Twitter  Quartz found an unlikely inspiration for its relaunched homepage: The email newsletter.  
Nieman Journalism Lab
Pushing to the future of journalism — A project of the Nieman Foundation at Harvard

It’s alive! What NPR learned from turning its @nprnews Twitter account from a bot into a human

Putting a person at the controls increased traffic and engagement by just about every measure. Now the question becomes: Is that boost worth the extra human effort?

May 22 was World Goth Day! To celebrate, NPR’s main Twitter account, @nprnews, sent out a tweet to its 2 million-plus followers sharing a 2013 NPR story on Goth Barbie. It earned 156 retweets — among the highest of anything NPR shared that day.

On a normal World Goth Day, NPR wouldn’t have sent that tweet, because its Twitter account is typically automated with an RSS feed of NPR stories. But last week, it ran an experiment by manually controlling the Twitter account during business hours.

How’d it go? During the five days of manual updating, there were 142,219 visits to NPR’s website from @nprnews tweets — a 45 percent increase from the average (98,213) of the five weeks leading up to the experiment, according to NPR’s Google Analytics data. Links tweeted by @nprnews were clicked on nearly 100,000 more times than links shared automatically the week before, information from its account revealed. And the account gained 5,010 followers — about 14 percent more than the week before.

Lauren Katz, an NPR intern turned temp, led @nprnews during the weeklong test. The week before, she ran a smaller experiment of sorts by manually tweeting a handful of stories at night after they’d been tweeted out by the RSS feed earlier in the day. The tweets she wrote out received more engagement across the board, inspiring the longer trial.

At the beginning of the week, Katz sent out tweets every eight to 10 minutes. Later in the week, she lessened the pace to only once every 20 minutes. During the trial, she sat near The Two-Way, NPR’s breaking news blog, in order to more effectively tweet out breaking news as it occurred. And though Katz was able to schedule some of the tweets ahead of time, she said managing the account, at least early in the week, “meant that I didn’t leave my desk, basically,” as she tweeted out nearly every story NPR published, up to 100 stories daily, at least once.

“Throughout the whole week that I was livetweeting, I was doing all the other social media for NPR News. I was still posting to Facebook manually, doing our Tumblr posts, pinning on Pinterest,” Katz told me. “So the rest of social media didn’t stop, and I found it really hard to write human tweets and keep up with everything else I was doing. I did it, and we were there all week, but it takes more manpower if we want to continue it.”

And that’s manpower that NPR might not have, even though the experiment was successful in increasing engagement. Including Katz, there are three people on NPR’s social desk, but Katz is likely leaving the network when her contract expires next month, meaning the desk will be spread thinner. Though they haven’t yet decided how to implement what they learned from their week of manually controlling Twitter, NPR social media strategist Melody Joy Kramer said she’s considering a few ways to have increased human control over @nprnews without it becoming an overwhelming burden.

Some news organizations, like The Wall Street Journal and NBC News, maintain manual control of their Twitter accounts around the clock with social media staffers spread across the world. That’s not something NPR has the capacity for, but it does have a news desk staffed 24/7, and the team that produces Morning Edition arrives at work in the wee hours as well — many of them already tweeting from their personal accounts, Kramer noted.

“The question is what can people take on additionally to what they’re doing. I think when you’re doing radio, it’s a lot more time consuming than when you’re at a print publication because you have to edit the audio,” Kramer said.

At a large legacy news organization, still focused on producing news for the radio, making changes like that can be a slow and arduous process. But since Kramer assumed her current role last year, she’s made a push to make social media and digital analytics a larger focus at NPR. Just last month, she helped oversee the release of an analytics dashboard to better inform the newsroom on how it can reach its audience.

To that end, @nprnews also retweeted more NPR journalists’ own tweets and tweets from other NPR shows or desks. How individual NPR reporters use Twitter has received more attention of late as NPR media reporter David Folkenflik unleashed a torrent of tweets in the wake of Jill Abramson’s firing from The New York Times. Vox created a Storify of Folkenflik’s tweets that it published on its own website, and the whole episode sparked a debate both within and outside NPR on whether it was best for Folkenflik to tweet out information like that instead of posting it directly on NPR’s website.

