The X (Twitter) algorithm

The X algorithm is the recommendation system that ranks and orders posts in each user's feed by predicted engagement, weighting replies and reposts heavily while penalizing signals like external links and blocks.

X ranks content in two feeds: the chronological Following feed and the algorithmic For You feed, where most discovery now happens. When a post goes out, the system scores it against a model of what each viewer is likely to engage with, then decides how widely to circulate it. Early engagement matters most. A post that draws quick replies and reposts gets shown to a larger audience; one that gets ignored in its first minutes rarely recovers.

The weighting is the part most people get wrong. A reply is worth far more than a like, and a repost more than a scroll-past. X has said conversation signals carry the most weight, which is why a post that starts an argument outruns one that just gets tapped. The model also reads negative signals: mutes, blocks, and reports suppress reach, and outbound links have historically been dampened because they pull people off the platform.

You cannot reverse-engineer the exact scoring, and chasing it is a losing game. What holds up is the underlying behavior the algorithm is trying to reward: posts people reply to. Write things worth answering, reply to others in your first hour, and keep the link in a follow-up rather than the main post. That aligns you with the ranking instead of fighting it.

Frequently asked questions

how does the x algorithm decide what to show

X scores each post against a model predicting how likely a given user is to engage, then fills the For You feed with the highest scorers. Replies and reposts count far more than likes, and early engagement drives how far a post travels. Negative signals like mutes, blocks, and reports cut reach, and posts with external links have historically been shown to fewer people.

does the x algorithm punish links

X has historically shown posts with external links to fewer people because links pull users off the platform, though the degree changes over time and X does not publish exact numbers. A common workaround is to put your main point in the post itself and drop the link in a reply, so the post is judged on its own engagement before the link is seen.

why do replies matter more than likes on x

X weights conversation over passive taps because replies signal that a post held attention and prompted a response, which is what keeps people on the platform. A like is one cheap signal; a reply is effort and often pulls in the replier's audience too. Practically, a post built to be answered will out-travel one built only to be liked, even at similar impression counts.
Put it into practice

Post Analyzer — Check a draft for the signals that actually earn reach. Open the free tool →

Related terms

← Back to the full glossary

Knowing the term is the easy part.

Liftli turns your voice notes, calls and commits into posts in your extracted voice — inside the AI you already use, with a one-tap approval gate.

Start free — no card