President Trump, your problem isn’t bias by Twitter. It’s that you tweet too much.


President Trump held a social media summit in the East Room at the White House on Thursday. (Jabin Botsford/The Washington Post)
July 12 at 11:57 AM

In retrospect, President Trump’s social-media sympathy session at the White House on Thursday was inevitable. It was essentially his social media existence made manifest, with his social media director Dan Scavino helping to gather a number of the people who power Trump’s, and to a larger extent Trumpworld’s, content and attitude online. Trump offering up IRL faves.

To welcome his Twitter crowd to the executive mansion, Trump gave one of his signature freewheeling speeches. The broad focus was his insistence that conservatives were being treated unfairly by social media companies, a line of argument that he’s gleaned from allies online and his son Donald Trump Jr.’s activism.

As Trump presents it, though, the assertion doesn’t make much sense. Here, for example, is how he says Twitter’s bias against him is manifested.

“I will tell you a lot about things that are happening,” he said. “People come up to me, ‘Sir, we want to follow you. They don’t let us on.’ And it was so different than it was even six, seven months ago. I was picking up unbelievable amounts of people, and I’m hotter now than I was then, okay?”

Trump’s made this claim before that people have approached him to claim that they aren’t allowed to follow him on Twitter. As CNN’s Daniel Dale has documented, Trump’s use of “sir” in his anecdotes is generally a tell that he’s making the story up. And in this case, he’s either making it up or dealing with remarkably inept people.

Not only does a search for “Trump Twitter” quickly bring you to Trump’s Twitter page, where there’s a big ol’ button that says “Follow” in the corner, but Twitter actually recommends Trump to new users as someone to follow.

I created an account this morning and gave it no information about my interests. On a list of recommended accounts to follow, Trump was listed seventh.


(Philip Bump/The Washington Post)

He also complained about losing followers, a complaint he’s made before. Twitter regularly purges fake or automated accounts, which can cause follower counts to drop, particularly among popular users. A big purge a year ago saw Trump lose 200,000 followers — a tenth of the number lost by former president Barack Obama (who’s more popular on the platform).

One thing Trump was particularly mad about on Thursday, though, was that his tweets aren’t doing as well as they used to.

“I noticed things happening when I put out something, a good one that people like,” he said. “Right? A good tweet, it goes up. It used to go up, it would say 7,000, 7,008, 7,017, 7,024, 7,032, 7,044, right? Now it goes 7,000, 7,008, 6,998. . . . I say, ‘What is going on? It never did that before.’ It goes up and then they take it down, then it goes up. I had never had that. Does anyone know what I’m talking about with this? I never had that before. I used to watch it. It would be like a rocket ship when I put out a beauty.”

It’s extremely unlikely that Trump has witnessed his retweet counts fluctuate up and down in the way he suggests. When he tweets, he gets a lot of retweets quickly, too many to observe with the naked eye. Is it possible that he’s had tweets retweeted a certain number of times and that, after a purge of bots, the figure has gone down? Yes. But in real-time, as a function of Twitter trying to mess with him? Nah.

There are a variety of reasons this doesn’t make sense, including that no one except Trump cares if he has 7,044 retweets instead of 6,998. This complaint, though, highlights how closely Trump tracks these metrics and that he feels as though things aren’t going as well as they used to be.

On that point, he’s not exactly wrong.

Trump is tweeting more than he used to. In the first half of 2017, he tweeted 937 times — counting only original tweets, and not retweets. In the first half of 2018, 1,276 times. This year? He’s tweeted 1,779 times. The peak was in May, when he tweeted 381 times, more than 12 times a day.


(Philip Bump/The Washington Post)

The number of retweets those tweets get, though, hasn’t gone up dramatically. In the first half of 2017, he got about 17.8 million retweets on his original tweets. (See the note on methodology below.) That’s an average of about 19,000 retweets per tweet. In the first half of 2018, 27.7 million retweets — but just a slightly higher average of 21,700 retweets per tweet. In the first half of this year, he got just over 40 million retweets, but the average per tweet was about 22,500.


(Philip Bump/The Washington Post)

That’s particularly weak because Trump’s gained so many followers over that time. On July 1, 2017, he had 33 million followers. A year later, he had 53.1 million. Now he has nearly 62 million.

This is what you might call diminishing returns.

There’s a correlation between the frequency of Trump’s tweets and the average number of retweets he gets. On the graphs below, which compare monthly tweet counts to retweet averages, the more sharply a trendline drops down and to the right, the less effective Trump’s tweets are at getting retweets.

In 2017, the line is fairly flat. In 2018, it dips downward — in months he tweeted more, he got relatively fewer average retweets. In 2019, the drop is much more significant. (The line on this graph excludes July, since data for the month aren’t yet complete.)


(Philip Bump/The Washington Post)

Let’s isolate the 2019 frame from that animation. Trump’s been tweeting more — more dots further to the right — but he’s seeing fewer retweets on average the more frequently he tweets.


(Philip Bump/The Washington Post)

In other words, if Trump’s concern is that his tweets lack the oomph they used to, the solution isn’t government intervention in Twitter’s business in order to ameliorate some unproven bias against the president.

It’s that Trump should maybe tweet less.

Note on methodology: The Twitter API, which allows data access to users’ tweets, only pulls 3,000 tweets at a time. I’ve been pulling Trump’s tweets regularly since he began his campaign, so some of the older tweets in this data set (for the period in light blue on the second chart) will be tallied as having fewer retweets than they actually do. The difference is minor, though; most retweets of a tweet happen shortly after it is published.

Leave a Reply