The short version: X feels more toxic than it used to because the algorithm was rebuilt to reward replies and conflict, the verification system now lets anyone buy influence for $8 a month, and the trust and safety team that used to catch the worst of it was gutted after 2022. None of that means your brand has to leave. It means you need a written protocol for how you respond, when you stay quiet, and when you escalate, instead of making it up in the moment while your notifications fill with strangers.
What changed on the platform
I have run brand accounts on Twitter since 2009. I watched it go from a place where a well-timed reply to a journalist could land you a client, to a place where the same reply now gets quote-tweeted by an account with an egg avatar and forty followers, and somehow that gets more traction than your original post.
The turning point most people point to is Elon Musk’s takeover in October 2022, and they are not wrong, but the specific mechanics matter more than the ownership change. In November 2022, roughly half of the company’s 7,500 staff were laid off in a single round, and trust and safety was hit especially hard, with reporting at the time putting the cuts to that team well above the company average. Fewer people moderating means more of the borderline stuff, the pile-ons, the coordinated dogpiles, the bot replies, simply stays up longer before anyone looks at it.
One antidote to the noise is a profile that leads with proof: here is how brands pin milestones at the top of their timeline.
At the same time, verification stopped meaning “this is confirmed to be who they say they are” and started meaning “this person pays $8 a month.” That single change removed a trust signal that had existed since 2009. You used to be able to glance at a blue tick and assume a baseline of accountability. Now a blue tick just tells you someone has a subscription, and some of the loudest, angriest accounts on the platform are paid subscribers whose replies get algorithmic priority specifically because they are paying.
The algorithm rewards conflict, on purpose
This is the bit most brand guides skip. X’s recommendation system, parts of which were made public when the company open sourced elements of its algorithm in 2023, weighs replies and reply chains heavily, far more heavily than likes. A post that gets 500 likes and no replies will often be outranked in the “For You” feed by a post with 40 likes and 60 replies, because replies signal “conversation,” and conversation is what keeps people scrolling.
The problem is that agreement does not generate replies the way anger does. Nobody quote-tweets a post to say “yes, I agree completely.” They reply to correct you, mock you, or pile on. So the system that decides what gets seen is mathematically biased toward content that provokes a reaction, and the cheapest, fastest reaction to provoke is annoyance. Your brand does not need to say anything controversial for this to bite you. A perfectly ordinary tweet about a product update can get pulled into a reply thread that has nothing to do with your business, because someone used it as a jumping off point for an unrelated argument.
A story: the “boring” tweet that wasn’t
A few years ago I posted a fairly dry observation about email open rates dropping across the industry. It was the kind of tweet I have posted a hundred times, aimed at other marketers, no opinion in it worth fighting over. Within about ninety minutes it had over 300 replies. I went through them out of curiosity later that evening, and I would guess maybe fifteen were about email marketing. The rest were people arguing with each other about something entirely unrelated that had spun off from one reply near the top of the thread, plus a handful of accounts using my notifications as free advertising for their own newsletter, plus two or three aggressive replies calling the whole post “cope” for reasons I still cannot explain.
Nothing bad happened to my business because of it. But I remember closing the app that night thinking: this is what brands are walking into every single time they post something with even mild engagement. It is not that Twitter hates your brand specifically. It is that the platform has built a machine that turns any moderately visible post into a small arena, and your brand’s tweet is just the stage, not the show.
The uncomfortable part nobody wants to say
Here is the bit that gets left out of most “how to survive Twitter as a brand” advice: a good chunk of the toxicity brands complain about is not aimed at them at all, it is aimed at each other, and the brand just happens to be standing in the blast radius because they posted at the wrong moment. Brands that pull the “we’re leaving X because it’s toxic” card are sometimes telling the truth, and sometimes they are using the platform’s genuine problems as convenient cover for the fact that their social team panicked over one bad afternoon and wanted an easy exit story.
The other uncomfortable truth is that some brand accounts make it worse for themselves by trying to be “relatable” or “funny” in a way that reads as bait. If your brand voice is deliberately spicy to chase engagement, you cannot then act shocked when the replies get spicy back. The algorithm did not create that dynamic out of nothing, it just amplifies whatever tone you already brought to the table. I have had clients ask me to write “edgier” tweets to compete with brands like Wendy’s or Ryanair’s social team, and I have talked most of them out of it, because that voice only works if you also have the appetite, the legal cover, and the community team to handle what it invites back.
