Deep Data and the Semantics of Local
In my commentary last week, “The Long Tail of Local Search,” I discussed the need for online and mobile search applications to provide a broader view of the local marketplace, expanding beyond the well-worn categories of pizza and hotels to provide a full representation of specialized stores and services. Today I’d like to focus on another pertinent need: the need for deep data on individual businesses.
Most U.S. residents live in suburbs and travel infrequently. Our habits keep us close to home, where we are already experts in the types of local businesses we visit every week. As I pointed out last time, we need the most help meeting less common needs. A truly forward-looking search application is one that can deliver rich, detailed information about the full range of local businesses.
My last commentary took up the example of lawn and garden services. In my midsized California town, the search results for this category and several others are so incomplete that depth of data is not really the primary consideration. What is needed most are more listings in more categories, plain and simple. But in a more populated area like San Mateo, California, the search results at least give us a reasonable set of options. See, for example, the blended local listings in Google’s first results page for “lawn and garden services san mateo ca.”
There are ten listings to choose from, five of them located in San Mateo. Two of the results seem to be for hardware or building supply, not quite what I was looking for; still, 80% of the listings, a decent percentage, appear to be relevant to my search phrase. But what if I were searching for a specific service within the lawn and garden category? Here’s a real-life example: the front lawn of my house is not getting enough water because the soil has become compacted. What fixes this problem is aeration, a service only some lawn and garden companies offer. If I wanted to compare lawn and garden companies in San Mateo that provide this service, I would have no clue how to do so given the results on this page.
I could try clicking on the “Google+ Page” link for each result to see if each page provides the needed level of detail, but the first listing does not give me much hope, showing only contact information and the keyword “Landscaper.” Instead, I will probably be forced to do what I would have done twenty years ago after looking up the same category in the yellow pages: call each of the companies and ask them if they provide this particular service.
Surely local search can do better. Businesses want to provide rich information about their products and services, and consumers want and need that information in order to turn local search from a frustrating experience into a useful and productive one. A lawn and garden company I’ve used, Caddie Shack Pest Control, has a great website at www.caddieshackpestcontrol.
Google itself has pointed the way toward at least one solution with its forays into structured data. Certain types of information about a business, such as basic contact information and review highlights, can be displayed in search results if web designers make use of microformatting. More recently, Google, Yahoo, and Bing teamed together to launch Schema.org in order to promote the use of structured data. Among the schemas on this site is a Local Business schema with markup for describing payment methods, price range, hours of operation, and certain business categories such as automotive, dry cleaning, and child care. It’s a somewhat anemic set of options at present, but the potential is enormous. If all directory services and website designers made use of semantic markup for the full range of business profile information, including products, services, brands, specialties, and the like, we would have a huge crowdsourced repository of local search data at our fingertips. Google could use structured data to provide useful differentiating information within search results like the example above.
Writers like David Weinberger have emphasized that the non-hierarchical structure of the internet makes it ideally suited to store and sort large amounts of data in very flexible ways. The notion of a Semantic Web, in fact, has long been the dream of Tim Berners-Lee, the inventor of the World Wide Web, who has for many years encouraged the development of markup conventions that would allow machines to understand the meaning of web pages. What Google has done thus far with microformatting is a very small piece of that puzzle, though as others have suggested, the wholesale conversion of all web pages to a semantic model is likely to be unrealistically ambitious. In contrast, targeted projects such like the markup of local business data are a practical way to move the semantic vision closer to reality.
In the case of local search, there is no prevailing reason beyond lack of attention to prevent the industry from doing a better job of serving the full range of consumer needs. For the time being, industry attention is still directed in a self-fulfilling way toward the activities that have always received attention. This is the safe bet, but safe bets don’t lead to progress.
Google in particular is uniquely positioned to move local search forward, given its central position in the local ecosystem and its connection through Schema.org to Yahoo and Bing. If Google were to support a complete semantic model for local businesses on Google+ Local, the market for that information would immediately and exponentially grow. We would see every SEO in the country helping local businesses to mark up their products and services for consumption by Google, and every local directory rushing to follow suit. Companies like UBL that house data on local businesses would be well positioned to switch to semantic formatting, making their repositories fully available for search.
Surely there would be concerns, as always, about gaming the system. But for the most part, business owners stand to gain more by feeding real, relevant data into the ecosystem. It’s a win-win for businesses and consumers.
Damian Rollison has served as VP of Product for Universal Business Listing since 2010. He holds degrees in English from the University of California, Berkeley, and the University of Virginia, where he did graduate work at the Institute for Advanced Technology in the Humanities. Damian’s articles on emerging technology have appeared in Venture Beat.
