How The Economist’s conversational tweets drive clicks, while Al Jazeera’s automation drives retweets

The conventional wisdom among digital journalists is that tweeting is better done by humans than robots – an argument that would cause @nytbot5000, the robot that automates Twitter feeds for The New York Times, to break down and cry. Luckily, he doesn’t have feelings.

The thinking is that people scanning Twitter are looking for something more than a pithy headline that includes a bunch of keywords and drops every “a,” “an,” and “the.” They like humor. They want their curiosity to be piqued.

A new study by SocialFlow substantiates and tests these beliefs, indicating that conversational tweets can spur clicks but headline-oriented tweets can be effective in sharing information regardless of whether users click on a link.

For part of the study, SocialFlow looked at clickthroughs and retweets for four media brands: The New York Times, Fox News, The Economist and Al Jazeera English. (Other parts of the study also looked at CNN and BBC News, but bit.ly data wasn’t available for those accounts.)

The report doesn’t address whether engagement is affected by the type of tweet: auto-generated headlines vs. handwritten. In fact, Gilad Lotan, the company’s vice president of research and development, didn’t know to what extent the accounts were automated.

But after looking at the figures and reviewing tweets for several of these organizations, I see a correlation between the type of tweet and how people respond. My conclusion: Yes, people seem to click more links in conversational, hand-crafted tweets. (Someone get @nytbot5000 a box of tissues.) But auto-generated headline tweets have their role, too, generating substantial retweets.

Two things support the idea that people click on conversational tweets:

  • @TheEconomist, which hand-crafts all of its tweets, had the highest engagement level of the four sites – much higher than @nytimes, which has three times as many followers.
  • When @nytimes was almost all automated, it registered fewer clickthroughs than when staffers manually tweeted during the day and left things to @nytbot5000 at night.

The evidence that people retweet headline-style links is Al Jazeera’s @AJEnglish. This account was mostly automated during the study period, but it registered the highest retweet rate.

Hand-crafted tweets encourage clicking on links

When I looked at tweets from @TheEconomist, I could tell immediately that they were handwritten, with a variety of approaches – sometimes incorporating a headline, but not always – and frequent use of hashtags.

Dave Humber, social media/marketing manager for The Economist Online, told me via email the thinking behind those tweets:

Our goal is to create compelling tweets that push the audience to click through the links to our material. Due to character limitations, however, we do not always include headlines from articles in our tweets. Regardless of whether or not we have a headline from our article though, we always include enough information within each tweet to ensure that the audience understands what that particular piece being mentioned is about. …

Further, we always try to increase the potential for sharing Economist content by ensuring that posts continue to make sense when re-tweeted outside of The Economist main Twitter stream. We also include hashtags when there is a specific and relevant trending topic directly related to our content.

The strategy appears to be working, according to the SocialFlow study:

Even though the size of The Economist’s audience is less than a third of the New York Times’, it is generating hundreds more clicks per shared link. This high proportion of clicks per tweet is suggestive of a high level of attention paid to the content shared by The Economist from its audience.

The study noted that The Economist’s median clickthrough rate was nearly as high as the average, indicating that the figure wasn’t thrown off by a few breakout tweets. (The Economist was a SocialFlow client during the study period.)

This SocialFlow chart shows that The Economist and Fox News, which handcraft their tweets, had the highest clickthrough rates. The Times and Al Jazeera English rely mostly on automation.

The New York Times’ “Human Tweet Week”

In May, the Times decided to try manual tweeting during the workday, dubbed “Human Tweet Week” by the Times’ Interactive News team. Social Media Editor Liz Heron said at the time that the experiment was meant to combat the perception that the account is entirely automated, when in reality some hand-crafted tweets are mixed in.

It’s “about changing the perception,” she said, “and it’s about being a little more strategic about what we put out there — finding the most engaging content.”

As it turned out, one of the two days that SocialFlow sampled fell during that week, providing something close to a controlled experiment on the impact of hand-crafted vs. automated tweets. (It’s not perfect, because the feed was automated at night.)

SocialFlow didn’t break out the figures for each day in the study, but at my request Lotan did some quick calculations for clickthroughs and shared them with me. (He didn’t immediately have figures for retweets.)

More people clicked on the links when humans were crafting the tweets.

Date Average CTs per tweet Median CTs
May 18: 971 631
May 25: 1,818 767
Increase: 87% 22%
Note: The number of tweets each day with links was the same, about 60.

“That’s really interesting, and quite a difference,” Pilhofer told me via email when I sent him the figures.

The Times is just now getting some data back on the experiment – nothing conclusive yet. “But if we see that big a difference between human/machine in our data too, then I think you are onto something,” he told me. “Hard to judge just based on one day, since there are so many lurking variables.”

Lotan told me he’s not so sure that handwriting was key, and unlike Pilhofer, he didn’t think the difference between the two median figures was that significant.

The study also noted that Fox News is doing well in clicks. (AllThingsD wrote about which keywords drove the most traffic to the Fox site, compared to the Times.)

A representative of Fox News wasn’t available in time to talk for this story, but that account also appears to be hand-crafted. The tweets that I reviewed often were similar to the headlines, but they used conversational language:

Headline-oriented tweets spur retweeting

When it came to clickthroughs, Al Jazeera’s @AJEnglish was the standout, with a retweet rate of more than twice the other media outlets.

