love
In addition to a destination strategy, Grayboxx will also market the technology as a platform for other local search destinations such as IYPs, according to Chandra.
This could help IYPs compete for user appeal against sites that are building a great deal of user-generated content, including Citysearch and Yelp. This is something in which IYPs have shown a definite interest, evidenced by review programs launched in the past year at Yellowpages.com and Superpages, among others.
This could in fact give IYPs an edge, given their breadth of content, when compared with the more focused vertical strategies — weighted toward restaurants — of most online pure plays. Indeed Grayboxx is well suited to this breadth of content, which comes back to the challenge of building consistent volumes of reviews across categories that elicit varying levels of usergenerated activity.
“Places like Yelp are social networks for foodies,” says Chandra. “I’d love to see a social network for people interested in plumbers.”
We’ll get to see how this all looks later this month when Grayboxx initiates a staggered launch of localities, starting in small towns and cities where it believes it has a competitive edge over other local search destinations that focus more on larger metros. We’ll await the launch and be all over it when it happens.