The best time to post on TikTok: four studies, eleven million posts, and one window they all agree on

Post on Friday between 5 and 8 p.m. in your audience's local time. That is not the loudest answer on this topic, it is the only one that survives reading all four of the large public studies instead of one of them. Buffer measured 7.1 million posts, Sprout Social measured about 2 billion engagements, Metricool measured 2.3 million and RecurPost measured over 2 million, and on every other day of the week at least one of them contradicts at least one other. Saturday is Buffer's single best day. Sprout Social tells you not to post on Saturday. Metricool calls Saturday the quietest day it measured. Those are three honest reads of large datasets reaching three incompatible conclusions about one day, and no guide that quotes only one of them can tell you that.

What the four studies actually found

Set the four side by side and the disagreement is not at the margins, it is structural. Buffer found its single strongest slot on Sunday at 9 a.m. and its weakest days in the middle of the week. Sprout Social found the opposite shape entirely: Tuesday through Thursday, 2 to 6 p.m., with the weekend written off as what it calls an algorithmic dead zone. Metricool put the peak of the entire week on Sunday at 8 p.m., eleven hours after Buffer's Sunday peak and on a day Sprout Social names as the worst of the seven. RecurPost landed on Friday and Saturday evenings around 8 p.m., which agrees with Metricool on the hour and with nobody on the day.

Day by day, with the disagreement printed

Most people arriving here are not asking about the week, they are asking about one day, usually because they are about to post on it. So here is every day with our reading of where the four studies overlap in the second column, and each study's own published window in the four columns after it. That second column is the only figure in this article nobody measured, which is why the evidence for it sits on the same row: check our arithmetic rather than taking it.

Wednesday, the day with the second best case

Wednesday is the most searched day of the week on this question and it has a genuinely defensible answer: 7 to 9 p.m. Buffer's top Wednesday slot is 10 p.m. with 9 p.m. third, Metricool puts it at 7 to 8 p.m., RecurPost has 9 p.m., and Sprout Social's Wednesday window is a wide 1 to 8 p.m. that contains the bottom of that range. Three of the four converge on the two hours after seven and the fourth does not exclude them. The complication worth knowing is that Buffer also names Wednesday as one of its two weakest days overall, so this is a good hour on a day one large study thinks you should not have chosen.

Tuesday, where the data falls apart

Tuesday has no answer and anyone who gives you one confidently has read a single study. Buffer's best Tuesday hour is 6 a.m. Metricool's is 8 p.m. Those are fourteen hours apart on the same day of the same week. Sprout Social and RecurPost both land in the afternoon, which is not either of the other two. If you have to post on a Tuesday, the afternoon is the least contested part of it, and that is the strongest sentence the evidence supports.

Monday, and the two peaks

Monday is the only day where two studies name the same hour independently: Buffer and RecurPost both put 1 p.m. at or near the top, and Buffer's Monday 1 p.m. is its second best slot of the entire week. Against that, Metricool and RecurPost both find a second peak after 7 p.m. So Monday genuinely has two windows rather than one, which is unusual and useful: early afternoon, or the evening after eight.

Thursday, split down the middle

Thursday splits into an early afternoon cluster and a late evening one. Buffer's best Thursday hour is 1 p.m. and Sprout Social's window is 1 to 5 p.m., so those two agree. Metricool says 8 p.m. and Buffer's second choice is 10 p.m., so those two agree with each other instead. Pick one and hold it rather than alternating, because the whole point of a slot is that an audience learns it.

Friday, the one everybody agrees on

This is the finding worth having. Buffer's top Friday hour is 6 p.m. with 8 p.m. third. Metricool says 8 p.m. RecurPost says 5 p.m. and 8 p.m. Sprout Social says 3 to 5 p.m., which touches the bottom of that band. Four studies, four methods, four datasets, and their Friday windows overlap in the same three hours. Nothing else in this article is that well supported, and if you only ever schedule one slot a week, this is the one the evidence actually earns.

Saturday, the day three studies fight over

Saturday is the clearest illustration of why this genre of article should be read sceptically. Buffer names it the best day of the week. Metricool names it the quietest. Sprout Social says do not post. If you post anyway, Buffer's 3 to 5 p.m. and RecurPost's 3 p.m. give you a mid afternoon window that two of the four support, and Metricool's 6 to 7 p.m. is close behind. What nobody can give you is confidence.

Sunday, best and worst at the same time

Sunday holds both Buffer's single strongest slot of the week, 9 a.m., and Metricool's, 8 p.m., and Sprout Social's verdict that it is the worst day of the seven. Two studies place their peak of the entire week on a day a third tells you to skip. The only part of Sunday with more than one study behind it is the middle: Buffer's second and third Sunday choices are 1 p.m. and noon, and RecurPost's are noon, 2 p.m. and 4 p.m. If you post on a Sunday, post at lunchtime.