“We had a few discussions about it, and the sentiment was we want our news to travel out and we want our reporters to be seen as experts in the field. And a lot of times you see news organizations picking up tweets from other reporters or citing other things, and Vox cited David appropriately and they linked back to his tweets,” Kramer said. “I was totally for that and would love to see other news organizations do that with our stuff in the future.”

But no matter how individual NPR staffers tweet, having a human being control the main NPR Twitter account created an opportunity for more interaction with the account’s followers to potentially allow them to engage with NPR content. Katz and Kramer said the feedback they received — both positive and negative — was helpful in figuring out how to best serve their Twitter followers. And they wanted followers to know there really was a human behind the account:

Photo of NPR headquarters by Eric Langhorst used under a Creative Commons license.

What to read next
Ken Doctor    Aug. 25, 2014
America’s largest newspaper company says it’s building for the future. But it’s hurting its own value proposition in the process.
  • Brian Krogsgard

    I should probably preface my comment by noting that I’m probably a fringe Twitter user, as I try to follow fewer people and read everything they post. That said, I find following traditional media sources’ main automated accounts to be nauseating.

    They almost all post every few minutes, clogging a feed if the user typically follows real humans that only tweet once or twice per day (if that). But once you get into media / news / politics, it’s a madhouse, with accounts sharing sometimes 100+ things in a day.

    I think organizations could serve themselves well (outside of the inner-news-circle dance party) if they consider the way their audience uses social media more. Primary accounts could share fewer items, but better curated “must read” kind of stuff. Keep a separate account for a raw feed.

    Anyway, this is what has led me to forego following most media companies at all, opting instead to follow some of the more mellow social media voices of the authors themselves (say, a Joshua Benton vs a Ben Smith).

    To make a long comment longer, here’s my question that I wonder if organizations have tested: Is there a follower cost to tweeting too much? I feel there’s considerable evidence of more tweets == more engagement, but I have a feeling more tweets (Some breaking point I can’t well define. Maybe 20 per day?) that I think costs followers. It at least costs my follow. I’d like to see that study.

  • Jon Cole

    I was sure I’d be reading about staffing when I saw this article pop-up on my Twitter feed, it’s a problem I can definitely sympathize with. That said, I’m one of those thousands of people who followed the NPR News account during this time and the human touch was noticed and appreciated. I hope that NPR manages to work in more manual tweeting into the mix, it really helps me feel more connected to your content and more likely to spread it to my own feed.

  • suziequeue

    I have to agree with this post. I follow several individual NPR reporters/anchors or program-specific Twitter accounts and not the main feed because I cannot handle the volume of tweets that it generates. I can see the value for some people of having a “raw feed” as Brian calls it, but that’s not what I am looking for. I’m looking for insights into how stories are put together, previews of upcoming pieces, and links to specific stories that are important or have generated a great deal of comment or response. A good example was following Steve Inskeep during his reporting of the Borderland series. Or following Scott Simon for WE Saturday.

  • wmstudio

    Big Cube brought all new gameplay.U never played before.We hate clones,dont be a clone.Get your time and your man under control!It was worth the try.App store:

  • Melody Joy Kramer

    Hi. I’m Mel and I work at NPR on the social desk.

    This is really helpful feedback as we think about transferring the account to actual humans (which is our goal.) I know that I prefer connecting with people and not with robots. We’ve been incredibly short-staffed as of late — and are using the RSS feed as a stopgap for the time being. Once we can transition the account to humans, I think the experience will be much like the one you’ve both described.

  • Melody Joy Kramer

    Thanks. I’m going to share your feedback with our newsroom.

  • Nathan Reimer

    If you have to tweet less but be more human, then do that. We have too much noise to wade through already. Make us want to read your news, humor, stories, etc…

  • Andrew Catalina

    This presupposes a particular consumption model, more akin to sitting down with the morning paper. I use Twitter more as a snapshot at any particular point in time of things that are happening (and I imagine quite a lot of journalists are moving this way). Thankfully, with a little work and time, these models aren’t mutually exclusive, if you set up lists and such.

