How the X Feed Works: Algorithm, Signals & Ranking Explained

• X (formerly Twitter) uses detailed data from user interactions to form relationships (Feature Formation) and then ranks posts accordingly.
• The for-you feed is built by sourcing around 1,500 candidate posts which are carefully ranked and refined using filters, heuristics, and ad mixing.
• Both positive signals (likes, reposts, follows) and negative signals (mutes, blocks, reports) affect what appears on your feed.

Understanding how the X feed works is essential for anyone who uses the platform—for content creators, marketers, or casual users. The intelligent design behind the feed leverages network effects and vast data points to tailor your content experience.

Additionally, by knowing the signals that boost or suppress content, users like Alex can better curate a feed that truly represents their interests and filter out content they find less valuable.

The Data Behind the Feed

When you scroll through X, the platform is not randomly selecting posts. It uses a host of data such as likes, reposts, bookmarks, clicks, and even the time you spend on each post.

This collection of data, which might include geolocation and trending topics, is what X uses to build a profile of your interests. For example, if Alex consistently interacts with Bella's content more than Charlie's, the algorithm assigns a higher weight to Bella's posts, influencing their positioning in Alex's feed.

More insights on targeting audiences on X can be explored in Targeting Audiences on X.

Feature Formation & Ranking

Feature Formation: Here, relationships between users and posts are formed using a wide variety of signals. The algorithm builds a graphical representation of connections among users based on interaction strength.

Verified profiles, user engagement (likes and reposts), and content quality (usage of images and audio) are all factors. If a post ranks high in positive signals—such as a 30x increase for a favorite or a 20x boost for a repost—it has a higher chance of appearing in feeds.

Ranking: Out of the selected 1,500 candidate posts, the system then predicts the likelihood that you will engage with each one. Posts are assigned ranks based on current engagement data and refined using further filtering rules. This stage also ensures diversity, so you don't end up with too many posts from one account, and it filters out content that violates legal or community standards. For deeper insights into the impact of engagement metrics on ranking, see Twitter X metrics deep dive.

For further optimization insights through experiments, see AB Testing on X: Optimize Twitter X posts.

The Role of Signals

Every action you take counts on X. Positive signals such as following, liking, bookmarking, and replying increase the strength of the connection between you and the content creator.

These signals are weighted differently—favorites and reposts provide much larger multipliers compared to a simple like. Conversely, negative signals such as muting, blocking, or reporting completely prevent relationship formation. Detailed insights into interaction metrics are explained in Twitter engagement strategies.

This dual design reinforces that creating engaging yet non-controversial content is valuable both for increasing reach and maintaining a healthy engagement rate.

Network Effects in Play

X's algorithm is not just about individual posts—it is also strongly influenced by network effects. The system considers not only who you follow, but also how your followed accounts interact with posts and other users.

This creates clusters of communities (or networks) that can lead to content being widely distributed if it resonates within a particular group. As X refines its feed, it balances posts from accounts you follow with those from accounts you do not follow, helping you discover new content while still seeing familiar voices.

The Mixing Stage

After Feature Formation and Ranking, posts undergo a further refinement stage where additional filtering, heuristic adjustments, and product-based features come into play. This stage ensures that the final feed is not just a list of ranked posts but also includes a mix of organically engaging content and advertisements, ensuring sustainability for the platform. This is where diversity is enforced to avoid overwhelming users with monotonous content that even if perfectly ranked might not provide variety.

For fundamentals on ad placements on X, see X Ads Basics.

• Overposting or Spammy Behavior: Flooding your feed may lead the algorithm to suspect spam. Focus on quality rather than quantity.
• Ignoring Audience Feedback: Regularly check engagement metrics and alter your methods. If posts continually receive negative signals, consider revising content style.
• Content Monotony: Repeatedly posting similar content without variation can lead to audience fatigue. Mix it up with different formats and topics. Consider diversifying your content presentation, as seen in Reddit post formats.
• Neglecting Timing: Posting off-peak when your audience is less active can diminish early engagement which is crucial for reaching broader audiences. More details on timing can be found in Early Engagement Windows on X.