At the time, its feed was handwritten about 5 to 20 percent of the time (now it’s 15 to 30 percent); the rest was automated, according to Riyaad Minty, head of social media for Al Jazeera Networks. (Al-Jazeera is now a SocialFlow client.)

Minty told me by email that the study shows a high level of trust on the part of @AJEnglish’s followers:

People who follow us trust our editorial values and our brand enough to share our links with their friends … Building and maintaining this trust is key to us; if it’s automated or handwritten, it does not make a difference if your audience does not trust you. At the end of the day, they want the news, and they want to be able to share it, and it’s our job to get it out to them in the most effective way.

The retweet rate for Al Jazeera English, which is mostly an automated feed, was far greater than any of the other accounts, according to this SocialFlow chart.

SocialFlow didn’t include @CNN in this part of the study, but I looked at its tweets. Most of those tweets mimic the headlines of the stories they  link to, but the account has been “human curated/managed” for years, according to Victor Hernandez, who recently became responsible for it. In part this is because humans can respond more quickly than a bot.

When I asked him about why CNN staff mimics headline language in most tweets, he told me:

Ideally there’s a healthy mix of the conversational tweets and the short news bursts with a link. Brevity is the key on Twitter ~ no secret there, but we’ve heard and seen much evidence that followers prefer to Retweet from this account than most others.

Like the Times, Al Jazeera and CNN are experimenting with the right mix of automated and hand-crafted tweets. Hernandez said he’s started to explore how @CNN can “increase the one-to-one interactions” with users, but he also said he’s conscious of the social behaviors that the account’s 2.4 million users have come to expect.

If indeed headline-oriented tweets – regardless of whether they’re automated – are retweeted more and clicked less, that would be because the tweets are sufficient in themselves. Which, if you think about it, is the point of a headline: to tell you very quickly what the story is about.

What are you trying to get your Twitter followers to do?

Lotan said one of his goals with the study was to spur media companies to think not just about “engagement” in general, but what they’re trying to achieve. The study doesn’t say whether clickthroughs or retweets are better. One helps site traffic among people who have expressed an interest in your content; the other helps build brand, puts content in front of new people and can bring clicks from outside your typical audience.

“If you’re a social data person at one of these media outlets,” he said, “how do you come to your boss and say, ‘I know we’re not bringing in clicks in traffic, but hey, we’ve got this really awesome, engaged audience on Twitter.’?”

“How is that OK? There’s no easy answer there, but I think it’s a really interesting problem that we have to solve.”

We have made it easy to comment on posts, however we require civility and encourage full names to that end (first initial, last name is OK). Please read our guidelines here before commenting.

  • http://termpaperwriter.org/ custom paper

    Generally I do not post on blogs, but I would like to say that this post really forced me to do so, Excellent post!

  • http://aheram.com Jayel Aheram

    I have kind of raised the point before, but the New York Times craft their tweets like they do their headlines. That is, they capitalize every word.

    “U.S. to Sing Songs About Birds [link]” versus “U.S. to sing songs about birds [link]” and when they do that, it makes it very hard to read their tweets when do show up in your Twitter timeline. You end up having slowing down to read each one and quickly get bogged down with the number of tweets they push out. When I used to follow them, I end up just skipping most of their tweets. Sure, I follow them, but they received zero engagement from me.

    I have a feeling that this is a major factor in the level of engagement they receive.

  • http://pulse.yahoo.com/_HRFHFJ7PVR2WCOHZRIWYSN2BHI Mitzi Patton

    nice

  • http://ultimateafricanmango.posterous.com/ultimate-african-mango-review-get-free-trial UltimateAfricanMango

    I am lucky because like your first answerer my wife and I are retired, so have money in the bank and the interest rates are going up….I drive very little now, and sold our house at the top of the market and have no mortgage on our present home.http://ultimateafricanmango.posterous.com/ultimate-african-mango-review-get-free-trial

  • http://ultimateafricanmango.posterous.com/ultimate-african-mango-review-get-free-trial UltimateAfricanMango

    I am lucky because like your first answerer my wife and I are retired, so have money in the bank and the interest rates are going up….I drive very little now, and sold our house at the top of the market and have no mortgage on our present home.http://ultimateafricanmango.posterous.com/ultimate-african-mango-review-get-free-trial

  • http://twitter.com/hummingbird604 Raul Pacheco

    I think you’re confusing correlation with causation. I’m not 100% sure but I could almost hypothesize that it’s Al Jazeera’s recognition of their groundbreaking stories and leadership in the online media world that drives the retweets. This means, Twitter users believe Al Jazeera English. it’s a trustworthy source. I don’t think there is a way to ascertain a direct causal link between the style of the tweet (or automation) and driving clicks or retweets. They are certainly correlated but you can’t prove causation.

  • http://www.facebook.com/profile.php?id=516475121 Fray Close

    If it matters, Since AJE is tweeting clips that they are playing on tv, I don’t have to click those to retweet, I have already seen them. I click the article/blog links, read them then retweet. 
    I pass along what my followers would be interested in.. most of the time I am waiting for them to post a link to something I want to pass on.