RDFa: The Inside Story from Best Buy
An Interview with Jay Myers
Late last year, I approached Jay Myers, Lead Development Engineer for Best Buy, and asked him if he’d be willing to be interviewed about his project of implementing RDFa on Best Buy’s web site. He accepted, and we began exchanging information, a process that culminated in a voice interview last Monday.
It’s a lengthy interview, so grab some popcorn;
1. I know you’ve been interested in RDFa for some time. Was this Best Buy effort your first effort at implementation of RDFa?
“Yes. It’s going on three years since our first foray into heavy duty Semantic Web coding. I initially started out using microformats, but found that RDFa (and subsequently microdata) was a more powerful and stable coding methodology based on many years of work in academia and by semantic web practitioners.”
Jay went on to say that in the beginning, he played a part in reviving a microformat called hProduct. He was attempting to address some issues using code and the microformats he was using weren’t meeting his needs. That’s when he realized that RDFa was more robust, offering more flexibility. That’s also about the time that he made contact with Dr. Martin Hepp, developer of the Good Relations ontology, with whom he began to communicate closely.
2. Many seem to fear that RDFa implementation is a formidable task. Others question its contribution to ROI. I would think that would make it difficult to convince some organizations to make the investment of time and/or money. How difficult was it to sell the concept internally at Best Buy?
“Initially we didn’t do any “selling” of the concept…it was something I worked into a project as an experiment. The great thing about RDFa is the ability to weave meaning and rich data directly into a web page without having any impact on the front-end user experience.”
Expanding on this, Jay says he was simply experimenting initially, in an effort to see if they could incorporate RDFa into some of their 1,100 stores’ pages, without any adverse affects. “Management doesn’t read code”, he said. So with no front-end impact, he didn’t have to sell the concept, he just did it!
3. What medium-to-long-term benefits did you expect to see for Best Buy?
“We really didn’t go into it with any expectations. We just wanted to see if it was something we might want to do. That’s why we were caught by surprise by the results… we weren’t really expecting any.”
4. What benefits turned up?
“Within just a couple of months, we began to see an increase in our organic search results. Before long, it had increased by 30% over historical rates. We also saw an increase in our click-through rate. Yahoo did a study a while back and found that people that had rich snippets on the results pages were seeing around a 15% increase in CTR, which has proven to be the case for us. And of course, it makes our web site “smarter” and more open to machines, which ultimately benefits customers.”
5. What led you to choose RDFa over alternative formats, such as microformats or microdata?
“I found that RDFa was a much more stable concept – based on the use of long established vocabularies (also known as ontologies) that have existed for years. My first foray into giving objects better definition on the web was revitalizing the hProduct microformat. Working through that implementation I found limitations in microformats that left me searching for a “beefier” solution. At the time, microdata hadn’t hit the scene yet, so RDFa was the natural choice.”
6. What architecture strategy did you employ, and what were the specific reasons for that choice?
“Initially, we deployed RDFa markup through our local stores’ WordPress blogs simply by weaving the rich markup and attributes into the WordPress themes. An advantage to “front-end semantics” is that a developer or team doesn’t have to use a particular platform or employ a particular arch strategy to populate their sites with rich data. A savvy developer could hand code the stuff into their work just as easily as adopting a semantic-specific platform.”
7. How hard was it for Best Buy to implement RDFa? What was the methodology?
“Initially, implementing RDFa was a personal challenge, as a developer has to shift their development mindset and consider not only what the visual output of the code is, but what the machine output is as well. Outside of a couple of extra validation steps to confirm the RDFa output could be successfully distilled by machines and software, the coding methodology really didn’t change that much. Over the past two years I have worked hard to automatically build semantics into my project methodologies – making it a standard rather than a separate implementation step.”
8. What are the greatest benefits you’ve seen from implementing RDFa?
“The rise in our organic search traffic, and the adoption by major companies like Facebook and Google.”
9. How fast did you begin to see benefits, and what were they, intially?
“Within 3 months of our initial deployment we saw a surge in our organic search engine traffic. That eventually reached 30% and held it.”
