Wood-Lewis: Local Grayboxx Results ‘Bad Advice’

Grayboxx ranks local businesses based on the number of mentions it receives on online consumer services. Almost a year after I first wrote about it, it has finally launched. The service, which assumes that people mostly mention services they like, has repositioned itself as the ratings service for small towns that couldn’t hope to get enough users to individually sign in and rate services.

To my eyes, it looks great. It is elegant, and intuitive, and it is really nice to see every category heavily populated with lots of listings – even in tiny Burlington,VT, population 38,000, which is its first market. Grayboxx’s system has 12,000 “neighbor recommendations” on more than 3,000 of Burlington’s businesses.

But do the mysterious algorithms really produce reasonable results? To find out, I was going to call Michael Wood-Lewis, the Burlington-based hyper local guy at Front Porch Forum who writes The Ghosts of Midnight blog. But then I thought I’d just read him.
Wood-Lewis says the results were pretty awful.

Wood-Lewis notes that he checked out a wide range of categories, including jewelers, taxis, computer repair, groceries, computer dealers, newspapers, shoes and pizza. “As a local, I haven’t had one search produce what I would call ‘good advice from a neighbor.’ Makes me wonder about the chef’s secret recipe (mysterious ranking criteria)… hard to trust when initial tests come up with what I’m seeing.

“Burlington, and all of Vermont, is so reasonably scaled, that lots of this kind of information is near at hand. Ask a few co-workers, friends, etc. Indeed, this is one of the most common types of messages posted among neighbors on Front Porch Forum. People get real ‘neighbor recommendations’ in Burlington everyday this way.

“At this point, I’d rather ask a couple hundred neighbors for a computer repair recommendation through FPF than turn to this kind of service. Or I’d sooner take a peek at the local successful reader survey that our weekly alternative paper runs, the much coveted Daisies, by Seven Days.”

8 thoughts on “Wood-Lewis: Local Grayboxx Results ‘Bad Advice’

  1. I agree with you, Peter, that Grayboxx “looks great. It is elegant, and intuitive, and it is really nice to see every category heavily populated with lots of listings.” However, my initial search results were not very encouraging. But I didn’t do exhaustive testing and my definition of “good” results include my own bias. I do think this service could add something.

    Cathy Resmer, Burlington-area journalist and blogger, found odd results too (see her comments).

    I should clarify, the taxi results were fine… but obvious. There’s one sizable cab company in town and then a small number of tiny ones, and that’s what the search results reflected… the big one got lots of “mentions,” while the little ones got a few. I happen to prefer one of the little ones (or some of the independents), so that would be my recommendation to a neighbor.

    I guess the bottom line is trust. I’ll use a service if I trust that the results will be decent most of the time. Since Grayboxx is new and hasn’t shared its formula or data sources, it’s starting with very low user trust. Perhaps they’ll earn it over time. I hope they do because I like what they’re trying to do and I love to see them taking on the big boys and aiming to serve the smaller markets.

  2. It’s interesting that Grayboxx is emphasizing small towns so much. It does look like a repositioning, doesn’t it? But I’m not sure this small-town focus has a strong business model on a nationwide basis — not until there are stronger distribution networks for small-business ads, anyways.

    (What is the stated business model, by the way? Have they said?)

    In general I agree with your comments and those of Michael Wood-Lewis. I posted my early assessment here today, based on limited use of their beta. I think the main task for Grayboxx in the next few months will be to tweak its algorithms — and its presentation, possibly — based on feedback from real people in the markets where it launches. While it doesn’t seem to be working particularly well right now, I think it has some potential. A listening tour worked for Hillary, and I suspect it’ll work for them, too.

    I’d very much like to hear some real information about the sources Grayboxx is plumbing, by the way, and I suspect their users would also. I think the company would get much more useful feedback if people understood what’s behind the curtain.

  3. Michael,

    As Laurence pointed out, I believe our sites are complementary rather than competitive. I hope you view grayboxx through that lense as well.

    Regarding accuracy of grayboxx’s recommendations, nothing is perfect – not Google Map’s algorithm nor Yelp’s reviews. Google uses URL references to rank local businesses which is problematic since many local businesses don’t have websites. And Yelp, while terrific in metro areas for categories like restaurants, is noticeably absent in small-midsize towns. While Michael has pointed out a few cases where grayboxx has not worked optimally, there are many examples where it has. For example:

    The first result on grayboxx for Burlington restaurants is “Cafe Shelburne” which won The Vermont Epicurious Awards for “Favorite Restaurant” and for “Best Restaurant Service.” it was also nominated for “Top for Ambiance,” “Favorite Wine List” and “Best Chef.” Per Michael’s point of bigger businesses placing first, there are probably larger restaurants with higher budgets; nonetheless Cafe Shelburne placed first according to our algorithm.

    The first result for architects on grayboxx for Burlington is Truex Cullings which has won more awards than I can list here (http://www.truexcullins.com/grapevine/awards.php). Its a highly reputable firm that has been in business for 30+ years that is clearly recognized for their first-rate work.

    Michael – we don’t claim to be experts everywhere, but I think the site should be judged in full, including where it provides accurate, useful results.


  4. Bob… No doubt Grayboxx sets an ambitious course. While I’ve never heard of Cafe Shelburne (so you’ve piqued my curiosity), I’m certainly aware of Truex Collins’ local reputation. I look forward to trying Grayboxx over time and welcome it to our town.

    I guess I’ve got a bias against computer-generated recommendations generally… I’d rather talk with someone I know or take the opportunity to meet someone new and ask for his/her two cents.

    But even for an odd character like myself, an online tool such as this can be a great asset, and I prefer Grayboxx’s approach over the big boys.

Comments are closed.