Best Time to Post on X: What's Confirmed, What Isn't

Search "best time to post on X" and you will find a specific hour, usually printed in bold, with no link to how anyone measured it. X has never published its own study on this. What can actually be said, from third-party analysis of the platform's ranking model, is narrower and less satisfying: the first 30 minutes after you post matter enormously, your post's reach decays for hours afterward, and neither of those facts points to a specific hour of the day.

There is no verified "best time" — here is what is

Nobody outside X has the dataset that would settle this. X itself has not published an hour-by-hour engagement study, and the pay-per-use API pricing that took effect on February 6, 2026 (no free tier, $0.005 per read — about $5 per 1,000 — a 2 million read monthly cap, and search limited to the last seven days) makes it expensive for an independent researcher to gather one at scale from the outside. One caveat on those numbers, since they get quoted everywhere: X's own developer pricing page returned a 402 Payment Required error every time we tried to check it, so every figure here comes from third-party developer blogs that agree with each other rather than from X's own documentation. X API pricing in 2026 has the full rate card and the same caveat spelled out. Every specific hour you see recommended is coming from somewhere else: one company's own account data, a scheduling tool's aggregate across its users, or a general-purpose "social media best times" roundup with no platform-specific evidence behind it at all.

That API pricing detail matters more than it looks. Before 2026, a research team with a modest budget could pull a large, timestamped sample of public posts and check for real patterns. At $5 per thousand reads with a monthly ceiling, gathering a dataset large enough to say anything reliable about "best hour" — across enough accounts, follower sizes, and timezones to be more than a single company's anecdote — now costs real money before a single insight comes back. That has not stopped the "best time to post" articles from being written. It has just made it more likely that the ones being written are repeating an old number rather than checking it against anything current.

Why most "best time" charts do not cite a source

This is the part worth saying plainly, because it is unusual for an article on this topic to say it: the confident hour on most best-time charts is not backed by a citation you can check. It may be true for the account that produced it. It is not evidence about your account, your audience's timezone, or what you post about — all of which change when the people who would engage with your post are actually online.

Look at how these charts tend to get built and the honesty problem compounds. A scheduling tool publishing "the best time to post on X" is, in most cases, aggregating engagement across its own paying customers — a self-selected group that already skews toward a particular account size, posting cadence, and audience makeup. (We look at how tools in this broader monitoring space compare in our comparisons.) A general "best times for social media" roundup often goes further and blends data across X, Instagram, LinkedIn, and TikTok into a single chart, as though a feed ranked by a machine-learned model processing over 100 million posts, replies, and reposts a day behaves the same way as a feed shown in chronological order. Neither approach is necessarily dishonest. Both are presented with far more confidence than the underlying method can support.

What X's algorithm release actually confirmed

On January 20, 2026, X published the code behind its Grok-based recommendation model on GitHub, under xai-org/x-algorithm — a confirmed, dated event, not a rumor, and the closest thing to primary evidence anyone outside the company has. What it confirmed is that ranking is now a model trained on behavior rather than a set of hand-written rules, described in coverage of the release as processing over 100 million original posts, replies, and reposts a day. It did not include a chart of best posting hours, because that was never the kind of thing the ranking model measures — it scores individual posts against each other in something close to real time, not slots on a clock.

Two further changes followed the initial release: a large update in May 2026, described as the biggest since the code was first published, and a "mutual-follower" change in July 2026 reported to favor visibility between accounts that already follow each other. Neither update came with a posting-time recommendation either. If a best-hour signal existed inside the model in any form X considered worth publishing, the two most likely opportunities to say so — the original release and the subsequent updates — would have been the moment to do it.

The first 30 minutes: the one signal with real teeth

Third-party analysis of that released model reports that a post needs to clear an early engagement threshold — roughly ten or more genuine interactions — within about its first 30 minutes to keep circulating at all. This has not been confirmed by X directly, but it is the single most specific, most repeated claim in the material we reviewed, and it explains something the hour-of-day charts cannot: two posts published at the exact same hour can have completely different outcomes, because what happens in their first half hour, not the hour itself, is what the model is reported to be scoring.

It is not only analysts making that point. Typefully, a scheduling tool — the kind of publisher with the most obvious reason to print an hour — leads its own explainer of the January 2026 update with the mechanic instead.

