[The following was published in Politics Magazine in July 2009 with title What do you look like on YouTube?]
Political parties and candidates bombard us with their messages and counter-messages; they always have. But with new social media thriving—from blogs to Twitter to YouTube—what matters more is how the audience of these messages responds to them and shares them.
For a long time there’s been no good way to measure that social media presence. But we at UNC-Chapel Hill have developed two programs that help keep track of what’s happening on YouTube and other social media.
The first, ContextMiner, is a framework to collect, analyze and present not just data but contextual information. The ContextMiner website provides tools to collect data, metadata, and contextual information off the web by automated crawls from blogs, YouTube, Flickr and Twitter. The second, TubeKit, is a toolkit for creating YouTube crawlers. These allow you to search YouTube based on a set of seed queries. Both are free, open-source projects distributed under Creative Commons licenses.
The 2008 election—which set a new standard for YouTube use—gives a good example of how the programs can be helpful. We monitored election-related YouTube videos beginning in early summer 2007. We developed a system to simulate a visitor who goes to YouTube every day, looks for election-related videos by running certain queries and then browses through the top 100 videos for each query. Just before the elections, we analyzed our data to see what our hypothetical YouTube visitor would have discovered about both the presidential candidates.
We found that running the queries “Barack Obama” and “John McCain” every day for about 18 months and looking at the top 100 results, we could have seen about 600 unique videos for the former and only about 100 for the latter. The official channel of both candidates had a similar disparity, with nearly 1,500 videos on Obama’s website but only about 300 videos on McCain’s.
More importantly, the Obama videos that we found using our query approach had significantly higher views than those of McCain’s. The videos found by running the Obama query totaled about 35 million views (60,000 per video), whereas the McCain query’s videos had only about 2 million (20,000 per video). What’s more, Obama videos had significantly more comments (70,000 vs. 24,000) and ratings (220,000 vs. 15,000). What all these numbers indicate is that while Obama had many more videos than McCain, community involvement around Obama videos was also much more successful.
Yes, sending more messages—including YouTube videos—and reaching out helps. But whatever your outreach strategy, it’s important to see how many bangs you are getting for your buck. Now you can measure those bangs online.