Best time to post on X (Twitter): what five studies and X's code say

Ask five studies for the best time to post on X and you get five answers: 9 a.m., 9 to 11 a.m., 12 to 6 p.m., 7 to 11 p.m., and a Sunday evening. Each is right about the accounts it measured and silent about yours. X’s own code is more useful than any of them, because it says how long a post has to find its readers: a small account’s boost lasts 2 hours, and the For You feed stops considering a post after 48.

We read the five studies with their dates, samples and methods, checked X’s published code at the same snapshot as our post on how the X algorithm works, and wrote down the test we’d run to find a slot for one account. We have no posting-time data of our own to add, and we’d rather say so than print a chart.

The short answer #

Post when your readers are awake and about to scroll, in their time zone, and leave hours between your originals. Three of the five studies land on weekday mornings, so a weekday morning in your readers’ zone is the place to start. Then measure your own posts, five per slot, and let the numbers overrule us.

What the studies say #

The worst times disagree as much as the best. Buffer finds the drop in the evening, 6 to 11 p.m. on every day of the week, and names Saturday and Friday the weakest days, which puts SocialBee’s best window inside Buffer’s worst. Sprout names Saturday. Hootsuite says anything before 5 a.m. SocialBee’s weakest hour is Thursday at 1 p.m., inside Sprout’s best window. SocialPilot calls weekends the least favourite days and still lists weekend hours from 9 a.m. to 8 p.m.

Why five studies give five answers #

• Different accounts. Buffer and SocialBee counted only posts sent through their own tools, and Sprout and SocialPilot counted their own customers. None of the samples is X, and none is you.
• Different questions. Sprout measured when engagement happens. Buffer and SocialBee measured which publishing hour earned a post the most. A 9 a.m. post is still fresh when Sprout’s 12 to 6 p.m. crowd arrives, so the two can both be right.
• Different clocks. SocialPilot reports Eastern time. The others use each account’s local time or normalise across zones, which hides where the audience sits.
• Different seasons. Sprout covers three winter months. SocialBee covers the first half of 2026. Hootsuite and SocialPilot don’t say.
• Loose pages. Sprout lists Sunday among the worst days and rates Sunday afternoon moderate in the same piece. SocialPilot gives Friday two different sets of hours. Buffer’s single best slot falls on a Tuesday while it calls Wednesday the best day, which can both be true, but the headline hour and the headline day get quoted as if they were one finding.

We’d read any such chart as the rough shape of a week, not as an appointment.

What X’s code says about timing #

X publishes its ranking code as xai-org/x-algorithm. We read the November 7, 2026 snapshot, the same one behind our algorithm post. Four things in it concern time.

48 hours. The For You pipeline runs an age filter set to a maximum post age of 48 hours. After two days a post can’t enter anyone’s For You feed, whatever it has earned, so whatever For You gives a post, it gives inside that window.

2 hours. The new-author boost lifts one fresh post per feed load from a small account. By the defaults synced on November 6, the post must be an original from an account with 50,000 followers or fewer, under 7,200 seconds old, and under 200 views in Home. If your account is small, the two hours after you publish are the only time this lift can reach you, so publish as your readers arrive rather than three hours ahead of them.

No weight for the hour. X’s published weights multiply predictions of what a viewer will do with a post: reply, like, copy its link. There is no entry for the hour the post went out. Timing matters because it decides who is awake to be predicted about, and for no other reason the code shows.

Your own posts compete. In one feed load, X lifts one fresh post per author and shrinks the score of an author’s second and third posts. Five posts in ten minutes fight each other for the same slot. The figures and their caveats are in the algorithm post.

How to find your own best time #

X’s analytics dashboard, now limited to X’s paid plans, shows impressions by day and by post. We found no view of your followers by hour. Since May 2026 it shows what share of your followers were active in the last 24 hours, and X’s head of product Nikita Bier said that 3-hour data would follow and that “People overestimate how many of their followers are active at a given time”. Until the hourly view exists, your own posts are the data, and every post shows its views in public, so the test below needs no paid plan.

1. Decide whose morning you’re posting for. If you write for US readers from Europe, Eastern time is your clock.
2. Pick three or four slots in that zone. Take them from the table above, for example 9 a.m., 1 p.m. and 7 p.m. on weekdays, plus one weekend slot.
3. Rotate them until each slot has five posts. At five originals a week that takes three or four weeks. Keep the kind of post the same, originals without a link or always with one, and write new text every time.
4. Read the views at 48 hours. Note each post’s view count two days after it went out, once For You has stopped serving it.
5. Compare slots, not posts. One post that travels can carry a slot on its own, so take the median of at least five posts per slot before you trust a gap.
6. Keep the winner and retest. Run it again each quarter, after each clock change, and whenever X moves its weights. The like weight alone changed three times between October 13 and November 5.

Common mistakes #

• Posting on your own clock. A 9 a.m. slot in Berlin is 3 a.m. in New York. The studies report local time for the account, which only helps if the account and its readers share a zone.
• Reading a chart without its row. Buffer’s 9 a.m. and Sprout’s 12 to 6 p.m. come from different customers, different metrics and different months, and SocialPilot’s hours are Eastern. Without its sample beside it, the number is a guess.
• Posting in bursts. X lifts one fresh post per feed load and discounts an author’s second and third in the same load. Space originals hours apart.
• Testing slots with the same text. X’s automation rules forbid duplicative posts, and a duplicate-content label keeps a post out of non-followers’ reply notifications. Change the words every time.
• Judging a slot on one post. A single post that catches a reply from a large account tells you about that reply, not the hour.
• Setting a slot once. Clocks move twice a year, and on different weekends on each side of the Atlantic, so a US slot scheduled from Europe drifts by an hour for a week or more each spring and autumn, and X retunes its weights more often than the clocks move.
• Writing off weekends. SocialBee’s top slot is Sunday 7 p.m. and Sprout rates Sunday afternoon moderate, while Buffer names Saturday the weakest day and SocialPilot calls weekends the least favourite. Test one weekend slot before you decide.
• Posting into the news. Hootsuite’s team says to watch real-world events when you pick a slot on X, because the feed runs on news, and when everyone is watching one thing your slot’s history counts for less.
• Treating timing as the main lever. Buffer’s 52-million-post report from March 2026 ranks what you post first, how often second, and when last. Timing gives a good post its first push. It doesn’t rescue an average one.

Rules we’d follow #

1. Post in your readers’ time zone, as their day starts, so the post is fresh when they open the app.
2. Publish originals when readers are awake. Replies skip the new-author boost, so they can go out any time.
3. Leave hours between originals, and never send a burst.
4. Write new text for every post, including tests.
5. Measure at 48 hours, five posts per slot, median not best.
6. Retest every quarter and after every clock change.
7. Spend the saved effort on the post itself, the one point where the studies agree.

Common questions #

Schedule in your readers’ morning #

Postmoderno publishes at the time you choose, so a post for 9 a.m. in New York can go out while your laptop is closed. Tell your agent how you want to post, say three originals a week in US mornings, and it keeps that as your posting strategy for the next drafting session. Every preview shows the time in your zone and in UTC before anything is scheduled, which is how a wrong offset gets caught.

How to schedule tweets from Claude or ChatGPT covers the setup. Our free writing skills say the same about timing as this post: there is no single best hour, so write for the readers you have and post when they’re there. Posting on LinkedIn too? Best time to post on LinkedIn reads the LinkedIn studies the same way.

← all posts