ME* Conference 2010

Five Things You Didn't Know About Social-Media Tracking

A Look at Some of the Data and Adaptations From Twitter's Front Lines

By Published on .

(From l.) Mark Ghuneim, Simon Dumenco and Josh Auerbach at Ad Age's ME* Conference.
(From l.) Mark Ghuneim, Simon Dumenco and Josh Auerbach at Ad Age's ME* Conference. Credit: Pete Kolonia
NEW YORK ( -- If you missed Ad Age Media Guy Simon Dumenco's panel on how social networks are fueling a real-time media flood, here are a few surprises from the talk.

Bieber's clutch on Twitter is scarier than you thought
According to Mr. Dumenco, who moderated a panel about real-time media, Justin Bieber has become the fresh, prepubescent face of Twitter. Citing a potentially dubious stat, he said that pop phenom accounts for 3% of traffic on Twitter at any given time, occupying racks of servers. Josh Auerbach, senior VP at Betaworks, a real-time media incubator, said that according to an analysis by a Betaworks company, 10% of Bieber reply tweets contain the phrase "I love you."

"Bieber is driving adoption [on Twitter] like nobody could," Mark Ghuneim, CEO of Wiredset, later said.

Digital can actually help TV
There actually is a correlation between all those tweets about TV programs and actual viewership. Mr. Ghuneim said he and MTV saw directly correlating spikes in tweets and viewership when Kanye West famously stole the stage from Taylor Swift at the 2009 MTV Video Music Awards.

That means networks can actually use digital and social media to get viewers back to TV.

"Spend digital dimes to affect analog dollars because it can affect social tune in," he said.

Machines can learn what's interesting
SocialFlow, one of the companies Betaworks has invested in, is using math to figure out when to get the most bang for your tweet. Here's how it works: For SocialFlow clients like The Economist, an algorithm looks at potential tweets queued up by a human (in The Economist's case, they're likely headlines and links to stories) and figures out what is most interesting based on current conversations on Twitter. For example, if Twitter explodes with conversation about the Irish bank bailout, SocialFlow will tweet a story related to that topic at that very moment.

"You can think of it as a human problem; there are things that are interesting and less interesting," said Mr. Auerbach. "If you look at it like a math problem, it becomes simpler....When a tweet gets to be sufficiently interesting, it goes out."

After using this system, The Economist has seen clicks per tweet spike about 50% and clicks per follower increase 30%, said Mr. Auerbach.

Malcolm Gladwell might be right
Malcolm Gladwell riled up a lot of social-media cheerleaders last month with his New Yorker column, "The Revolution Will Not Be Tweeted." His piece focused on how dissatisfaction and revolt are much easier on social networks, but that they'll never amount to the real guts that make real revolutions, like the Civil Rights Movement. Well, Mr. Gladwell has one more case study to add to his artillery. Wiredset's Mr. Ghuneim pointed to the TSA-body-scan fiasco that erupted on Twitter and other social networks leading up to Thanksgiving. As the busiest travel day approached, tweeters complained about the new security measures, which sometimes resulted in full-body, extra-personal pat downs. On top of claims that the new measures were "gate rape," there were threats of boycotts and protests to hold up security lines during the busy time. While he said one in 10 tweets prior to Thanksgiving had to do with the pat downs, or contained the hashtag #TSA, no actual protests actually surfaced. Security measures went unhindered.

"It was a really good example of talking the talk, but not walking the walk," Mr. Ghuneim said.

Click behavior is already starting to change content in real time
Betaworks project, the URL shortener, gets 7 billion clicks per month and tens of of billions of unique urls shortened everyday. But beyond gauging what links get the most love, Mr. Auerbach said content companies are starting to modify content in real time based on click behavior.

For example, a news organization is already changing the content of articles based on what readers do when they click. For example, if a certain link gets a lot of clicks, but readers don't scroll below the third paragraph, editors can see that a headline might be strong, but that the content of the story needs a boost. "If scroll depth is not that impressive, you can change the content of the page based on what you're seeing," he said.

Most Popular