Your engagement bait is worthless on X now
If your X strategy includes telling people to engage so the algorithm sees your post, you have been building on a foundation that X just publicly confirmed it ignores.
Affiliate Partner|Viktor.com: Viktor is the AI employee that lives in Slack and MS Teams, connects to 3,200+ tools, and does the actual work: reports, dashboards, code, campaigns. Start free.
Nikita Bier, head of product at X, posted something yesterday that most creators scrolled past without fully understanding what it meant for their content strategy.
Replying to a creator who had noticed something unusual about their post performance, Bier confirmed it directly: "We do not count impressions from posts that attempt to manipulate the algorithm, for example by instructing users to engage."
One sentence. Quietly posted. Enormous implications for every creator who has been using engagement bait as part of their growth strategy on X.
Here is what this actually means, which specific tactics are now pointless, and what X is rewarding instead.
What Bier actually said and why it matters
The reply was short and unambiguous. X does not count impressions from posts that attempt to manipulate the algorithm by instructing users to engage. He also noted in the same reply that a bug with photo rendering had been patched, suggesting the context was a creator asking why a post designed to drive engagement had underperformed in their analytics.
The key phrase is "attempt to manipulate the algorithm by instructing users to engage." That is precise language that points to a very specific category of content that has been widespread on X for years: posts that explicitly tell their audience what action to take in order to boost the post's algorithmic performance.
You know exactly what these posts look like. "Like this if you agree." "Repost if this resonates." "Reply with your thoughts below to help spread this." "Press and hold to charge this post." "Drop a reply so more people see this." These are not calls to action designed to start a genuine conversation. They are instructions designed to generate the engagement signals that trigger algorithmic amplification. And X is now saying explicitly that those impressions do not count.
This is not a new policy. It is a public confirmation that a long-standing platform position is being actively enforced. The significance of Bier saying it openly is that creators who were still using these tactics, or who were unsure whether they were being penalized for using them, now have a direct answer from the head of product himself.
The context: a year of X closing every gaming loophole
To understand why this confirmation matters, it helps to see it as part of a pattern rather than an isolated announcement.
X has spent the past twelve months systematically identifying and closing every mechanism creators were using to game their way to impressions and revenue without producing genuinely valuable content.
Earlier this year, X confirmed that replies do not count for revenue sharing purposes. The platform identified that reply rings, coordinated groups of accounts replying to each other's posts to trigger algorithmic amplification, were generating fraudulent impressions and directing revenue share payments toward accounts that had manufactured their metrics rather than earned them. Removing replies from the revenue calculation closed that loophole directly.
Then came the engagement solicitation enforcement. Bier announced that directly asking for engagement, using phrases like "I'll follow everyone who replies," three or more times would result in removal from the revenue share program and forwarding of the account to the policy team for suspension. Nearly 4,000 accounts were removed in a single enforcement cycle.
The duplicate content detection was upgraded to catch stolen posts at three times the rate of the previous model. Watermarks, edited intros, and other minor modifications that previously allowed aggregators to steal content and collect its impressions were no longer sufficient camouflage. 1.5 million stolen posts were detected in a single review cycle.
AI written content was confirmed as grounds for account restriction, with Bier's direct reply to the AJ Investment Research account establishing publicly that X's detection systems are operating at the account level.
And now this: impressions from posts that instruct users to engage do not count. Every significant tactic that creators have used to manufacture algorithmic performance on X without earning it organically has now been explicitly addressed, and in most cases actively enforced.
The picture that emerges from connecting these dots is not a platform making random policy decisions. It is a platform that has identified every major gaming mechanism and is methodically closing each one in turn, while simultaneously building the detection infrastructure to enforce those closures at scale.
Which specific tactics are now worthless on X
Given the confirmation that manipulation-instruction impressions do not count, it is worth being specific about which behaviors fall into that category so there is no ambiguity about what needs to change in your content strategy.
Any post that explicitly instructs your audience to perform an engagement action for the purpose of boosting the post's performance is in this category. This includes asking people to like, repost, reply, or bookmark specifically so the post reaches more people. It includes the "press and hold" trend that spread through X earlier this year, where posts instructed users to press and hold on the post to "charge" it, generating a stream of replies in the process. It includes any variation of "engage with this so the algorithm shows it to more people" regardless of how creatively that instruction is phrased.
It also likely includes more subtle versions of the same tactic. A post that says "drop a word in the comments" or "reply with one emoji" with no genuine conversational purpose, designed purely to generate reply volume, falls into the spirit of what X is targeting even if it does not use explicit algorithmic language.
What does not fall into this category is a genuine call to engage based on the substance of what you posted. Asking your audience a real question you want their perspective on is not engagement manipulation. It is conversation. Ending a post with "what do you think?" when you are genuinely interested in the answer is not the same as saying "reply below so the algorithm shows this to more people." The distinction X is drawing is between posts that invite genuine participation and posts that instruct mechanical action for algorithmic purposes.
If the engagement you are asking for would be meaningless to you without the algorithmic benefit, it is manipulation. If the engagement would be valuable to you as genuine audience input regardless of its algorithmic effect, it is conversation. That distinction is where the line sits.
What X is actually rewarding instead
Understanding what is no longer working is only useful if it leads to understanding what does work. And X has been surprisingly transparent about what it is building its algorithm around.
