Testing live-tweeting with @MLB

Thursday, 3 October 2013

As any stat-obsessed, Moneyball-loving baseball fan knows, Major League Baseball (@MLB) and data go hand in hand. And we feel the same way, especially about the numbers around Tweets.

We recently embarked on a data-driven analysis of the Twitter strategies of a select group of @MLB teams. Our goal was to see if by live-tweeting games — in which teams post updates, reactions, photos, videos and more on Twitter in real time — teams could boost follower engagement and grow their audience on Twitter.

Which strategies were most effective at helping teams meet their specific social media objectives?

To answer that question, our sports partnerships team worked with @MLB to help us examine the efforts of five teams using five different methods of tweeting during games. We collected key data for each method, including follower growth for team accounts, retweets, favorites and other forms of engagement. To obtain a clean dataset, teams could only experiment with one ‘type’ of live-tweeting for each game.

The five teams in the analysis: the Cleveland Indians (@Indians), Oakland Athletics (@Athletics), Los Angeles Dodgers (@Dodgers), Seattle Mariners (@Mariners) and Texas Rangers (@Rangers). And the live-tweeting formats we studied were:

  1. Play by play: Tweeting each play of the game in real time
  2. Takeover: Turning over a team’s Twitter account to a high profile fan or expert (for example, the Dodgers asked long-time announcer Vin Scully to tweet his thoughts and reactions during a game)
  3. Vine videos: Using Vine to post six-second clips of action during the game, or in the stadium, to complement the TV coverage
  4. Behind-the-scenes photos: Tweeting out pictures from the locker room, the dugout, and other spots to give fans access to parts of the game-day experience they rarely see
  5. Fan engagement in-game conversation: Responding directly to fan Tweets during the game, engaging fans in ongoing conversations in real time

To help establish a baseline (control group) we also analyzed the results of games in which the teams engaged in their normal Twitter behavior, which usually combined several types of live-tweeting, rather than applying just one. We examined features common across all types of live-tweeting, such as using a single official hashtag.

The first takeaway was that any concerted effort by a team to engage its fans led to an increase in followers and engagement. From follower growth to Retweets, mentions and favorites, by simply communicating with their fans via Twitter, teams saw dramatic increases in user activity.

The top-level results are that live-tweeting of any kind results in:

  • Follower growth: 1.6x the average
  • Retweets: 1.9x the average
  • Mentions: 1.5x the average
  • Favorites: 1.5x the average

And a closer look at each individual type of live-tweeting reveals even more nuanced results, which you can see in the following set of charts:

1. Followers

For teams looking to increase the number of people following their team account, letting celebrities or experts take over their account during a game led to the highest increase in follower growth (+125% compared to average). As an example, the @Rangers had team president and general manager Jon Daniels, answer fans’ questions:

2. Retweets

While every method of live-tweeting resulted in more Retweets, it’s most striking to see the impact of Vine videos (each team posted Vine videos throughout the day when implementing this strategy). One standout: The @Indians gained a staggering 454.6% more followers than their standard rate of follower growth on the days when they live-tweeted Vine videos.

Here’s how they used Vine to show what was happening around the stadium that night, and to go into the dugout itself.

3. Mentions

 

Vine videos had the greatest impact on mentions as well — with an increase of 81% over the average. But this wasn’t the case for every team, suggesting that content makes a big difference. Behind-the-scenes photos had the smallest effect on overall mentions — although there was still growth compared to the average.

Vine videos had the greatest impact on mentions — with an increase of 81%

4. Favorites

Once again, posting Vine videos had the biggest overall impact (+112% over average) — this time on favorites. But there was variation between teams: The @Dodgers had the most favorites when implementing fan engagement strategies, while the @Rangers got more favorites from behind-the-scenes photos. But when looking at the teams as a whole, Vine outperformed the rest.

One Vine video in particular stood out from the rest: The @Dodgers posted a Vine video of @GirlsGeneration singing the national anthem, which gained more than 500 favorites.

5. Hashtags

Using the team name as a hashtag makes following the game easy for fans, but it’s a hard thing for the team to control, as followers often come up with their own hashtags. Users retweeting play-by-play Tweets were the most likely to use the official hashtag. “Takeover” retweeters were least likely to use the official hashtag; for example, the Dodgers changed the hashtag they promoted from #dodgers to #VinScully when the famous sports announcer took over its account.

The Vine effect

Across all other measures of engagement, Vine videos dramatically increased the ways that followers interacted with a team. The overall impact was that Vines got:

  • 2.3x more retweets than average
  • 1.7x more follower growth
  • 1.8x more mentions
  • 2.1x more favorites

People tweeted more often when the teams posted Vine videos than when they used any of the other live-tweeting strategies — a nearly 5% increase compared to the control sample.

Teams tweet more when they’re winning

There were an average of 41 Tweets per day for a winning team, compared to 32 per day for each losing team — a 56-44% split.

The @Rangers bucked the trend by tweeting more when they were losing. In fact, tweeting when you’re losing isn’t bad for a team, especially when you’re playing a team with a bigger fanbase. The @Rangers lost 8 games out of 11 during the period we measured — but actually gained proportionately more followers than any other team, an average of +16% above the control week.

The Twitter Media team will be doing more of these focused data dives in the future. What area would you like to see us examine? Contact us at mediablog@twitter.com.