Last updated: August 31, 2026
How to get into X Custom Timelines
X launched Grok-powered Custom Timelines on April 22, 2026: more than 75 topic feeds you pin beside For You, Premium-only, iOS first. Grok reads every post and assigns topic labels itself. Hashtags are explicitly not the mechanism. To get surfaced, post consistently about one recognizable topic so your account reads as coherent to the classifier.
This is the first surface on X where a language model, rather than a graph or a keyword index, decides where your post belongs. That changes what is worth doing and makes a decade of hashtag advice obsolete in one release.
What Custom Timelines actually are
A Custom Timeline is a topic feed. X shipped more than 75 of them on April 22, 2026, and they pin at the top of the app next to For You and Following. A Premium subscriber picks the topics they care about and gets a dedicated feed for each one instead of hoping the main algorithm guesses right.
The interesting part is how the feeds are assembled. Grok reads posts and assigns topic labels to them. No author declares a topic, no editor curates the list, and no tag routes anything. Placement is a model output, computed from what the post says.
Two constraints on availability. The feature is gated to X Premium, which is $8 a month or $84 a year, and it shipped on iOS first. Neither affects whether your posts get classified. Premium determines who can read the topic feeds, not who can appear in them.
Why this is different from For You
The For You ranker, open-sourced as Phoenix on August 13, 2026, scores a post using engagement signals: a reply is worth 5.0, a like 0.5, a report -234. Custom Timelines are upstream of that. They decide what a post is about before anything decides how good it is.What works, and what does nothing
Sorted by whether it affects a topic classifier at all.
Why hashtags stopped mattering
A hashtag was a manual routing instruction. It told a keyword index where to file a post, and for years that was the only signal available. A topic classifier does not need the instruction, because it can read the post.
Worse, tags are trivially gamed, which is exactly why a model-based system ignores them. If #ai routed posts into an AI topic feed, that feed would be nothing but people typing #ai. The mechanism X chose is the one that cannot be declared, only demonstrated.
What replaces the hashtag is the body of the post. Write about the subject in the plain vocabulary a reader would use. Not a keyword list, not a density target. A model reading for topic responds to a post that is clearly and specifically about one thing, which is also what a human responds to.
Coherence is a scheduling problem
Here is the uncomfortable part. Nobody fails at topical consistency because they do not understand it. They fail because consistency is measured over months and enthusiasm is measured in weeks. Three good on-topic posts in a burst, then a fortnight of nothing, then some unrelated posts about a conference, and the pattern a classifier could have locked onto is gone.
The fix is structural, not motivational:
- Plan the subject, not the post. Decide the topic for the next six weeks and fill the slots against it. A content calendar makes drift visible before it happens.
- Schedule ahead so gaps cannot open. A queue that is full three weeks out does not care how busy you were. Post at the times your audience is online rather than whenever you happened to open the app.
- Let an evergreen queue carry the quiet weeks. An evergreen queue recycles your best on-topic posts, which keeps the subject signal alive without inventing filler.
- Split genuinely different subjects across accounts. If you cover two things properly, run two accounts. OpenTweet supports up to 10 connected X accounts on Agency, so this is a scheduling decision rather than a logistics problem.
A useful test
Take your last 30 posts and try to name the account topic in one phrase. If you cannot, a classifier cannot either, and no posting frequency fixes that. Start by cutting the off-topic half rather than adding more posts.7-day free trial. Cancel anytime.
Frequently asked questions
What are X Custom Timelines?
Topic-based feeds that X launched on April 22, 2026. There are more than 75 of them, they pin next to For You at the top of the app, and Grok assembles each one by reading posts and assigning topic labels. They are available to X Premium subscribers and shipped on iOS first.
How does X decide which posts go into a Custom Timeline?
Grok reads the post and classifies it. The label comes from the content of what you wrote, evaluated by a model, not from anything you attach to the post. That is why the same post can be surfaced in a topic feed with no explicit signalling from you at all.
Do hashtags get you into X Custom Timelines?
No. Hashtags are explicitly not the mechanism. Custom Timelines are built from Grok topic classification, so adding tags to a post does not route it anywhere. Hashtags were already close to inert for reach on X and they do nothing here.
Do I need X Premium for Custom Timelines?
To browse them, yes. Custom Timelines are a Premium feature, and Premium is $8 a month or $84 a year. Your posts can be classified into a topic regardless of your own subscription. Premium determines whether you can read the feeds, not whether you can appear in them.
How do I get my posts surfaced in a topic feed?
Post consistently about one recognizable topic so the classifier can place your account without ambiguity. A feed of unrelated posts gives a topic model nothing stable to work with. Consistency of subject matters more than volume, and neither hashtags nor keyword stuffing help.
Does posting about many topics hurt my chances?
It dilutes them. Every off-topic post is a data point that makes your account harder to classify confidently. If you cover several subjects seriously, running a separate account per subject is a cleaner answer than mixing them into one stream.
Can I schedule posts to keep a topic consistent?
Yes, and that is the practical mechanism. Topical consistency over weeks is a planning problem, not an inspiration problem. A content calendar that holds a subject steady, and an evergreen queue that keeps posting on that subject when you are busy, is what makes coherence survive a bad month.
Related guides
Consistency, timing, and how the rest of X ranking works.
Build a content calendar
Plan a subject across weeks so topical coherence survives a busy month.
Find your best time to tweet
Your own posting data, not generic best-time charts.
Evergreen queue
Recycle your best on-topic posts so the quiet weeks do not break the pattern.
How often should I tweet?
Cadence that is sustainable, which is the only cadence that compounds.
Tweet algorithm scorer
Score a draft against the published Phoenix ranking weights.
Grow your following
What actually moves the number once tricks stop working.
Stay on topic for six weeks straight
Plan the subject, schedule the posts, and let an evergreen queue hold the line when you are busy. That is what a topic classifier is looking for.
7-day free trial. Cancel anytime.