Celebrity Tweets May Not Pay Off
Yahoo researcher says predicting influence is not a science.
Yahoo researcher says predicting influence is not a science.
Paying a celebrity to tweet on behalf of your brand may not be the most cost effective approach to delivering a marketing message.
Duncan Watts, principal research scientist at Yahoo Research, said today that marketers may get better results if they reach out to a large number of “ordinary influencers” rather than one high-profile person on Twitter.
Watts, speaking at SES New York (a sister organization to ClickZ), said it’s next to impossible to predict what makes a Twitter message go viral – or be retweeted by other Twitter users. On average, he said, one message is retweeted 1.2 times.
With odds like that, Watts said paying a celebrity to tweet or retweet a message might not be a good investment. He referenced a 2009 report that said Kim Kardashian was paid $10,000 for an endorsement tweet.
“It’s better to trigger many small cascades,” he said. “Ordinary influencers are promising. Grind away [in] a very systematic manner.”
Watts also shared other trends and factoids from his team’s research:
* The top 20,000 Twitter users in four categories (media, bloggers, organizations, and celebrities) created 50 percent of all tweets received by Twitter users. These accounts also represented 45 percent of all following relationships.
* Celebrities tend to follow other celebrities. “They don’t pay attention to others. The media is focused on media. The only people who care about anyone else are the organizations – they are on Twitter, not to only to tell people about things, but to listen and to get feedback,” he said.
* Of the tweets tracked by Watts, bloggers were responsible for 1.36 million tweets – or 272 tweets per account; the media was responsible for 5.1 million tweets, or 1,023 tweets per account. And “ordinary” people on Twitter were responsible for 244 million tweets, or 6.1 tweets per account.
Messages travel on Twitter in much the same way that information was disseminated more than 50 years ago. Watts pointed to research in the 1950s by Elihu Katz and Paul Lazarfeld; they found there’s a “two-step flow of information,” whereby opinion leaders and influential people were more likely to be exposed to media and then disperse that information through the social strata. “There was no Twitter. There were no followers. Yet they used language that’s very amenable to Twitter today,” Watts said.
“The lesson for marketers? Give up on predicting individual events… focus on the typical event size. Try to optimize many, many times.”