The best time to post on X (Twitter) is weekday mid-morning, around 9am to 11am. That is where Buffer's 8.7 million tweets and Hootsuite's cross-network study both land. Sprout Social disagrees and says noon to 6pm, a window that opens after the other two have closed, and the deeper you read the published tables the less any of them holds together.
Key takeaways
- Buffer publishes only 3 distinct rows across 7 days, then ranks three identical days 2nd best, 2nd worst and worst.
- SocialPilot publishes two different Eastern Time schedules on one page. They disagree on all 7 days.
- Metricool's 9pm measures connected users, not engagement, so it is not the outlier it looks like.
- Of the platforms in this series, only X documents post age as a ranking feature, in open-source code.
What is the best time to post on X?
Four studies rank posts by performance, and a fifth measures something else entirely.
- Buffer analysed 8.7 million tweets sent through Buffer and names 9am on Tuesday the single best slot (Buffer, 2026).
- Sprout Social identified peak windows between 27 November 2025 and 27 February 2026 and lands on Tuesday to Thursday, noon to 6pm (Sprout Social, 2026).
- SocialPilot analysed around 700,000 posts from 50,000 accounts, headlines 8am to 11am on weekdays and names Tuesday to Thursday its best days. Its own day table then adds 3pm slots on Wednesday and Thursday and a 7am on Friday, so the headline window does not cover its own numbers (SocialPilot, 2025).
- Hootsuite analysed over 1 million posts with a data-science partner and says 9am to 11am, Wednesday to Friday (Hootsuite, 2025).
Three of those four point at mid-morning. Sprout is the odd one out, and the gap is not small: its window does not open until noon, an hour after Hootsuite's has shut.
On the other end, Buffer names evenings from 6pm to 11pm the weakest stretch across every day, and Hootsuite names before 5am. Those are the only two worst-time windows any of the four quantifies.
Worth knowing: a fifth number circulates widely and measures something the other four do not. Metricool says 9pm, which reads as a flat contradiction until you check what it measured. Its study identifies 9pm as "the time of day with the highest average connected users" (Metricool, 2024). That is audience presence, not post performance. More people are online at 9pm; that is not the same claim as posts doing better at 9pm. Metricool's study publishes no engagement-ranked hour, so there is no like-for-like figure to put in the table alongside the others, and it is a 2024 study against four from 2025 and 2026.
Buffer ranks three identical days 2nd best, 2nd worst and worst
Buffer's is the largest stated X corpus and the most confidently ranked. It names Wednesday the best day, Tuesday second, Thursday third, Friday second-worst and Saturday the worst.
Then you look at the table it publishes.
Across seven days there are only three distinct rows. Monday, Thursday and Sunday all read 9am, 10am, 8am. Tuesday, Friday and Saturday all read 9am, 10am, 11am. Wednesday alone differs.
So Tuesday, Friday and Saturday publish exactly the same three slots, and Buffer ranks them second best, second worst and worst. Whatever separates them is not in the table. The ranking comes from data Buffer did not publish, which means the day ranking and the times cannot both be used as evidence for each other.
There is a smaller wrinkle in the same place. Buffer names 9am Tuesday the single best time of the week, and Wednesday the best day, but Wednesday's own top slot is 10am.
SocialPilot publishes two Eastern Time schedules
SocialPilot's page carries a full seven-day schedule, footnoted "The timings given above are in the Eastern Time Zone (EST)". Further down it carries a second seven-day schedule under a USA heading, closing with "All the above-mentioned times are in EST".
The two lists disagree on every single day. Monday is 8am to 11am in one and 11am to 12pm in the other. Friday is 7am and 10am against 10am and 2pm. Sunday is 8am to 3pm against 9am to 5pm. Same page, same stated time zone, seven mismatches.
One of them is wrong and the page never reconciles them, which makes any SocialPilot figure you see quoted elsewhere a coin flip between two schedules.
Sprout says avoid the weekend, then schedules it
The third pattern is the one this series keeps finding. Sprout's summary is unambiguous: worst days are Saturdays and Sundays, Saturday is marked "Avoid posting", and readers are told to "avoid Saturday launches entirely".
