How the X algorithm ranks posts (and what that means for reach)
What X published about its ranking in 2023, which signals matter most, and why a drop in reach is usually ranking, not a shadowban.
In March 2023 X published the source code of its recommendation system, including the weights its main ranking model used to score posts for the For You timeline. The system has changed since and X does not publish current weights, but the published version still explains the shape of how reach works.
The short version
For each candidate post, the model predicts how likely you are to engage with it in different ways, multiplies each probability by a weight, and adds them up.
| Predicted action | Published weight (2023) |
|---|---|
| Reply that the author engages with | 75 |
| Reply | 13.5 |
| Opening the profile and engaging | 12 |
| Opening the conversation and engaging | 11 |
| Repost | 1 |
| Like | 0.5 |
| Negative feedback ("show less", mute, block) | -74 |
| Report | -369 |
What it means
- Conversation beats likes. A reply is worth many likes, and a reply the author answers is worth far more.
- Negative signals are brutal. One report outweighs hundreds of likes. Content that makes people mute or block you sinks fast, and accounts that collect these signals get filtered.
- Time on X matters. X has said publicly that posts sending people off the platform get less attention, which is why posts with external links often reach fewer people. Putting the link in a reply is a common workaround.
- Subscriptions affect ranking. X Premium includes prioritized ranking in replies and search.
Ranking vs. shadowban
Ranking decides how many people see a post. A shadowban removes the post from places it would otherwise appear. If your reach dropped but all four FSB tests are clean, your problem is ranking: posting time, topic, format or audience, not a restriction.
Check an account
Four live tests from a logged-out view. No password, no app permissions.
By searching, you accept our Terms of Service and Privacy Policy.