A protocol for managing your brand on X without losing sleep
I give every client the same basic structure now, adjusted for size. It is not complicated, and that is the point. Complicated plans fall apart in the moment a real pile-on starts.
- 1. Separate “criticism of the product” from “noise about anything else.” Write this distinction down as a rule for whoever runs the account. A complaint about your delivery times needs a reply. A stranger arguing about politics under your tweet does not.
- 2. Set a reply window, not a reply rule. I tell clients to check brand mentions three times a day, at set times, rather than leaving notifications on all day. Real-time monitoring for anger creates real-time anxiety, and it produces worse replies because you are answering from a jumpy place instead of a considered one.
- 3. Draft your three worst-case replies in advance. Before anything happens, write the reply you would give to a customer complaint, a bad-faith troll, and a genuine PR problem. Having the tone already worked out means you are not composing under pressure while forty people are watching your account.
- 4. Use mute and block without guilt. Muting a reply thread does not delete it, it just stops it feeding your own anxiety, and it stops your engagement metrics from being warped by a conversation you were never going to win. Blocking bad-faith accounts is not censorship, it is basic house-keeping.
- 5. Never argue in public with someone who has nothing to lose. An anonymous account with 200 followers has no reputation on the line in a public argument with your brand. You have everything to lose. If a reply needs correcting, correct the facts once, calmly, and then stop engaging with that specific account.
- 6. Decide your escalation line before you need it. Agree in writing what counts as “get a human decision-maker involved” versus “the community manager handles this alone.” A single insulting reply does not need the CEO. A coordinated pile-on with hundreds of accounts, or anything that touches a legal or safety issue, does.
None of this stops toxicity existing on the platform. It stops your team making it worse through improvisation, and it stops one bad hour turning into a week of bad decisions.
Should your brand even stay on X
Plenty of well-known brands scaled back after 2022, and several major advertisers paused spending in 2023 after reporting from groups including the Center for Countering Digital Hate flagged ads appearing next to extremist content. That is a real business risk worth weighing, particularly around ad placement rather than organic posting.
My honest advice for most small and mid-sized brands is not to leave, it is to stop treating X as your main platform. Post there for reach, customer service replies, and industry conversation, but build your actual community somewhere the algorithm is not structurally biased toward conflict. If you are already investing energy into a visual platform, it is worth doing that rather than half-heartedly, and my breakdown of the Instagram tools I use is a decent starting point if you want a channel that still rewards consistent, calm content instead of arguments.
I also tell clients this bluntly: if managing your brand’s presence on X is eating more hours than it is worth in leads or reputation protection, that is a sign to shrink your ambitions there, not a sign you have failed. Posting twice a week and replying to genuine customers is a perfectly good X strategy in 2026. You do not owe the platform a daily fight.
Frequently asked questions
Why does Twitter feel more toxic now than it did a few years ago?
Two things changed at once: staff cuts after 2022 reduced the trust and safety team that used to catch abusive or coordinated behaviour, and the algorithm was rebuilt to weigh replies more heavily than likes, which mathematically favours conflict-driven content over agreement.
Should a small business leave X entirely because of the toxicity?
Not usually. Leaving entirely means giving up reach and customer service visibility. A better move for most brands is to reduce posting frequency, avoid deliberately provocative content, and put more energy into a platform where the algorithm does not reward arguments as heavily.
How should a brand handle a pile-on or coordinated attack?
Stop replying to individual accounts in the thread, mute the notification storm so it does not affect your judgement, correct any factual errors once in a single clear statement, and escalate to a decision-maker only if the volume or content crosses into genuine safety or legal territory.
Is X Premium (the paid blue check) worth it for a brand account?
It gives your replies more algorithmic reach, which can help visibility, but it also puts you in the same trust category as anonymous paid accounts, since the checkmark no longer signals verified identity the way it did before 2022. Weigh the reach benefit against that credibility trade-off for your specific audience.
