How Does the New X Algorithm Rank Your Posts?

TL;DR

X ranks posts by predicting how likely each user is to engage positively, subtracting predicted negative actions, and applying filters for relevance, diversity, safety, and spam. Creators can improve their chances by staying within a clear niche, earning meaningful engagement, posting quality content at a tested cadence, spacing posts apart, and avoiding rage bait, repetitive comments, copied material, and excessive hashtags.

Transcript

Over the past 6 months, X has completely changed the algorithm. And as a result, many creators have seen their reach take a hit as they've adapted to the new system. So, in this video, I'm going to explain how the new algorithm works, how it's different than the old algorithm, and what it means for creators, all based on the most recent update to t... Read More

Key Insights

• The For You feed is built from two sources: in-network posts from accounts a user follows and out-of-network posts selected according to predicted interest. Following still matters because it increases the likelihood that an account's content will appear, although followers will not necessarily see every post.
• Engagement history teaches X what each user prefers. Likes, replies, reposts, quotes, clicks, profile visits, video views, photo expansions, shares, dwell time, and follows are positive signals, while not-interested selections, mutes, blocks, and reports are negative signals.
• Initial filtering removes posts that should not become active recommendations. Examples include duplicates, old posts, posts written by the viewing user, previously served content, paywalled material, posts containing muted keywords, and posts from muted authors.
• AI scoring predicts how likely a user is to take each engagement action. Positive predictions, including likely replies, reposts, and quotes, add weighted value, while predicted mutes, blocks, and reports subtract value, producing a personalized score for every remaining candidate post.
• Author diversity prevents the feed from showing consecutive posts from the same account. Even when a user regularly likes one creator's work, the system introduces variety because data cited in the transcript indicates that diverse feeds keep people using the app longer.
• The new algorithm differs from the old system because distribution depends on AI matching rather than hard-coded engagement rules. Strong early replies previously made promotion more predictable, but the current prediction system can produce inconsistent results because its matches are not always correct.
• A consistent niche helps creators retain audience attention. When users repeatedly see posts and choose not to engage, they may become less likely to receive that creator's future content, even if they already follow the account.
• Quality, spacing, and originality can protect reach. The creator reports success with two to three daily posts and gaps of at least three hours, but recommends testing each account's cadence. Multiple hashtags, copied content, repetitive comments, and other spam signals can prevent recommendations.

Questions & Answers

Q: How does the X algorithm decide which posts to show?

The X algorithm combines posts from followed accounts with out-of-network recommendations. It records each user's positive and negative engagement history, filters unsuitable candidates, and uses AI to predict possible reactions to every remaining post. Weighted positive predictions increase a post's score, while predicted negative actions reduce it. Author diversity and safety checks are then applied before content reaches the For You feed.

Q: What positive engagement signals does the X algorithm track?

Positive engagement signals include likes, reposts, quotes, replies, clicks, profile clicks, video views, photo expansions, shares, dwell time, and following the author. X examines which types of content produce these actions for each user. If someone frequently replies to personal finance posts, for example, the system learns that similar content is more likely to interest that person in the future.

Q: What negative signals can reduce a post's reach on X?

Negative signals include marking a post as not interesting, muting its author, blocking the author, and reporting the content. The recommendation system gives these predicted actions negative weight when calculating a post's score. Creators therefore risk harming future distribution when tactics such as rage bait generate substantial negativity, even if the same posts initially receive millions of impressions or attract new followers.

Q: Why is choosing a clear niche important on X?

Choosing a clear niche helps creators keep delivering the material that attracted their followers. Because X personalizes recommendations from engagement history, every ignored post can decrease the likelihood that the same user will see that creator again. This effect can apply even when the user follows the account, so consistent subject matter supports continued attention and repeated positive engagement.

Q: Do followers still matter under the new X algorithm?

Followers still matter because in-network content is one of the two sources used to build the For You feed. The recommendation system also treats following an author as a positive engagement action, making that author's content more likely to appear for the follower. This does not guarantee visibility for every post, but the source code discussed in the transcript indicates that building followers remains worthwhile.

Q: How often should creators post on X?

There is no single posting frequency presented as correct for every creator. The creator reports that posting around two to three times per day has produced the best results for him, but recommends testing different cadences for each account and niche. Posting excessively can cause posts to compete under the author diversity rule, which prevents the feed from repeatedly serving content from one account.

Q: Why should X posts be spaced apart?

Posts may take around 24 hours, or sometimes longer, to reach their full potential on the For You feed. Publishing several posts together can therefore make them compete for distribution, particularly because the author diversity rule limits consecutive content from the same account. The creator typically leaves at least three hours between posts, while advising others to test the best gap for their niche.

Q: What content may be treated as spam by the X algorithm?

Content containing multiple hashtags, material copied and pasted from another person, repetitive comments, and other spam-like activity may fail to reach the For You feed. The transcript also describes users being ghostbanned after posting AI-generated or repetitive comments. Since every post passes through AI filtering that includes spam detection, creators should prioritize original, useful content instead of repetitive posting patterns.

Summary & Key Takeaways

• The For You feed combines posts from followed accounts with out-of-network recommendations. X learns from positive signals such as replies, reposts, clicks, views, dwell time, shares, and follows, while also tracking negative signals such as not-interested selections, mutes, blocks, and reports to personalize what each user sees.

• Candidate posts first pass through filters that remove duplicates, old or previously served posts, self-authored content, muted material, paywalled posts, and unsafe content. AI then predicts each possible engagement action, applies positive or negative weights, calculates an overall score, and considers author diversity before recommending a post to the user.

• Creators should consistently publish valuable content for the audience they attracted, because repeated non-engagement can reduce future visibility, including among followers. Quality matters more than volume. Posting too frequently, reposting heavily after publishing, using spam-like tactics, or attracting negative reactions through rage bait may limit reach and weaken later distribution.