What TikTok itself says, which is less than you would expect

TikTok publishes an explanation of how the For You feed recommends videos, and it names three kinds of signal: user interactions, meaning the videos you like, share, comment on and create; video information, meaning captions, sounds and hashtags; and device and account settings, meaning language, country and device type. Read that list again and notice what is not on it. Posting time is not a signal. Time of day is not a signal. Recency is not named as a signal. TikTok also states plainly that neither follower count nor a history of high performing videos is a direct factor in recommendations, which is a stronger claim than most platforms are willing to make about themselves.

What TikTok does say is that stronger indicators of interest carry more weight than weak ones, and the example it gives is whether somebody finishes a longer video from beginning to end. That is the sentence to build a strategy on. It means the system is reading completion, and completion is a property of the video, not of the hour you uploaded it. The four studies above are measuring when audiences happen to be awake and scrolling, which is real and worth using, but they are not measuring a rule inside the algorithm, because on the platform's own account there is no clock in there to please.

The UTC trap in your own analytics

Your own follower activity data outranks all four studies, because it is the only one measuring your audience. On a Creator or Business account it sits under Profile, then the menu, Creator tools, Analytics, and the Followers tab, as an hour by hour and day by day view of when your followers are actually on the app. Before you schedule anything around it, check one thing. Multiple third party guides report that TikTok presents these hours in UTC rather than in your local time, and that the desktop view and the in app view do not always agree. We could not confirm that in TikTok's own help documentation, which is exactly why it is worth checking yourself rather than trusting either us or them.

Why every guide gives you a different hour

Because they are downstream of a handful of datasets like these four, and the caveats fall off in the rewriting. A median engagement rate pooled across seven million posts becomes "post at 9 a.m. on Sunday", loses the words median and pooled on the way, and lands on your desk as an instruction. Then a second guide does the same thing to a different dataset, contradicts the first, and both put the current year in the title. Notice also the fifth row of the first table: one of these four publishes its hours in US Eastern time. A global median expressed in one country's clock is a much weaker claim than the same median expressed in each audience's own local time, and you would never know from the article which kind you were reading.

The same disease runs through every platform's timing folklore, and it is the reason our piece on YouTube timing ends where this one does, and why the honest answer to how often a small business should post is also less precise than the internet's answer. It is the same shape as the question about hashtags: a practice everybody follows, a confident number attached to it, and a body of evidence underneath that turns out to be thinner than the advice built on top. The same four study treatment for Instagram is here, where the spread between studies is wider than on any other platform in this cluster.

The thing that moves more than the clock

Buffer, having ranked all 168 hours of the week, ends its own study by saying that the quality of the content will always be the most important factor and that timing is rarely the main reason a video underperforms. That is the study author telling you the study is not the point, and it agrees exactly with what TikTok says about completion carrying more weight than weak signals. So the arithmetic is uncomfortable but simple: the hour decides how many people are awake when the video enters its first small test audience, and the first two seconds decide whether that audience stays long enough for the test to pass.

Which means the useful work is upstream of the calendar. Finding out what already holds attention in your niche is a better use of an afternoon than any further study of the clock, and the tools that mine what is already working on TikTok exist for that. So does writing an opening line that earns the second second, which is the same craft as writing a caption that survives the first line, compressed into speech. And if you are staring at an empty calendar rather than a suboptimal one, the harder question is what a small business has to say at all, not when to say it.

What to actually do on Monday morning

• Take Friday 5 to 8 p.m. as your anchor slot, in your audience's local time rather than your own. It is the only window in this entire article with all four studies behind it.
• Add Wednesday 7 to 9 p.m. as the second, which three of the four support, and stop there. A third slot chosen from this evidence is a coin toss with extra steps.
• Open your Followers tab and work out the timezone offset before you trust the heatmap. Post at a time you wrote down and see which hour it appears on. Then shift every number in the chart by that difference.
• Post one to two hours before your own peak, not at it. The video needs to already exist and have collected its first signals when the wave arrives, not be uploading into it.
• Hold the slots for eight weeks before judging them. Consistency is the one timing effect nobody in the literature disputes, and Buffer says as much in its own conclusion: posting consistently matters more than posting every day.
• Then stop optimising the clock and go and shorten the first two seconds. TikTok says completion is a strong signal. It has never said the hour is any signal at all.

So: Friday evening, Wednesday evening second, your own heatmap read in the right timezone over anybody's median, and the hours you were going to spend hunting a better hour spent on the two seconds that decide whether anyone watches to the end. If TikTok is one part of a wider presence rather than the whole of it, how the rest of that presence gets found is the companion to this piece, the same four study comparison for Facebook shows a field that agrees far less than this one does, and the vertical 9:16 frame itself is worth getting right before you worry about when to publish inside it.