Key takeaways
- The large public studies do not agree with each other. Buffer's analysis of 1.8 million videos puts long-form's best slot on Sunday at 10 a.m.; SocialPilot's 301,000-video study points at 2–4 p.m. on weekdays; RecurPost's two-million-video set favours midday. They cannot all be describing the same effect.
- They disagree because they measure publish hour against early views, and early views are not the outcome. Most long-form views arrive later, through Browse and Suggested, on a schedule the launch hour has almost no say in.
- YouTube's position, repeated across its documentation, is that the system cares about how people respond to a video, not when it went up — and that many viewers never look at the publish date at all.
- The one number that is genuinely yours sits in Studio under Analytics → Audience → When your viewers are on YouTube: a 28-day heat map of when your audience is on the platform, in your local time, which is not necessarily theirs.
- Timing does real work in four narrow places: the subscriber notification burst, Premieres and live streams, trend and seasonal races, and avoiding a collision with your own back catalogue.
- Long-form and Shorts want different clocks, because one is announced to subscribers and the other is dropped into an algorithmic feed. Scheduling them together wastes one of them.
Search for the best time to post on YouTube and you will be handed a specific hour within about four seconds. Sunday at 10 a.m. Wednesday at 3 p.m. Weekdays between noon and three. Every one of those numbers comes from a real dataset, some of them enormous, and every one of them contradicts the next. That should be the first clue that the question, as usually asked, does not have the kind of answer people want from it.
It does have an answer. It is just a narrower and less satisfying one: there is a window that is slightly better than the alternatives for your channel specifically, you can derive it from your own analytics in about six minutes, and it will change your view count by a smaller margin than almost anything else you could spend that afternoon on. The reason is structural. A long-form video's view count is mostly decided over weeks by recommendation surfaces, not over hours by a launch rush, and the launch hour touches only the first, smallest part of that.
This piece works through the whole thing: what the big studies actually found, why they had to disagree, what YouTube itself says about publish time, the four situations where the clock genuinely matters, how to read your own window out of Studio without kidding yourself, and how to test a posting time properly — which turns out to be much harder than the advice suggests, and is the reason the advice keeps contradicting itself.
What the big studies actually found
Start with the evidence rather than the conclusion. Four of the most-cited sources are built on genuinely large samples, and they are worth laying side by side, because the disagreement between them is more informative than any one of them.
| Source | Sample | Headline finding |
|---|---|---|
| Buffer | 1.8 million videos | Long-form strongest 8–11 a.m., single best slot Sunday 10 a.m.; Shorts strongest 6–11 p.m., with Wednesday 4 p.m. on top |
| SocialPilot | 301,000 videos across 27,000 channels | 2–4 p.m. local time on weekdays; Wednesday, Thursday and Friday the strongest days |
| RecurPost | Over 2 million videos | Weekday uploads between noon and 3 p.m. earned two to three times the early views of early-morning uploads |
| Hootsuite | Your own last 30 days | No universal answer — the tool computes each account's best times from its own impressions, engagement rate and comments |
Look at the first three rows again. Buffer's best long-form window is the morning and its peak is a Sunday. SocialPilot's is the middle of a weekday afternoon and its peak days are midweek. RecurPost's midday finding sits between them but is stated as a two-to-three-times effect on early views, which is an enormous claim — if publishing at noon rather than 7 a.m. really doubled or tripled a video's performance, it would be the single highest-leverage decision in creating for YouTube, and it plainly is not.
The fourth row is the interesting one. Hootsuite sells scheduling software, has every commercial reason to publish a universal best hour, and instead built a tool that refuses to give one and computes an answer per account from the last thirty days of that account's results. That is a vendor telling you, through its product design rather than its blog, that the universal number does not exist.
Why the studies had to disagree
Three problems sit underneath all of these datasets. None of them is a criticism of the people who ran the analyses — they are properties of the thing being measured.
Early views are not the outcome
Every large timing study has the same shape: bucket videos by the hour they went live, compare some view metric across buckets. To make that tractable at the scale of a million videos, the view metric has to be measured early — first day, first few days. But for most long-form channels the bulk of lifetime views arrives well after that, through Browse and Suggested, and third-party estimates commonly put those two surfaces together at the majority of views for an established channel, with Search a smaller slice that varies enormously by niche.