10. Did you see a rankings increase on Google?
“This is hard to tell – I’ve had a difficult time separating search traffic from scraper applications utilizing traditional web analytics. To my knowledge, there are a handful of interested individuals running distilling and parsing software against our RDFa pages.”
11. You said you saw about a 15% increase in CTR. Did you notice any appreciable change in your bounce rate?
“Not really. Some stores, of course, may be less active on their blog, and they see a higher bounce rate because of that, but that’s something we want to watch closer.”
(This started out as an experiment, so the results caught them by surprise. They hadn’t really been monitoring to the extent they might have, had they expected such results.)
12. What are your plans now?
“We’re continuing to implement RDFa on product pages across our various web properties and have our attention focused on the perceived impact that rich RDFa markup could have on sales numbers.”
13. Do you foresee any major impact to the acceptance and implementation of RDFa by the release of Drupal 7?
“Kudos to the Drupal folks for implementing semantics into their CMS. I believe that the Drupal 7 release will make semantics even more accessible than ever before to everyday developers (and many people with no dev experience working on the web) , and allow them to easily open up and release useful data through RDFa.”
14. Any specific comments or recommendations you’d like to offer those that are either on the fence about implementing RDFa for their site, or have decided it doesn’t offer them sufficient ROI to make it worth the effort?
“There’s definitely a benefit in being an early adopter, that’s one thing. Additionally, with RDFa, we’re getting closer to the Semantic Web. Tim Berners Lee is the one asking for our data, and I think it makes sense to do that. RDFa is one of the… what I call a gateway drug… to the Semantic Web.
You don’t have to be an expert, you can be an every day developer or somebody that uses a tool like Drupal or another CMS to put out rich data in RDFa. So I think it’s not as hard as people make it out to be. Again, it was more of an experiment for us, it wasn’t a hard-core effort, and the hope is that we can make this part of the everyday development, just like coding HTML… it should be just that easy.
We’re seeing a lot of adopters, including some big names, like Google with its rich snippets and also Facebook. Facebook has something called the Open Graph, that also uses RDFa and rich data. So there’s a benefit there, and I see more adoption by the larger and more important firms. It just makes sense. We’re living in a world where there’s a huge amount of data on the web, and it’s only going to continue to get larger. And I don’t believe that current or traditonal SEO practices, page sculpting, things like that… I think that can only go so far. But traditional SEO efforts, in combination with RDFa and some more development-centered efforts, can really go a long way for people. So that would be my words of encouragement.
We’re starting, as the Semantic Web community, to engage the SEO community a little bit more, to try and get everybody on the same page. So I’m going to be speaking in Austin on March 1st, at the Semantic Web Meetup there. What I’m trying to do is tailor this message not only to a technical audience, but to the SEO audience, too. There is an effort now to try to tie development and SEO together. The more rich data we have on the web, the better off we’re going to be.”
Making the Case for RDFa
As you can see by the results achieved by Jay’s “experiment”, the effort was well worthwhile. 30% higher organic search results and 15% higher CTR are nothing to turn your nose up at. The results showed up fast and have been sustained.
I think Jay made a very important point, when he said that implementing RDFa allowed him to, “weave meaning and rich data directly into a web page without having any impact on the front-end user experience.” Users will see nothing different on our pages. Machines, however, will be able to understand more of what our pages are about.
The second very important take-away, I think, is this: “… it wasn’t a hard-core effort, and the hope is that we can make this part of the everyday development, just like coding HTML… it should be just that easy.
“Just that easy.” Simply a different mindset, for the everyday developer. Or if you prefer, you can use Drupal 7, which automatically includes RDFa, or one of several WordPress plugins for the purpose. The point is, utilizing this technology is not rocket science. It can, however, take you to the stars.
Your comments and questions are welcomed. Jay will undoubtedly be stopping by, and should be able to offer clarification or further detail. You can also visit his blog, Random Musings of Jay Myers or follow him on Twitter.