Its own industry table then gives Education and Schools a best-days value of "Weekends" with Saturday 1pm to 11pm and Sunday noon to 6pm, and a worst-days value of "Weekdays". Food and Beverage gets Saturday 3pm to 7pm. Healthcare gets Saturday 1pm. Travel gets Sunday noon to 2pm.
The same page also calls Sunday a worst day while giving it a 2pm to 6pm window, and excludes Friday from its best days while giving Friday the identical noon-to-6pm window the best days get.
When is the best time to post on X each day?
The four engagement studies, side by side. Do not average these columns. Only SocialPilot is on a genuinely different clock: it reports Eastern Time, while Sprout reports local time and Buffer and Hootsuite both claim times that need no conversion. That matters for the Sprout gap, because it means Sprout's noon really is three hours after Buffer's 9am rather than a time-zone artifact. The best time to post on X tool gives you Sprout's windows on their own if you would rather not choose.
| Day | Buffer (8.7M tweets) | Sprout Social | SocialPilot (main list) | Hootsuite (1M+ posts) |
|---|---|---|---|---|
| Monday | 9am, 10am, 8am | 2pm to 3pm, 5pm | 8am to 11am | 10am |
| Tuesday | 9am, 10am, 11am | 12pm to 6pm | 8am, 10am | 10am |
| Wednesday | 10am, 9am, 11am | 12pm to 6pm | 9am, 3pm | 9am to 11am |
| Thursday | 9am, 10am, 8am | 12pm to 6pm | 8am, 3pm | 9am to 11am |
| Friday | 9am, 10am, 11am | 12pm to 6pm | 7am, 10am | 9am to 11am |
| Saturday | 9am, 10am, 11am | Avoid posting | 8am, 8pm | 8am to 10am |
| Sunday | 9am, 10am, 8am | 2pm to 6pm | 8am to 3pm | 9am to 12pm |
Read down the Buffer and Hootsuite columns and they broadly agree on mornings. Read across any single row and Sprout is somewhere else entirely.
What is the best day to post on X?
Wednesday and Thursday. Take the three days each study names and intersect them rather than counting votes: Buffer says Wednesday, Tuesday, Thursday; Sprout and SocialPilot both say Tuesday to Thursday; Hootsuite says Wednesday to Friday. All four include Wednesday and Thursday. Nothing else survives all four.
That intersection is worth more than the wider Tuesday-to-Thursday reading, because Buffer's contribution here is precisely the ranking whose provenance it does not publish. An intersection of four is checkable against what each page prints; a vote count leans on the one artifact this article has already shown cannot be audited. The best time to post tool lines the same pattern up against the other networks.
Does X say when to post?
Not that anyone can verify. No X-authored timing guidance turned up on any X property, and X's marketing and help sites return errors to automated requests, so their contents cannot be checked either way. Any article telling you "X recommends posting at…" is not quoting X.
What X does publish is far more interesting, and almost nobody uses it.
Does posting time affect the X algorithm?
Yes, and of the platforms covered in this series X is the only one that documents it in code you can read.
X open-sourced parts of its recommendation system. The feature list for its heavy ranker, the model that scores posts for the For You timeline, includes time_since_tweet_creation and time_since_source_tweet_creation, which are the age of the post and of the post it replies to. It also includes time-decayed counts of favourites, replies, reposts and quotes, and features measuring how long since the last engagement of each kind (X).
That is a meaningful contrast with the rest of this series. Instagram files post timing under "mundane information about the content itself". TikTok's published description of its For You recommender lists no time-of-day factor at all. X puts post age in the model and names the variable.
Two limits worth keeping in view. First, X publishes no decay rate: no half-life, no "posts stop being shown after N hours" figure appears anywhere in the repository. Claims of that shape circulate widely and are not X's. Second, the weighting configuration in that repository is dated April 2023, so it describes the system as open-sourced then, not necessarily as it runs now.
Be careful where X timing numbers come from
Two traps specific to this topic.