So the studies are measuring the launch rush, which the publish hour genuinely does move, and treating it as a proxy for performance, which it mostly is not. A video that opens quietly and gets picked up by Suggested in week three does not register as a timing success in any of these datasets, and there are a great many of those. This is the same distinction that trips people up when they read a view graph — the version we worked through in why your views dropped applies exactly here.
Who publishes when is not random
The second problem is confounding, and it is severe. Channels do not choose upload times by coin flip. Full-time creators with editors and a production calendar publish on schedule, in the working day, on midweek days. Hobbyists upload when the edit finally finishes, which is disproportionately late at night. Bucket a million videos by hour and you have partly bucketed them by how professional the operation behind them is.
The 2 a.m. bucket underperforming does not tell you that 2 a.m. is a bad hour. It may be telling you that the sort of channel which finishes an upload at 2 a.m. differs from the sort which schedules one for Wednesday lunchtime in every respect that actually predicts performance: production budget, subscriber base, thumbnail craft, topic selection. No public timing study controls for that, and it is not obvious how one could.
"Local time" is carrying most of the weight
Then there is the phrase that appears in nearly every one of these findings and quietly voids it: local time. Whose local time? A video published at 10 a.m. in London goes live at 5 a.m. in New York and 6 p.m. in Sydney. When a study aggregates across a global corpus, it either normalises to the uploader's time zone — in which case the finding is about creators, not viewers — or to some assumed viewer geography, which for the English-language corpus that dominates these datasets means it is largely a statement about the United States.
If your audience is 70% American and you are in Manila, none of the headline numbers apply to you as written, and applying them literally will put your upload at the wrong end of the day.
What YouTube itself says
YouTube has never published a best upload hour, and its documentation points steadily away from the idea that one exists. Its performance FAQ notes that many viewers do not watch videos in chronological order and do not decide what to watch based on when a video was published — which is the same observation any creator makes the first time an eighteen-month-old video outperforms everything they published that quarter. The system's stated objective is long-term viewer satisfaction, not the promotion of whatever is newest.
Guidance from YouTube's creator-facing channels has been consistent in the same direction: the system is far more interested in how people respond to a video than in the hour it went up. A video published at an odd time can still perform, because the response signals it generates are what drive distribution, and those accumulate over days and weeks rather than in the first ninety minutes.
There is a weaker version of the timing argument that survives all of this, and it is worth stating fairly because it is the one serious people make: publishing when your audience is awake produces a better early sample. If YouTube shows a new video to a slice of your likely viewers and grades the result, you would rather that slice be composed of people who are on the sofa with time to watch than people who are half-awake and about to leave for work. That is a real effect. It is a second-order one — it improves the quality of an early signal, it does not create demand — and it is bounded by the fact that the recommendation system keeps testing your video long after that first window has closed.
The claim that should make you suspicious
Any source telling you a posting time delivers a multiple — twice the views, three times the engagement — is describing a correlation in a dataset it could not control. Effects that large from a variable that cheap do not exist on YouTube. If they did, every channel would already be publishing in that hour, and the advantage would compete itself away within a month.
Where the clock genuinely matters
Having taken the general case apart, here are the four cases where publish time does real work. They are narrow, and all four are mechanical rather than algorithmic.
The subscriber notification burst
When you publish, some of your subscribers get a notification. That is the one distribution event the clock fully controls, because a phone notification at 3 a.m. is a notification cleared from the lock screen without being read. Publish while your subscribers are awake and a meaningful share of them see it in the moment.
The size of that burst is smaller than most creators assume. YouTube's own subscriber notifications documentation is explicit that subscribers choose between all notifications, personalised notifications and none, that the share who have opted into all of them is shown in Studio as its own card, and that not every upload results in a notification being sent. Published estimates put the typical all-notifications share in the single digits — one widely cited figure is around 8.5%. Your own card is the only number that matters here, and it is worth looking at before you rearrange your week around a burst that might involve a few hundred people.
