The curious case of the Election Day follower exodus

Are large post-election drops in follower counts on X (formerly Twitter) the result of real people quitting or spambots being shut down? A bit of both, depending on what account you're looking at.

In the hours and days following the 2024 United States presidential election, a multitude of (mostly liberal) X users commented that they had rapidly lost hundreds or thousands of followers. Two popular hypotheses have emerged to explain this: some users have speculated that the follower losses are the result of people on the left side of the political spectrum leaving the platform in response to the election outcome, while others have put forth the proposition that fake accounts are being shut down en masse. As it turns out, there are two distinct phenomena at work, and both theories have some truth to them.

Data used in this analysis was obtained from SocialBlade; changes in follower count from November 4th through 7th were used, as some of the accounts studied are missing daily updates from the 5th and 6th.

in almost all cases, large pro-Harris accounts lost followers on or shortly after the night of the election, while pro-Trump accounts gained them

First, there is a notable shift in follower counts in the period from November 4th through 7th, 2024 for large USA political accounts at both ends of the spectrum. Prominent liberal accounts generally lost significant numbers of followers, while prominent conservative accounts gained them. The only notable exception in the above chart is @KamalaHarris, who gained roughly 81K followers during the period in question; other large liberal accounts, including @JoeBiden, lost thousands or tens of thousands. @realDonaldTrump gained over a million followers in the same timespan; other large right-wing accounts experienced smaller but consistently sizable growth. (A similar shift occurred when Elon Musk announced he was purchasing Twitter, now known as X.)

With the free X/Twitter API no longer available, it is difficult to be certain exactly what happened here, but with a combination of old follower lists, archives, and the current state of the platform, I was able to spot check examples of lost followers on the left side of the aisle, and almost all are self-deactivations rather than bans. (Others have reported similar results.) It’s possible that some of these are fake accounts that were voluntarily deleted by their operators, but there are also a number of demonstrably real people who simply quit the site. As far as the followers gained by prominent right-wing accounts, there unfortunately isn’t a large enough sample size of the new followers available to perform a useful analysis in the absence of the aforementioned API.

an exception to the pattern: right-wing account @DallasJusticeN1 lost over a thousand followers

Although none of the massive right-wing accounts studied lost significant numbers of followers, an interesting anomaly emerged when looking at slightly smaller accounts. @DallasJusticeN1, an X account that belongs to Texas right-wing group “Dallas Justice Now” and has a history of astroturfing, lost almost 1500 followers during the period studied. Past research has demonstrated that the majority of this account’s followers are fake, and that these fake followers all came from the same spam-for-hire network, bringing us to the second phenomenon in play.

SocialBlade shows that many accounts followed by a massive spam network lost large numbers of followers around the time of the election

X cracked down on at least one large spam/fake follower network at some point between midnight November 5th and 1 PM CST November 6th, 2024, when follower loss was first observed on the @DallasJusticeN1 account. Review of SocialBlade statistics for 30 other accounts followed in early 2023 by the same network as the @DallasJusticeN1 account shows that every single one lost between a few hundred and tens of thousands of followers between November 4th and November 7th, 2024. Spot checks of known fake followers from this network identified in prior analyses show that these accounts were suspended rather than self-deactivated, indicating that X took action against the network as a whole at some point on Election Day or the following morning.

The majority of the accounts followed by this spam network over the course of its existence have been cryptocurrency or porn-themed accounts rather than political accounts, so this particular ban is unlikely to have contributed to most of the recent losses seen by (mostly liberal) political influencers. That being said, there are exceptions, such as the aforementioned “Dallas Justice Now”, and there are likely other such cases here and there as well.

these lost followers were suspended rather than voluntarily deactivated