Engagement Signals: The 19 Factors X Uses to Score Your Content

X's recommendation algorithm doesn't treat all engagement equally. The open-source code reveals a hierarchy where some signals carry far more weight than others. The algorithm predicts the probability of each action for every user-post pair, then multiplies each probability by the signal's weight to produce a final score.

The signals fall into three tiers, plus a negative category that can override everything:

Tier 1 — The algorithm builds distribution around these:

1. Reply: The highest-weighted signal. Replies require the most cognitive effort from users, so the algorithm treats them as the strongest indicator of valuable content. Reply-to-reply chains (when someone replies to a reply on your post) carry even more weight.

2. Retweet / Quote Tweet: Sharing your content with their audience is a strong endorsement signal. Quote tweets carry slightly more weight than plain retweets because they add original commentary.

3. Favorite (Like): The lowest-effort Tier 1 signal, but still significant. A like confirms that both the hook and the payoff delivered value.

Tier 2 — Structure drives these:

4. Dwell time: How long a user spends reading your content before scrolling. Long-form articles with maintained momentum generate high dwell time. The algorithm rewards this because it keeps users on the platform longer.

5. Profile click: When someone reads your content and then clicks your profile to see more. Triggered by distinctive voice and specificity that makes readers curious about the author.

6. Share (DM, external): When users share content through direct messages or external channels. Content that provides both personal resonance and reference value gets shared most.

7. Link click: If a link exists in the post, clicking it counts as a positive signal. But links in the main post body reduce overall reach, so the best practice is placing links in the first reply.

Tier 3:

8. Follow: When someone follows the author after reading the content. Driven by consistent, distinctive point of view across multiple pieces. A lower-weight signal individually, but compounds over time.

Negative signals (distribution killers):

"Not interested," Block, Mute, and Report. These signals don't just reduce your current post's score. They actively suppress the algorithm's willingness to show your future content. Clickbait that doesn't deliver, AI-sounding prose, aggressive selling, and spam formatting are the most common triggers.

Understanding the signal hierarchy changes how you write. If replies carry the highest weight, your article's ending matters as much as its hook. An ending that invites replies (an arguable statement, a complexity admission, a moral verdict) generates more algorithmic value than a neat summary.

If dwell time is Tier 2, then pacing matters. Varied sentence length, open loops, and 2-4 line paragraphs keep readers scrolling. Three sentences of similar length in a row signal monotony and invite the scroll-away.

If negative signals suppress future distribution, then avoiding AI tells (words like "pivotal," "leverage," "furthermore") isn't just about sounding human. It's about protecting your account's algorithmic standing.

Write Better Articles structures every article around this three-tier hierarchy. The hook is designed to earn favorites (Tier 1), the body sustains dwell time (Tier 2) through pacing rules, and the ending is engineered to generate replies (Tier 1, highest weight). The Algorithm Brief's "Primary signal" field tells you exactly which signal your article targets.

See engagement signals in action — generate an article and examine the ending. Every article ends with either an arguable statement, a complexity admission, or a moral verdict, all designed to make readers reply rather than just like. Write my article →

How many engagement signals does X use?

X's algorithm scores 19 engagement signals, organized into three tiers plus negative signals. Tier 1 includes replies (highest weight), retweets, and favorites. Tier 2 includes dwell time, profile clicks, shares, and link clicks. Tier 3 covers follows. Negative signals like Block and Mute suppress distribution entirely.

Which engagement signal matters most on X?

Replies carry the highest weight in X's engagement scoring. This is because replies require the most effort from users, which the algorithm interprets as the strongest indicator of valuable content. A post with 50 genuine replies typically outperforms a post with 500 likes in algorithmic distribution.

Do negative signals affect future posts?

Yes. Negative signals like 'Not interested,' Block, Mute, and Report don't just reduce your current post's score. They actively suppress the algorithm's willingness to distribute your future content. This makes avoiding AI-sounding prose, clickbait, and spam formatting a matter of long-term account health, not just individual post performance.