Posts on x.com are not X guidance. Search for X posting times and you will get x.com/…/status/… results carrying confident hours. Those are tweets, several of them from the very vendors whose studies are cited above, and one is a person's personal impression. An x.com URL in a citation list does not mean the platform said it. The same pattern shows up as instagram.com/popular/… and tiktok.com/discover/… on the other networks, covered in the Instagram and TikTok versions of this question.
The consensus is manufactured. While researching this article, search summaries repeatedly asserted that the studies "converge on 9 to 11am, Tuesday through Thursday". They do not. Hootsuite says Wednesday to Friday. Sprout's window does not open until noon. SocialPilot contradicts itself. The convergence exists in the summaries, not in the sources, which is exactly why the table above prints each study separately instead of averaging them.
How do you find your own best time in 14 days?
The mid-morning answer has three corpora behind it and the afternoon answer has one, so test the disagreement rather than reading more about it.
- Take one slot from each camp. Wednesday 9am represents Buffer, Hootsuite and SocialPilot. Wednesday 2pm represents Sprout. Same day, same content type, six hours apart.
- Add an evening slot if your audience is consumer-facing. Metricool's connected-users peak at 9pm is a real measurement of a different thing, and whether presence converts into engagement for you is exactly the sort of question your own account can answer.
- Hold everything else constant and let a queue fire the slots. Same rough length and topic mix. A slot missed by an hour is a data point lost, which is what recurring weekly slots are for.
- Read every slot at the same post age. X's own ranking features include time-decayed engagement counts, so early response matters more here than on most networks. Analytics refresh every few hours rather than continuously, so compare posts at matching ages instead of at one wall-clock moment.
- Drop the weakest, double the strongest, repeat. Three cycles gives you a schedule built on your audience rather than a vendor's customers.
How do you hold those slots without doing it by hand?
This test has an awkward shape the other networks do not. Both slots fall on the same day, six hours apart, which means one of them always lands mid-morning on a workday. That is the slot people miss, and missing it three weeks running is how a fortnight of comparisons quietly becomes an opinion. A queue that fires both without you is the difference; Zilfu checks for due posts every minute, so a 9am slot goes out at 9am.
Read both slots at the same post age rather than at the same moment. Per-post analytics pull impressions, likes, replies and reposts back from X, and they refresh every few hours rather than instantly, which is exactly why comparing a 9am post and a 2pm post at 6pm would flatter the later one. If a post fails to publish it lands in an Error state with the reason attached, so a broken slot never reads as a weak one.
Driving the whole thing from a script suits X better than most networks, and the API and MCP server are on every plan.
Start mid-morning, then test the afternoon
Three of the four engagement studies put the answer in the weekday mid-morning, roughly 9am to 11am, and three of them agree on Tuesday to Thursday for the day. The day answer is the firmer of the two, because Wednesday and Thursday are the only days all four studies name and that intersection holds whatever you think of Buffer's unpublished ranking.
Then run Sprout's afternoon against it for a fortnight. Zilfu's free plan carries 2 connected accounts and 20 posts a month, which covers two weekly slots for the full two weeks with room to spare.
X is the only network in this series that documents post age in its ranking code. The cross-platform comparison sets that against what the other six platforms disclose.
Sources
All sources retrieved 31 July 2026.
- Buffer. The best time to post on Twitter/X in 2026. Published 13 March 2026, 8.7 million tweets, no collection window stated. Note its page heading says 8.7 million while its title tag still says 1 million.
- Sprout Social. Best times to post on X in 2026. Published 31 March 2026, covering 27 November 2025 to 27 February 2026. Sample spans six networks; no X-only figure is given.
- SocialPilot. When is the best time to post on Twitter/X in 2026?. Published 7 December 2025, around 700,000 posts from 50,000+ connected accounts, no collection window stated.
- Hootsuite. Best time to post on social media. Published 19 November 2025, over 1 million posts across networks; no X-only figure is given.
- Metricool. 2024 X/Twitter study. Published 16 October 2024, 23,561 accounts and 2,144,853 posts. Source for the connected-users measure.
- X. Recommendation algorithm ranking features. Source for post age and time-decayed engagement as model inputs.
- X. The recommendation algorithm. Source for the candidate-generation and ranking pipeline.