This post was first published by Hanah Johnson on March Communications’ blog, PR Nonsense, and may be viewed here.
Determining a tweet’s popularity – and essentially its spread-ability – used to be based on follower numbers and the user’s Klout. But, striving for a Gaga-level following (over 25 million!) or Bieber’s perfect Klout score is no simple feat, if even possible for your average-Joe tweeter. Luckily, researchers over at HP are offering an alternative solution to achieve maximum attention and impact for news-related tweets.
An article on Mashable details how HP has developed an algorithm that allows news organizations to predict the popularity of outgoing tweets (with incredible 84% accuracy), even before posting them. With this tool, news organizations can calibrate their content to ensure high rates of sharing and retweeting. To determine the key components of this tool, researchers first explored four factors potentially influencing a tweet’s popularity.
- The news source – Which outlet created and published the story?
- The category of news – Technology, Entertainment or Sports, for example.
- Name recognition – Are celebrities, famous brands or notable institutions mentioned?
- Language – Is the tweet written objectively? Or is it emotionally-charged?
After analyzing a dataset of over 40,000 news tweets against each of these factors, researchers developed a prediction algorithm to determine a tweet’s shareability success. The findings used to create this algorithm demonstrate some interesting results that both support and challenge conventional social media wisdom.
First off, news category made a significant difference to a tweet’s popularity. Tech news was found to be the most tweetable type of content, followed by health and general “fun stuff.” Name recognition was also noted as an important factor. So, while a tweet about Lady Gaga is very likely to be passed around the Twittersphere, a tech story about Lady Gaga might double that tweet’s sharing potential.
But, the factor that led most predictably to the popularity of a tweet was its source. It’s interesting to consider how a news story tweeted by @NYT might fare better than the same story as tweeted by the actual subject! This particular finding goes to prove that even on the Internet, brand matters. Furthermore, a sense of trust and credibility is what makes that brand matter.
Contrary to journalistic belief and probably the most surprising result of the study was that emotional language doesn’t seem to have any effect on a tweet’s popularity – in fact, it’s just as effective as objectively-framed content. Whether your tweet is simple and straightforward, or uses hyperbolic language detailed with colorful adjectives, makes no difference. This especially is worth noting, since brands engaging in social media often adopt a particular style of “voice” that resonates with their audience. On Twitter however, this doesn’t seem to matter at all. What really matters is content. Pair a conversation-worthy topic with a credible source and you’ve got the most popular tweet around.
What kinds of news tweets are YOU most likely to share? Do any these four factors play into what you see as a share-worthy tweet?