You’d think if there’s one thing a location-based app can do, it’s getting locations right. But that’s actually proving quite a challenge. Location-based ad company Placecast, says that discrepancies in location can lead to millions of incorrect listings. And there is no single, standardized database for location, nor will there ever be, Placecast believes.
Here’s the issue: An application can access one source for restaurant listings, another for bars and a third one for movie theaters. These sources may have different ways of identifying a location that can be hard for the application to reconcile. Placecast released an application programming interface in March, called MatchAPI, which goes through these different sources and “cleans” them up, trying to make sure all the different references to a location point to the same physical place.
According to Placecast, when MatchAPI cleans a data set, there’s an average fault rate of 8%. This grows to 40% when dealing with data sets that include high proportions of user-generated content. Placecast’s platform has over 20 million records, and these error rates translate into at least 1.5 million incorrect listings.
In the two months MatchAPI has been out, more than 200 developers have signed up for the free API. Placecast sees itself as providing the “plumbing” developers need to monetize their products. “At SXSW in March, we realized the need developers have for clean location data. We had already solved that problem for our business, so we decided to put it out for the community”, said Placecast CEO Alistair Goodman (pictured).
Besides keeping location data sets clean and up-to-date (in the U.S., more than 20,000 business listings change every month in some way; some businesses open, some close, some change locations, some change phone numbers, etc.), there is a bigger question of location-data providers trying to push their proprietary services on developers. Nokia wants developers to use its data, and the upcoming Google Places API is Google’s way to push location data. If a company doesn’t already have location data, that may be all well and good. But if a company is aggregating data from multiple sources, why would it want convert its identification systems to an exclusive system, Goodman asks.
“The thesis is that a company uses an ID system at the expense of somebody else’s ID system, and this creates lock-in.” And as Placecast is a middleman between advertisers and consumers, it makes sense for the company to create a system that’s as frictionless as possible for marketers.
VB's research team is studying web-personalization... Chime in here, and we’ll share the results.