how to

How the X algorithm scores posts: the engagement it actually rewards

the short answer

The X algorithm weights engagement types unequally, rewarding replies, quotes and bookmarks more heavily than passive likes; x-signal estimates a score for each of your posts against that known weighting so you can write for what the algorithm actually counts.

Most people picture the X algorithm as a single hidden number that decides whether a post flies or sinks. The reality is closer to a weighted tally: different kinds of engagement are worth different amounts, and the actions that signal genuine interest are worth far more than the ones that cost nothing. A like is cheap, so it counts for little. A reply, a quote, a bookmark or a long dwell on your post is expensive attention, so it counts for a lot.

You cannot see X's internal score, and nobody outside the company can. What you can do is estimate one from the engagement weighting that is publicly known, which is exactly what x-signal does for each post. This page explains how that scoring works so you can write for the engagement that actually moves you, rather than the engagement that merely looks good.

every postscored against X's known engagement weighting

Why some engagement is worth more

The logic behind the weighting is simple once you see it. The algorithm is trying to predict whether a post is worth showing to more people, and the best predictor is costly, deliberate engagement. Anyone can tap a like in passing; far fewer people will stop to write a reply, quote your post to their own audience, or bookmark it to return later. Those actions are stronger evidence that the content earned real attention, so they pull more weight in the score.

This is why chasing likes is a trap. A post can be liked thousands of times and still be judged mediocre by the algorithm if nobody replied, quoted or saved it, while a post with a fraction of the likes but a thread of genuine replies can be scored much higher and pushed further. Understanding the hierarchy changes what you aim for: you write posts that are worth replying to and worth saving, not just worth a thumb.

Reading the score x-signal gives you

x-signal scores each of your posts against the algorithm's known weighting, so you can compare two posts on the engagement that counts rather than on raw impressions. Treat the number as an estimate, not X's secret internal figure: it is built from publicly understood weighting, and its value is in the relative comparison between your own posts, not as an official verdict.

Used alongside your engagement history, the score tells you which posts the algorithm was likely to favour and which it quietly buried. Over time you learn which of your formats consistently earn the expensive engagement, and you lean into them deliberately instead of hoping a like-heavy post breaks out.

how it works

  1. 01

    connect your X account

    Link X to x-signal read-only so it can read engagement per post.

  2. 02

    let it score your posts

    Each post is scored against X's known engagement weighting as an estimate, not X's internal number.

  3. 03

    compare your own posts

    Rank posts by score to see which earned the costly engagement the algorithm rewards.

  4. 04

    write for the heavy weights

    Aim for posts worth replying to, quoting and bookmarking, not just liking.

How different engagement types tend to weigh in the algorithm

Engagement typeRelative weightWhy
ReplyHeavyCostly, deliberate; signals the post sparked a real conversation.
QuoteHeavyThe reader stakes their own audience on your post.
BookmarkHeavyStrong intent to return; treated as a save, not a passing tap.
RepostModerateEndorsement that spreads reach, but lower effort than a quote.
LikeLightCheap and passive; weak evidence the content earned attention.

frequently asked

Is x-signal showing X's real internal score?
No. X's internal score is secret. x-signal's score is an estimate built from the algorithm's publicly known engagement weighting, useful for comparing your own posts to each other.
Do replies really beat likes?
In general yes. Replies, quotes and bookmarks are costly, deliberate actions and tend to carry more weight than passive likes, which is why like-heavy posts can still underperform.
Why does my high-impression post score low?
Impressions are reach, not engagement quality. If a post got views but few replies, quotes or bookmarks, the algorithm has little costly engagement to reward, so the estimated score stays low.
Can I use the score to predict a viral hit?
It is a relative guide, not a crystal ball. The score tells you which of your posts the weighting tends to favour; it does not guarantee reach or replace your own judgement.

Last updated June 6, 2026

ready to try x-signal?

open x-signal