Dwell time is now one of the platform's primary ranking signals. The amount of time a user spends on a piece of content, scrolling through the replies, watching a video, lingering on a thread, is weighted heavily in the algorithm's evaluation of content quality. This rewards content that is genuinely interesting or useful enough to hold attention, not content that generates a quick mechanical interaction and gets scrolled past.
Shares via direct message are the highest-value engagement signal on the platform right now. When someone sends your content to another person directly, they are endorsing it to a specific individual rather than passively double-tapping as they scroll. That action requires genuine conviction about the value of the content and is almost impossible to manufacture artificially. X's algorithm recognizes this and weights it accordingly.
Genuine replies that generate conversation threads are rewarded over reply volume alone. A post that generates ten replies that themselves generate replies, creating a real conversation, performs better algorithmically than a post that generates fifty identical one-word replies from people following an instruction. The depth of the engagement matters more than the volume.
Original video content, particularly talking head videos where a creator looks at the camera and speaks in their own voice, is being actively pushed by the platform. Bier posted explicitly earlier this year that talking videos are an easy way to get millions of impressions even with basically no followers. This is the format X wants to see more of, and it is the format its algorithm is currently most aggressively rewarding.
And genuine mutual connections, the bidirectional follow relationships between accounts that actually read each other's content, remain the most reliable distribution engine on the platform. The mutual algorithm update confirmed that these relationships are now formally weighted in how content is distributed. Building a real network of people in your niche who follow you because they want to read what you say is more valuable algorithmically than a large following of passive accounts who followed you because you asked them to.
What this means for the Iron Slopdome
Earlier this year, Bier coined a term that deserves more attention than it received: the Iron Slopdome. He used it to describe X's multi-layered defense system against low-quality AI-generated and algorithmically manipulated content flooding the platform.
The term is useful because it frames what X is doing not as a series of individual policy decisions but as an integrated system. Each enforcement action, the reply exclusion, the engagement solicitation removal, the duplicate content detection, the AI written content restriction, the impression exclusion for manipulation posts, is a layer of the same defensive structure. Every layer closes a specific vector through which low-quality content was finding its way into the feed and extracting value from the platform's engagement and monetization systems.
The Iron Slopdome is, in essence, a bet that X can engineer the conditions that force genuine quality to the surface. If every mechanism for gaming the algorithm is closed, the only way to generate real impressions and real revenue is to produce content that people genuinely want to see, share, and engage with authentically.
For creators who have been building on genuine quality all along, this system is working in your favor. Every layer of the Iron Slopdome removes accounts and content that were competing with yours for attention and revenue without deserving either. The enforcement environment that feels punishing if you have been gaming the system is protective if you have not been.
The practical audit every creator on X should do right now
Given the confirmation that manipulation-instruction impressions do not count, the immediate practical question is whether any current element of your X content strategy falls into that category.
Go through your last thirty posts and identify every post that included any instruction to engage. Not every call to engage, the distinction between those two things matters, but specifically every post where the instruction was designed to generate engagement for its own algorithmic benefit rather than to start a genuine conversation.
For each post you identify, ask yourself whether the engagement you were asking for would have had any value to you without the algorithmic effect. If you were asking people to reply purely because replies boost distribution, that is manipulation. If you were asking people to share a perspective because you genuinely wanted to hear it and the content prompted a real question, that is conversation.
For any future post where you want to invite engagement, build the invitation into the substance of the content rather than attaching it as an instruction at the end. Make the content itself interesting enough that people want to reply, share, or discuss it. Then, if you want to invite that participation, do it in a way that reflects genuine interest in the response rather than algorithmic strategy.
The test for any engagement invitation you include in a post is simple: if the algorithm were not involved at all, would you still want people to do this? If yes, include it. If no, remove it and make the content strong enough to earn engagement without asking for it.
The bigger signal for creators building on X
Bier's confirmation that manipulation-instruction impressions do not count is a small announcement with a large implication. It tells you that X's measurement infrastructure is now sophisticated enough to distinguish between genuine impressions earned by content people chose to engage with and manufactured impressions generated by content that instructed people to engage with it.
That distinction is the foundation of everything X is building toward. A platform where the metrics reflect real audience behavior rather than manufactured signals. A revenue share program that rewards creators who produce genuine value rather than creators who game the system most effectively. An algorithm that surfaces content people actually want to see rather than content that has been engineered to appear in their feed.
Whether X succeeds in building that platform is an open question. The enforcement actions are real and the infrastructure is genuinely sophisticated. But platforms have attempted to build more authentic algorithmic environments before and have found that the gaming always evolves faster than the detection. The Iron Slopdome is the most determined attempt at this problem X has ever made. Whether it is determined enough is something only time will tell.
What is not an open question is how to navigate the environment X is building regardless of how successful it ultimately is. Create content that earns its engagement rather than manufacturing it. Build real relationships with your mutual network rather than transactional ones. Produce original work in your own voice rather than optimized output designed to trigger algorithmic signals. Measure your performance by the quality of the engagement you receive rather than the volume of the impressions you generate.
Those things work in every algorithmic environment because they reflect what audiences actually want. And in the environment X is building, they are increasingly the only things that work at all.
