Sabitlenmiş Tweet
🗞️ X algorithms are now public.
How do you grow an account? A quick breakdown from Sally.
• Interest matching.
The algorithm does not even look at likes at first. It compares the user’s interest vector with the meaning vector of your post. If the post does not match what the user has interacted with before, it never enters the candidate pool. It is not ranked at all. Content focused on a single topic almost always performs better than random posts about everything. Pick one niche and stay in it.
• Why some posts suddenly take off.
X predicts multiple actions at once: likes, replies, reposts, quotes, saves, and watch time. What matters is not one action, but the combination. When a post gets likes, comments, and saves at the same time, its total score grows nonlinearly. That is why some posts accelerate instead of growing steadily.
• Photos and videos are separate signals in the algorithm.
The system explicitly tracks image opens, video views, and video completion. This is why carousels and screenshots that make people open the media can work well.
• The algorithm aggressively cuts off old posts.
If content is not fresh, it simply drops out due to age filtering. Real-time posts almost always have an advantage over outdated topics.
• Why posting too often hurts.
X deliberately reduces the weight of posts from the same author so the feed does not turn into a stream from one account. In practice, posting too frequently lowers reach. One post per hour or less works better.
• What kills reach?
The algorithm subtracts points for negative signals: mute, block, report, and “not interested.” Even with good engagement, a spike in these actions can completely stop distribution.
• Watch time and sharing matter more than likes.
The algorithm strongly values read-through and time spent on screen. Quotes rank higher than simple reposts.
• Keywords still matter.
Posts with clear topic framing and high semantic density are more likely to get recommended outside your audience. There is talk on X about hashtags becoming important again. There is no hard confirmation yet, but hints from Nikita Bier suggest a shift toward so-called cashtags. If that happens, tags may become relevant again.
To sum it up:
• stick to one topic to lock into the algorithm’s interest graph;
• create posts that trigger multiple actions, not just likes;
• use media as an attention retention tool;
• post less often, but at the right time;
• avoid sharp negativity and low-quality formats;
• experiment with hashtags, but do not overdo it;
• optimize for reading time and shareability, not likes.
What do you think about this info? Alpha?
English
