X's recommendation system is one of the most complex in social media — and it's open-source. Here's exactly how it scores, ranks, and distributes your content to new audiences.
Timeline Types
X serves content through two distinct timeline modes. Understanding which one you're optimizing for changes your entire strategy.
The "For You" feed uses machine learning to surface tweets from accounts you don't follow — weighted heavily by engagement signals, topic relevance, and author credibility. This is where viral reach happens.
The "Following" feed shows tweets only from accounts a user explicitly follows, ordered by time. It's predictable, transparent, and immune to algorithmic suppression — but limits your discovery potential.
Key insight: Most impressions on X come from the "For You" feed. Optimizing for the algorithm — not just your followers — is where significant growth happens in 2026.
Ranking Factors
X's open-source recommendation code reveals a weighted scoring system. These are the top signals that determine where your post appears.
Replies > Retweets > Likes in the weight hierarchy. Conversations indicate genuine interest and boost distribution significantly.
X's interest graph matches post topics to user behavior. Topic coherence across your account builds stronger relevance signals.
Account age, follower quality, historical engagement rates, and Premium status all contribute to an author trust score.
Scoring Formula
Based on X's open-sourced recommendation code (github.com/twitter/the-algorithm), here's how the scoring pipeline actually works.
Step 1
System pulls ~1,500 tweet candidates from followed accounts + interest graph to evaluate for each user session.
Step 2
A neural network scores all candidates using the weighted formula, ranking them from most to least likely to get engagement.
Step 3
Rules applied: max 2 tweets per author, content diversity, NSFW filtering, block/mute lists, and spam removal.
Content Performance
Average engagement rates across X content formats, based on aggregated data from 2025–2026.
Posting Strategy
Timing, frequency, and format recommendations to maximize algorithmic distribution on X.
| Strategy Element | Recommendation | Why It Works |
|---|---|---|
| Best Times (EST) | 8–10 AM, 12–1 PM, 5–7 PM | Peak user activity windows drive faster early engagement accumulation |
| Daily Frequency | 3–6 posts per day | Consistent volume increases chances of catching the algorithm window |
| Thread Length | 4–8 tweets per thread | Threads keep users on platform longer, boosting author credibility score |
| Reply Speed | Reply within 30 min of posting | Early replies from the author signal activity and boost first-hour scoring |
| Hashtags | 1–2 max, highly relevant only | Overuse signals spam; 1–2 targeted tags improve topic categorization |
| Media | Native video or images preferred | Native media keeps users on X vs. external links which are penalized |
| Engagement Bait | Avoid explicit "like/RT if" asks | Manipulative engagement is actively penalized by X's classifiers |
| Link Placement | Add links in first reply, not tweet | Keeps main tweet free from the 50% link distribution penalty |
X Premium
X's subscription tiers directly influence algorithmic reach. Here's what the evidence shows about the distribution advantage for paying users.
X explicitly states that Premium subscribers receive priority ranking in "For You" recommendations. This creates a meaningful organic reach advantage over non-subscribers.
Premium+ users earn ad revenue from replies to their posts. This incentivizes creating content that generates replies — aligning monetization with algorithmic signals.
Premium users can post up to 25,000 characters and upload 3-hour videos. Long-form content encourages more dwell time, which is a positive ranking signal.
Ability to edit posts within 60 minutes means you can fix errors without deleting and reposting — preserving accumulated engagement that counts toward your score.
Accounts without Premium still appear in For You feeds, but are ranked below Premium equivalents. Engagement quality becomes even more critical to compensate.
The blue checkmark is now purely a subscription marker, not a credibility signal. Algorithm trust is still built through historical engagement quality and account age.
Algorithm Tips
X shows only the first ~280 characters before a "Show more" cut-off in feeds. Your opening line determines whether users engage or scroll past — this directly affects your engagement rate score.
X's algorithm applies a significant distribution penalty to tweets containing external URLs. Place your link in the first reply and reference it in the main post to avoid this penalty.
The open-source code reveals replies are weighted 27x more than likes. Ask genuine questions and create debate-worthy content that naturally generates conversation.
Activity signals boost your account's engagement score. Spend 15 minutes replying to others before you post to warm up your account's activity status in the algorithm.
Threads with valuable payoffs at the end generate more retweets (users share the full thread via the last tweet). Longer dwell time on your thread cluster boosts your author score.
Native video uploaded directly to X performs 4.2x better on average than tweets containing YouTube links. Upload video files directly for maximum distribution.
X's interest graph builds a topic model around each account. Accounts that post on 1–3 consistent topics receive stronger relevance signals than accounts that post about everything.
Deletion patterns are tracked and can flag accounts for spam behavior. If a post underperforms, analyze why and apply the learning to your next post instead.
X Analytics shows which posts earned the most impressions from non-followers. Reverse-engineer the format, topic, and hook of your top performers and repeat the pattern.
Ready to grow?
Explore how other platform algorithms work and build a unified multi-platform strategy.