Source: typefully.com/blog/x-algorithm-open-source — captured 2026-08-20

Read that bullet closely and it is not a best-hour claim at all. "When your audience is active" is a different quantity for every account, which is the opposite of a universal hour, and the sentence after it points at the same threshold the analysts describe. Published the same day X opened the code, by a company that sells scheduling, it still declines to name a time.

That threshold also explains why a small account and a large account cannot share a "best hour" even if they post at the exact same minute. Ten genuine interactions inside thirty minutes is a very different bar for an account with forty followers than for one with forty thousand. A best-time chart built from a large account's data will recommend hours when that account's audience happens to be online — which tells a much smaller account very little about whether its own first thirty minutes will clear the same threshold, because the two accounts are not competing against the same bar in the first place.

Time decay: why a post keeps losing reach for hours afterward

The same body of third-party analysis describes visibility as decaying on a roughly six-hour half-life after a post goes up — whatever hour that happens to be.

This is the honest answer to "what does timing actually do" on X: it resets where you start on this curve, not how steep the curve is. A "best hour" recommendation, if it does anything at all, is really a guess about when your specific audience is online enough to clear the 30-minute threshold before the decay sets in — which is a statement about your followers, not a universal fact about the platform.

It is also worth being clear about what this curve is not. It is not a measured plot of any single account's actual post performance over 24 hours — nobody outside X has published that dataset. It is the direct arithmetic result of applying the reported "halves roughly every six hours" rule forward from an assumed starting value of 100, drawn to show the shape of a claim that shows up consistently across the third-party sources we reviewed. The shape is the useful, defensible part. The exact numbers on the axis are an illustration of a reported rule, not a fact X has confirmed about your account or anyone else's.

Why "reply within 15 minutes" beats "post at 9am"

If you are choosing between memorizing a recommended hour and building a habit, the habit that the evidence actually supports is speed after an event, not a slot on a clock: reply to a fresh post within 15 minutes rather than two hours later, and check on your own posts in their first half hour rather than walking away. What the reply data actually says covers the reported engagement weights behind that, and how to build the ten-minute version of the habit. Both of those are things you control every single time, regardless of what hour it happens to be. A fixed "best hour" is something you can only act on once a day, if that. The bottleneck in that habit is not typing speed, it is having a post in front of you worth answering — which is the only part we sell (disclosure: WarmList is ours, and it is a search, not a scheduler).

What we found testing this on Reddit

We ran the same question against our own Reddit data — 1,175 posts across 8 subreddits, gathered over roughly six months — and found no detectable best time or day at all; almost every post sat at a score of 1 no matter when it went up. That is a separate platform with a separate ranking mechanism (chronological within a subreddit, rather than a single cross-platform model), and the full measurement, including the parts of our own method that failed, is written up on its own rather than here. The short version travels well to this article too, though: which community you post in mattered far more than when, and that finding came from directly measuring the outcomes rather than repeating a chart someone else had already published.

What actually decides whether people see your post

Strip away the specific hour and what is left is not nothing — it is just a different kind of advice. Post when you can be present for the next 30 minutes to help it clear that early threshold, expect it to lose most of its reach within the first day either way, and put your actual effort into replying fast to other people's posts — the method we cover in full elsewhere — where the timing effect is both larger and something you can act on every time. None of that fits on a chart with a bold hour printed on it, which may be exactly why so few articles say it.

We keep this same distinction — confirmed versus reported — in everything else we publish on our blog, because it is the difference that actually matters when you are deciding what to act on. If you take one thing from this piece, take the distinction it keeps drawing rather than a number: "confirmed" means X said it, in its own words, somewhere checkable. "Reported" means several independent parties studying the platform agree on it, without X having said so itself. Every claim above about the first 30 minutes, the six-hour decay, and the 15-minute reply window sits in the second category, clearly labeled as such. The only claim in the first category is that the ranking model exists, was trained on real behavior, and was published on a specific date. A "best time to post on X" article that cannot tell you which category its own numbers belong to is not being dishonest so much as skipping a question worth asking before you act on anything it tells you.

The part of this you can actually control

Strip out every unverified hour and what survives is a habit: be present early, and reply to other people fast. The hard part of that habit was never speed — it is having something in front of you worth being fast about. That is the entire job WarmList does: paste a product URL, and one run reads the public timeline, scores 500+ posts, and returns each candidate with the post, its score, and the reason for the score. $9 a run, one time, no subscription. And in the same spirit as the confirmed-versus-reported line this article keeps drawing: it runs once, when you ask it to. It does not watch X for you afterwards, and it does not send alerts.