Barack Obama may have been the original digital-first president, but Donald J. Trump is likely to leave a legacy as the first tweeter-in-chief. His embrace of the social platform is already the stuff of legend, and his frequency and reach have played no small part in defining his personal brand and spreading his populist message.
Since taking the oath of office, however, change has crept into the president’s Twitter account. After spiking two weeks into his term, total engagement has steadily declined during his first 100 days, even while his number of followers has surged by 40%, from 20 million to 28 million. Part of the explanation comes from tweet frequency. Early in his term, Trump sent out more than 40 tweets per week, very close to the frequency on his account in the weeks leading up to the inauguration. However, after a month in office, his activity quieted somewhat, which would naturally contribute to fewer total engagements.
To explore this phenomenon, Huge DC exported all tweets from the @realDonaldTrump account since the beginning of his presidency, from January 20th to date, and appended data to each. These data codes delineate date, time of day, and type of engagement (reply, retweet, like). We then manually scored each tweet across a range of attributes—voice, tone, objective of the tweet, at whom it was aimed, policy content—and overlaid external data, including the president’s approval rating at the time of the tweet, and his location at the time of the tweet (the White House vs. Mar-a-Lago, for instance).
Tweeting less frequently is only part of the explanation for diminished engagement. In addition to fewer tweets contributing to engagement totals, the president’s average engagement per tweet has dropped sharply—a whopping 66% decline in just three months, from more than 206,000 engagements per tweet at the outset of his presidency to fewer than 71,000 more recently.
Total engagement is measured by replies, retweets, and likes. Replies and retweets are a mix of positive and negative sentiment, but likes are typically a purer indication of support. One hypothesis for the engagement erosion is that opponents of the president or his policies may be flagging in their opposition efforts and engaging less frequently. The data suggests otherwise, however. At the beginning of the term, 77% of engagements were likes, or clear signs of support. This metric has declined as well, down to 64% in the most recent week. This means that the drop in engagement per tweet is attributable at least in part to fewer people showing support with a like. The total number of likes per tweet has also fallen sharply—a 72% drop since the president took office.
What could be responsible for more than two-thirds of the president’s most engaged Twitter supporters checking out? The answer may be found in an analysis of the tweets themselves, and in particular the voice (or, voices) in which they are written.
In our analysis of the tweets, we scored each according to the type of voice used. Out of the 476 tweets analyzed, 34% exhibit what we called an “agitated” voice. This voice is characterized by an emphasis on certain words using all caps (i.e. “FAKE NEWS”), exclamation points, threading thoughts into multiple tweets, and improper grammar and punctuation. This is the voice that most authentically resembles the president when he is speaking without prepared remarks. It is strident, defiant, and occasionally aggressive.
The greatest concentration (41%) of tweets we defined as “calm.” These tweets have less emphasis and aggression and are typically used for announcements (“Next Saturday night I will be holding a BIG rally in Pennsylvania. Look forward to it!”) or observations (“Very interesting election currently taking place in France.”).
We marked an additional 20% of tweets as “prepared.” These share many of the same characteristics as “calm” tweets (so much so that 20% is conservative—many “calm” tweets could actually be “prepared,” as well). But prepared tweets appear to have been written by someone other than the president, likely his communications staff, owing to their use of social media marketing conventions such as hashtags, emojis and “@” mentions. Their content is also more “business as usual,” featuring congratulations and thank you messages. The final 5% of tweets are retweets.
This distribution of tweets by voice type demonstrates consistent trends over the first 100 days. After spiking at 44% of all tweets in February, the number of “agitated” tweets has dropped to 30% in March and 26% of all tweets in April. “Prepared” tweets show an inverse trend, rising steadily from 15% to their peak of 27% in April—the first point at which they eclipsed the hallmark “agitated” tweets.
A new editorial strategy—that is, finding a new way to talk to users—is not uncommon when an organization or brand reaches a new stage in its development or tries to appeal to a broader user base. But that is not what appears to be happening with the president’s Twitter. Analysis of day of the week, time of the day, and location of the president compared with type of tweet voice reveals a lot. Tweets in the president’s trademark strident and “agitated” voice occur much more frequently on weekends when he is at Mar-a-Lago, and in the early morning during the week. Similarly, more than 90% of “prepared” tweets are posted Monday through Friday.
A vacillating voice by day of the week is not suggestive of a new strategy being phased in, but of two competing editorial strategies for the same account. During business hours, the president’s Twitter account sounds more like an official White House account. During the early morning and on weekends when the president is left to his own devices, it reverts to a personal account. Are we watching a tug of war over the voice of the president’s Twitter? The data indicates this is a possibility.
The plummeting engagement data also sheds some light on this question. Typically, a new editorial strategy is designed to lift engagement, not suppress it. And yet, the president’s Twitter was firing on all cylinders before the introduction of the prepared tweets, and this new strategy is causing it to sputter. Surely some part of this new strategy has gone wrong. Or has it?
Accounts from within the White House have claimed that staffers are trying to curb both the president’s use of Twitter and its influence. But a strategy that disregards a user’s preferred activity is designed to fail. Despite the flagging metrics, the president seems uninterested in abandoning Twitter anytime soon. A more strategic approach, then, is to set tactics aside for a moment and focus on outcomes. If curtailing the president’s use of Twitter is not an option, what approach would at least mitigate its impact by diverting attention away from it?
In social media, nothing alienates a core audience faster than an inauthentic voice. Watering down the president’s Twitter account with static and predictable prepared tweets has chipped away at its authenticity, giving its fans and followers license to look away. In effect, it turns down the volume on his Twitter account and saves staffers from potential damage control—at least during business hours.
Whether this altered timbre and its striking suppression of engagement is evidence of an internal power struggle or simply the growing pains of a brand in transition, the outcome is the same: the power of the president’s Twitter account to directly reach and engage his fans and other followers is waning.