Best Buy may be considered by many as a poster child for why businesses should use Semantic Web technologies. An early adopter of the GoodRelations vocabulary (just recommended for product and price information usage in web pages by Google), the retailer has been at the forefront of the Semantic Web wave – but it’s only just begun.
Semantic Web technologies, says Jay Myers, lead web development engineer at Best Buy, are a portal to better insight, in more ways than one (thoughts he’ll be sharing in more detail with attendees at this week’s Semantic Web Summit in Boston). “There is a vast amount of data that corporations produce or are able to consume,” he says – and within that a huge amount of insight most of them aren’t tapping into, which is a real loss considering the challenging environment that businesses such as retailers confront.
Marking up information using RDFa and vocabularies such as GoodRelations obviously can deliver more search insight to consumers, which works to the benefit of the business. As Myers points out, Best Buy.com has some 600,000 SKUs available at any one time, yet the majority of revenue comes from a small number of those products – an experience that can be very different when more attributes of other, perhaps less well-known products, are facilitated to more easily surface in their searches. “The Semantic Web and Linked Data give us the opportunity to start exploring some of those things,” he says, “for drilling deeper down into some attributes of products.”
But there’s so much more to explore, including getting more out of internal information, that so often is locked into its own silos within a company’s four walls, by making it accessible beyond its application. “It’s almost built into the Semantic Web to do that,” he says. And also, to potentially combine that with new sources of external data to broaden that insight a step further. “The ability for us to have our data about products that are selling well and trends, and mash that up with other forms of data – external sources on the web that identify different trends – into one structured source and query off of that – that lets you get more contextual,” he says.
It’s all good stuff, but how do you prove that the vision really will pay off? Myers agrees it is a challenge to help connect those dots for the business, but as a starting argument Best Buy can point to “a decent uptick in organic search traffic” following several projects to release RDFa. “I won’t point directly at adding semantic technology as the cause but I do believe that producing better, more structured data and displaying it on the front-end and making it accessible to machines plays a very large part.”
More detailed and more involved projects he has in mind to take its catalog data and mash it up with other data sources ideally will get a boost from what’s gone ahead so far. That includes one of the major Semantic Web kick-off projects Best Buy did, which was marking up with RDFa its network of active local store blogs. “That was interesting – we found since we started that we have got more traffic to localized store pages and we are hoping to see [that translate to] more store traffic.” While there’s not a formal analysis of that, he can point as one success to the “open box’ project that relies on those RDFa-enabled store blogs to bring in customers in search of better deals on returned items.
The open box problem is a pretty big one for the industry, he says, with between 12 to 20 percent of overall inventory returned at some point – usually for reasons other than actual problems with the product, so in most cases these are fully functioning and perfectly usable goods. Best Buy hadn’t been doing the best job communicating to potential buyers about these items on the web, but now a Semantic Web plug-in to the local store WordPress blogs makes it easy for the application to grab data about the SKU, plug in the Open Box price and the reason returned, and publish it to the site, usually on a weekly basis. “There are some 200 to 300 open box products sitting on shelves so it can be beneficial to get a web presence and markup for the machine to parse that this store with these coordinates and location has this product,” he says. And Best Buy knows, at least anecdotally for now, that consumers are taking advantage of this: “People actually have found these pages on the local site and printed out a copy of the product page and taken it to the store,” he says.
Sem Web Data, Everywhere
It’s an interesting time, with the Google announcement obviously set to draw more eyeballs to RDFa and the GoodRelations vocabulary, for early onboarders to the technology such as Best Buy. Will they face the innovator’s dilemma?
Myers imagines the opportunity for every web user if even just a fraction of those that publish HTML put more structure around it, given the interest in that by search engines like Google and Yahoo. Should developers worldwide choose to embrace coding in RDFa as a standard, it could be a very powerful way to contribute to the large web of data. “Looking at this from a customer perspective, I think having better data of all products will benefit those who purchase from these companies that do this,” he says.
And what for the companies who set this stage, as more of their own competitors join in the fun? “It runs on a bell curve – the early adopters have the advantage, then when everyone else does it something else will come in and give another advantage,” he says pragmatically. That said, he thinks those businesses that are as much real-world as they are virtual can perhaps wring more value out of the Semantic Web even as others step into the arena.
“In comparison with, say, Amazon and other primarily online retailers, the advantage is that we actually have physical stores and people who interface with customers every day,” he says. “If we can harness those interactions in data, that will give us the advantage in terms of what other sorts of data we can plug in and gain insights.”