Loading ...« Local Connections and Garden Doors | Home | Local Search: Think Data Mining »
If a Business isn’t in a Search Engine?
By Cameron Ferroni | September 24, 2007
…..Does it really exist? From my perspective, the single hardest problem in local search is knowing what businesses actually exist. There are lots of second-order problems: location, fuzzy search, what data to display, how to moderate user comments, scale - and I’ll touch on a bunch of those in future posts. But at its foundation, it all starts with whether or not the business exists, has existed, or now exists in a new location.
I looked at two examples through the following lens - if I type “business name, Seattle, WA” into the search boxes of Google, MSN, Yahoo, Open List, Internet Yellow Pages, CitySearch, and UrbanSpoon (for restaurants only) - what do I see? First, I did “Marriot, Seattle WA” - on this one most everyone did pretty well - there are 4 actual hotels, and all but one of the searches came up with all 4. Depending on the geographic boundaries used this was easier on some (www.openlist.com restricted it to just Seattle, so there were only 7 results to filter through, 4 of which were clearly the hotels we were looking for) and harder on others (www.google.com used a pretty broad range, and ended up with 2,641 results, and you had to scroll through 2 pages of results to get the actual 4).
My second example was a little more tricky - the Pho Bac restaurant in Seattle, WA. Turns out there are 4 instances of it in Seattle, if you include Pho Bac III down in the International District. In this test, only www.urbanspoon.com listed all 4. The others all ranged from 1-3 hits in Seattle proper. Even more interesting was the fact that the Pho Bac III turned up on almost all of them, but it was just called Pho Bac.
What’s my point? My point is that this is an incredibly hard problem to solve. As a maintainer of this information, you can only be so accurate. You get a list of businesses from a source (usually a company that specializes in this sort of thing), and you build from there. But then what? With 15,000,000 local businesses in the United States, you certainly can’t call all of them. And as you can see above, you can’t just take ones that seem weird and search online, because there is no definitive reference. The best thing you can do is do a combination of automation, with customer input, and editorial validation. Finding that right balance is the first critical step to building a foundation for a leading local information service.
Topics: Local Search |