The premise of Ad.ly is to be the Twitter advertising network that matches celebrities with brands. The goal is to drive audience click-through to those brand destinations thanks to the conversations carried on by the stars with their followers. So far it’s racked up some 1,000 celebrities, about 150 brands, and 24,000 celebrity-delivered tweets around brand engagements – tweets that can generate a lot of interest. As head of engineering Chris Testa explained at SemTech this week, a top celebrity like Kim Kardashian can get some 15,000 clicks on a tweet, vs. about 400 on average for a NY Times or Wall Street Journal tweet. For a brand to really talk to people, he says, “it’s good to have a Kim Kardashian tweet, ‘This is what I am into,’ vs. having an ad in the NY Times.”
Interview with Ad.ly's Chris Testa:
To grow the business, the Semantic Web came into play to help determine of some hundred thousand prospects other influential personalities – to reach out beyond the big-name celebrities that have huge scale to others from the worlds of acting, athletics, music, and more that could be Twitter-iffic for specific brand campaigns in different niches. Testa says he had had his skepticism about the Semantic Web back in the days when he worked with Professor Jim Hendler as a student at the University of Maryland on an early SPARQL program. But he needed a scalable way to discover data about candidates and make it actionable, integrating it with the business from pre- to post-campaign analytics, so that Ad.ly can determine who might be good fits for a campaign or how musicians, for instance, compare to athletes in delivering end results.
Thanks to Freebase, and its tera-sized datasets that include well-annotated sets of celebrity data, he found himself a convert from skeptic to enthusiast when it comes to using and integrating Linked Data for Ad.ly’s business intelligence needs. “There’s a huge corpus of [celebrity] information so you could apply the Semantic Web to interesting problems with a lot of data,” he says. “It’s a goldmine from Freebase.”
Testa has created a 5-step plan for integrating Linked Data that it applies to its work and that other businesses could take advantage of, as well. It includes
- Understanding what your own ‘things’ are
- Choose a Linked Dataset that matches that
- Reconciliation of your data with Linked Data
- Build business intelligence
- Get feedback and perform maintenance.(including giving back information you learn to the global knowledge world).
Next week, in Part II you can learn more about the Five Steps In Depth as the video interview with Testa continues. Also be on the lookout for Blingalytics app from Ad.ly this summer, which Testa says will provide a friendly way for business people to look at data sets (which may or may not include data drawn from the Linked Data world).