This summer The Atlantic wrote an article about Elon Musk ruining X, formerly known as Twitter. On July 3, they wrote: “Twitter may have just had its worst weekend ever, technically speaking.”  This was in response to a new change to the platform that forced users to login in order to view tweets - or what are they called now? Posts? There were also new restrictions on how many posts per day could be done by one user. 

For those of us who have been on the platform since 2007, this felt antithetical to some of the social network’s main benefits. But by this point, how much did it really matter to twist the knife that was already in? 

Here’s when I think the knife originally went in: Fall 2016. It was during the presidential debates and I was working at a company called Tracx, an enterprise social media listening company. We had direct API connections with Twitter and the CMO at the time asked me to create content she could use to demonstrate the power of the software. 

I was already tracking social media mentions from Facebook, Instagram, and Twitter for brands like Revlon, ebay, Jockey International and KraftHeinz - and not just mentions from all users, but the replies and comments, even replies to comments. One thing I noticed, across almost every brand, was the random insertion of “realdonaldtrump” and then the explosion of reactions afterward. 

This insertion made pulling data relevant to the brands nearly impossible since the exclusion capabilities were limited to just removing the original comment with the Trump mention and not the subsequent reactions that followed. And this was several months before the election.

(A couple of years later, Twitter had to testify before Congress that this problem spiraled out of control the following year. They detected and blocked 523,000 daily logins coming from automated systems instead of real users - that’s 6.4 million suspicious accounts globally per week.) 

The use of “realdonaldtrump” and tagging his handle became so widespread that I eventually stopped reporting on Twitter data for my clients. It was throwing off impression counts and sentiment scores for nearly every brand, which was a shame because we were learning a lot about their consumers on this platform, who were different from their consumers on other platforms like Facebook, which skewed more female. 

Twitter uses, at that time, were more likely to be male, blue collar, and in their 30’s and 40’s. There wasn’t another social media platform out there with those demographics reflected, so Twitter became a great research tool for us marketers looking to reach that audience, an audience that tended to watch NFL and NHL games loyally. 

Fall 2016 

When I sold our CMO on a report around the debates, she agreed. I set up the queries in our system and then I watched history happen. In fact, I remember getting a call late at night in October that I broke the bank for Tracx because my boolean search had greatly exceeded the number of mentions we were allowed to capture in a 24-hour period. The search term was “#electTrump.” 

A scientific report done in 2018 confirmed what I was watching unfold on my Tracx dashboards: The number of tweets about the NFL season were shrinking while the number of tweets - from these same blue collar workers who probably did not normally watch presidential debates - were on the rise. I knew it was a lot, but since I was limited in how many mentions I could pull in, I didn’t know how many, until this report came out a couple of years later stating that 9.6 million tweets had the Trump hashtag while only 3 million had the Hilary Clinton hashtag (which makes sense knowing the demographic makeup of the platform, but still, that’s a wide margin). 

Trump certainly knew the demographics of his key supporters. He even made up a story about the NFL trying to change their schedule around the debates, so there were no viewing conflicts. The NFL denied this claim. No such effort was ever made, but since I was tracking both the debates on social media and the NFL hashtags, I could see where the same audience had to choose between their beloved teams, team they probably wagered money on, and Trump as a Presidential hopeful. 

Our CMO never published my findings. In November 2016, I tried to tell her that Twitter was predicting the next President, but she didn’t want to believe me, or the data. To this day, I don’t blame her. Nobody else believed me at the time either. I wasn’t taking sides. I was reporting an unprecedented change in user behavior and use of a social media platform. 

In my opinion, that’s when Twitter broke. It was really too bad because it was a great way to follow journalists and PR professionals who had actual news and breaking headlines to share in real time. So if Musk is trying to break it now, he’s really just twisting the knife. The bots came for it like locusts years ago. 

“The situation has made conspiracy theorists out of onlookers who can’t help but wonder whether Musk’s plan has been to slowly and steadily destroy the platform all along.”  — The Atlantic