Anatomy of a Twitter thread: is long-form tweeting great for journalism?
Twitter is often linked to journalism for two reasons. Firstly, its real-time nature makes it a perfect place to discover breaking news. Secondly, well, it is full of journalists. However, Twitter has been trying for a while to go over real-time and quick conversations. One of its biggest steps towards this goal is Twitter thread, a function that allows users to publish a series of tweets at the same time.
Twitter has praised the news organisations that have been using threads in the most effective way: in a recent interview with WIRED, Twitter CEO Jack Dorsey complimented ProPublica for the way they thread together key parts of their articles (see an example here). Dorsey will always welcome any extensive use of his platform. However, I started wondering if Twitter threads are actually good for journalism or not. Specifically:
- How do readers engage with threads?
- Do people read from the beginning to the end of threads?
Anatomy of a Killing, the BBC Africa case
At the last Hacks/Hackers London, a monthly meetup of journalists (“hacks”) and technologists (“hackers”), the BBC Africa Eye journalists Daniel Adamson and Aliaume Leroy spoke about ‘Anatomy of a Killing’, a brilliant open-source investigation to verify a video that had gone viral among African social media users.
The whole investigation was published as a 10-minute video but BBC Africa also unpacked the investigation in 34 tweets.
If you haven’t seen it yet, it is worth it:
Daniel Adamson, Series Producer at BBC Africa Eye, focused on the format they chose to tell this story. He said the use of Twitter made a huge difference in the publication of this story:
Twitter is underrated as a visual storytelling platform. There is not another platform where we could have broken that investigation the way we did.
The structure of the thread, where each finding of the investigation occupied a single tweet, mirrored the structure of the investigation. A series of individual findings were linked together into a chain of visual evidence that the audience could follow towards a conclusion.
I think Twitter Thread is great for investigative journalism.
This piece of journalism had a great impact and helped highlight the potential of open source investigations. The result is visually impressive and highly engaging.
As attention-grabbing as it might look, we still need a bit more context to decide if a thread like this is effective for journalism.
How to find out the impact of a Twitter thread
BBC Africa disclosed the the number of impressions (how many times a tweet has been seen on Twitter) gained by the first and last tweets.
The first tweet was seen 15,049,504 times, but the impressions dropped by 90% to 1,031,305 in the last tweet.
Commenting on the impressions for tweets in the thread Leyla Najafli, senior social media producer at BBC Africa said:
We can assume that most people who looked at the second tweet looked all the way down to the end of the thread. From the second tweet onwards, the impressions numbers are quite consistent.
However, the number of impressions can be a misleading measure, as it shows just the amount of views and does not give back a clear picture of how readers related to that content. Often engagement is more revealing in terms of readers’ behaviour.
Measuring the engagement of a Twitter thread
The only report that focuses on the engagement of threads is by Daniel Grijalva, a student at Universidad de Sonora, Mexico. He looked into the engagement of 500 threads about different topics and published by a variety of accounts, collected through Thread Reader from October 2017 to May 2018.
What he found is that the engagement usually drops dramatically within the first two tweets of the thread:
Sounds familiar? I have compared the study with the BBC Africa thread to understand how it performed in terms of likes and retweets per single tweet:
The result is not different from Daniel Grijalva’s findings. The engagement trend dives between the first two tweets, with just a few spikes over a general plateau. The likes’ spikes look higher than retweets’ ones but that is just a problem of proportion. If we look at the percentage change of retweets and likes, it is easier to see a correlation.
It is not clear if likes boost retweets or vice versa. But the two types of engagement appear to be correlated. However, here is what happens when we compare the two trends with the replies one:
The rise of quality engagement
Replies are not the most used function on Twitter as they require people to put a bit of themselves into the interaction and write something on their own. It is way easier to hit the retweet button or, even better, to just like the tweet without committing too much.
Replies, instead, are a stronger and more valuable engagement (talking about a different platform, Facebook seems to agree with this statement, as its new algorithm favours comments and replies to comments).
The chart above is impressive because the increase in replies does not look linked to a rise of likes and retweets. That means those are replies to tweets that readers found genuinely interesting, without tons of likes to boost engagement and visibility of those tweets.
This is a strong evidence that people who didn’t stop at the first tweet, generally went on reading. More than a few reached the end of the thread. It is probably not possible to state that Twitter threads are great for journalism. However, they can definitely be a powerful tool to make people engage with good stories.
Originally published at www.interhacktives.com on November 7, 2018.