  • Melody Joy Kramer

    I completely agree. But with a staff of 2….we’re responsible for posting to the main accounts, managing third-party relationships, building and releasing products, training our newsroom, consulting on series and reporting trips, and coding some things….something’s gotta give. Right now, this is our solution but I’m hoping we can figure out a better one.

  • Ashley Lusk

    I’m an avid NPR fan and a digital media strategist myself, but I admit that I didn’t quite love the human-run feed. I turn to the correspondents and hosts for that type of “human” interaction and preferred the filtered, headline-driven feed. I feel the same way about the NYT feeds. If I clicked through on any of the stories, it might be because a) I thought it was a joke or b) I couldn’t imagine why it would be of relevance to NPR.

  • dougom

    Tweets by a living, breathing human being are more interesting than those by an automated system? Who’d’a thunk?


  • Nathan Reimer

    Hopefully you’ll be able to add more staff soon. Keep up the great work!

  • Christy Robinson

    Human is how we do it at the Dallas Morning News. I hear the staffing issue loud and clear, however. Twitter is a demanding platform, mainly because of the opportunity for frequency coupled with a daily’s volume of content. We have multiple pairs of hands across teams making our Twitter strategy work (and work cohesively). Good luck to Melody as she works out a strategy that fits NPR’s needs.

  • Marco Toledo Bastos

    We actually run this test on a much larger scale. We tested the engagement levels in RSS and human written tweets on 40 news outlets over 5 countries. The results confirm Katz account. Here’s an excerpt:

    We found a significant correlation between news articles relayed by automatic feeds and news articles that failed to engage the audiences at the national level and with respect to each news outlet. We performed a statistical correlation analysis on the two variables and found that the proportion of singleton have a significant correlation with the proportion of tweets relayed by automatic feeds (r = .78, p < .001).

    The study is available here:

  • haddadme

    I wonder to what extent the success of this experiment came down to writing better headlines for social media as opposed to the actual feed automation process. Someone has to write these “automated” headlines, why not get them to tailor specific messages for social media?

  • Jason Keller

    I’ve worked with global brands on social media and I can say that indeed most do pay attention to the follower/unfollower ratio as it relates to tweet volume. The challenge is that there are a lot of variables that go into it, so odds are their ideal mark will never align with what you are hoping to receive.

    That said, there are solution for you to consider to unclog your feed. I am a huge advocate of using a tool like TweetDeck to unclog your feed. You could utilize Twitter lists to place all news/media outlets in one feed while not obstructing your main feed – if you don’t want to build this, I’ve already created a couple of lists like that, just follow mine.

    Regardless, great comment as I know every social media marketer is looking for feedback on things like this!

  • Jason Keller

    Mel I admire your job so much! I know it is a challenge given you are so shorthanded, but the volume of excellent content you have to work with seems like a good problem to have!

    I hope you don’t limit the stream of content you produce via @NPRnews – regardless of the feedback above. However, an interesting model could be something like NPRnewsLite – a channel that limits content to most read stories and summaries – I don’t think I’ve seen another outlet test a model like this.

    Also, for the bandwidth problem, have you thought about have the writers of articles provide you with draft tweets? That way your team becomes more of a content syndication organizer than a content production engine. …just food for thought, I’m making sure I’m following you on Twitter if you want to chat more. :)

  • Girl#2

    Have volunteers run the twitter feed. Train them to maintain quality. Maybe give the opportunity as a gift with a certain dollar pledge.

  • Girl#2

    Another thought, have local journalism or communication majors run it.

  • Courtney Shelton Hunt

    I felt the same way, Ashley – maybe even more negatively. My main Twitter newsfeed account only includes @nprnews, @NYTimes, and @Economist. I found the “human” tweets silly and bordering on the unprofessional. It’s not what I’m looking for from @nprnews, and I seriously contemplated unfollowing the account. Along those same lines, the “must listen” (or whatever they’re called) music-oriented tweets have always bugged me too. I’m pretty sure they don’t qualify as news.

  • Jen Westpfahl

    All St. Paul Pioneer Press tweets are non-automated as well. Every web producer on staff is involved. One possible complication when many people are involved is an inconsistent voice. We’ve avoided that by specifically discussing what are voice will be and what kind of things we will and won’t say.