What we are working from
Everything below is grounded in what xAI actually publishes. Since 2026 the For You feed has been ranked by Phoenix, a Grok-based two-tower transformer that handles retrieval and ranking in one model, replacing the Heavy Ranker, SimClusters and RealGraph stack from 2023. The repository states the team eliminated every hand-engineered feature and most heuristics.
One caveat we will repeat wherever it matters: the numeric engagement weights are not public. They sit in a params module that has never shipped, in 2026 or in 2023.
Anyone quoting you an exact multiplier is repeating a leak or an inference. The ordering is reliable; the ratios are not. Full breakdown here.
Why it compounds
Because Phoenix scores relative to your account's own baseline, your recent performance is an input into your next post's distribution. A good week raises the bar you are measured against and the audience you start from; a silent month lowers both.
This is the single most important structural fact about growing on X, and it is why the accounts that grow are rarely the ones producing the best individual posts. They are the ones that never let the baseline go cold.
Consistency beats volume
Posting ten times on Monday and nothing until the following week is worse than posting twice a day every day, even though the totals are similar. The burst competes with itself for the same audience in the same window; the steady cadence keeps the baseline warm and gives the model repeated fresh evidence.
There is a ceiling on this. Posting far above what your audience signed up for drives the negative heads, not interested, mute, block, and those cost more than the extra posts gain. The right cadence is the highest one you can sustain without your reply quality falling off.
The profile-visit loop
Profile visits are one of the nineteen signals the model predicts. They are also the actual mechanism of follower growth: almost nobody follows straight from a timeline post; they click through, read three or four more posts, and decide.
Two implications. Your pinned post and recent few are doing more work than any individual new post. And a post that makes someone curious about you is worth more than a post that simply gets agreement.
What stalls accounts
- Topic drift. The model has learned an association between your account and a subject. Drifting produces posts your existing audience does not react to, which lowers the baseline for everything after.
- Only broadcasting. Replies, especially conversations the author takes part in, sit high in the ordering. An account that only posts and never replies leaves the cheapest signal on the table.
- Chasing the outlier. One post doing 100× your median is luck. Rebuilding your strategy around reproducing it usually costs you the consistency that was actually working.
- Optimising for likes. Likes sit at the bottom of the ordering. Reply rate and profile visits predict growth far better.
Questions we get asked
How often should you post on X to grow?
Consistently rather than in bursts. Because ranking is relative to your account's own recent baseline, a steady cadence keeps that baseline warm while a burst competes with itself for the same audience. The right frequency is the highest one you can sustain without reply quality dropping, since posting far above what your audience expects drives the negative signals the model tracks.
Do likes matter for growing on X?
Less than most people assume. Likes sit at the bottom of the published ranking ordering, below replies and reposts. Reply rate and profile visits are far better predictors of growth, because profile visits are the actual mechanism by which people decide to follow.
Why did my X account stop growing?
The most common causes are topic drift, which breaks the association the model has learned between your account and a subject, and broadcasting without replying, which forfeits one of the highest-value signals available. A gap in posting also lowers the baseline that everything afterwards is measured against.