Premieres and live streams
This is where YouTube's own framing is instructive. In the Understand your YouTube audience documentation, the "When your viewers are on YouTube" report is introduced as a tool for building community, scheduling a Premiere and planning a live stream. It is not presented as an upload-time optimiser. That is the correct emphasis: a Premiere or a stream is an appointment, everybody has to be present simultaneously for it to work, and getting the hour wrong empties the chat. An ordinary upload has no such constraint — it waits.
Trend and seasonal races
When demand is time-boxed, being early is worth more than being polished. A reaction to something that broke this morning, a review timed to a product launch, a seasonal video whose search demand peaks on a known date — in all of these you are competing for a window that closes, and hours matter because the videos published before yours accumulate the signals that get them recommended when the surge arrives. This is a content-timing question rather than a clock question, and it is covered from the demand side in our guide to how the algorithm decides what to recommend.
Not colliding with yourself
The least discussed one. If you publish two videos within a short span, they compete for the same subscriber attention, the same Home-feed slots and, on some channels, the same Suggested placements. Creators who move to a heavier upload schedule sometimes find total channel views flat and conclude the algorithm punished them, when what happened is that two videos split one audience's available viewing time. Spacing uploads is a timing decision that pays, and it operates on days rather than hours.
Find your own window in six minutes
Everything above argues that the useful answer is channel-specific. Here is how to get it, using reports you already have.
- Open Studio → Analytics → Audience. The "When your viewers are on YouTube" report shows when your viewers were active over the last 28 days as a heat map, darkest where activity is highest.
- Read the caveat, not just the chart. The report covers your viewers' activity across YouTube as a whole — not their activity on your channel. It tells you when they are reachable, not when they are in the mood for you.
- Check the geography report in the same tab. If your top countries span more than about six hours of time-zone spread, there is no single peak to hit and you should stop optimising for one. Pick the largest single-country block and serve it.
- Note that the heat map is drawn in your local time. If you do not live in your audience's main time zone, do the conversion once and write the result down, because you will otherwise re-derive it wrongly every week.
- Look at the subscriber bell notifications card. It sizes the only audience your publish hour directly reaches. A low number is not a problem to fix; it is a reason to care less about the hour.
- Check your traffic sources. If Browse and Suggested dominate and Search is small, your views are being distributed on a timeline measured in weeks and the launch hour is close to irrelevant. If Search dominates, it is even more irrelevant — search demand does not care what day you published.
That is the whole exercise. It produces a window of two or three hours on two or three days, not a magic minute, and the correct way to hold it is loosely.
Publish before the peak, not at it
One recommendation that recurs across the serious analyses is to publish two to three hours ahead of your audience's peak rather than at it. SocialPilot's 2026 analysis puts it in exactly those terms. The reasoning is mechanical and sound: a freshly uploaded video needs to finish processing its higher-resolution renditions, its metadata needs to be indexed, and YouTube's early testing needs somewhere to start. Arriving with all of that already settled, just as your audience comes online, is strictly better than arriving mid-processing.
The practical version is simpler than the theory. Upload and schedule rather than uploading and publishing. Give the file a couple of hours between arriving and going live. That costs nothing, removes an entire class of avoidable problem — the 4K rendition that was not ready when the first thousand people clicked — and makes the publish moment a decision you make calmly rather than one you make while an upload bar crawls.
Long-form and Shorts want different clocks
The single most consistent finding across the large studies is not a time at all. It is a split: Shorts and long-form peak at different points in the day, and the studies that examine both report it independently. Buffer's data puts long-form in the morning and Shorts in the evening. SocialPilot's reports the same directional split from a completely different sample.
The mechanism is obvious once stated. A long-form video is announced — to subscribers, into the Home feed — and it asks for a block of committed attention, which people have in the evening and at weekends. A Short is not announced to anybody. It is dropped into an algorithmic feed that works like a for-you page, gets shown to a small test audience, and expands or does not based on what that audience does. Nothing about a Short's distribution is chronological, so the only thing publish time changes is who happens to be scrolling when the test runs.
That gives you one concrete scheduling rule worth following: do not publish your Short and your long-form video in the same slot out of convenience. They are competing for different moments in a viewer's day, and pairing them means one of the two is being published at the wrong end of its own curve. If you are building a Shorts habit, the packaging rules differ too — we covered those in the Shorts thumbnail guide.
Defaults for a channel with no data yet
A new channel has no meaningful 28-day heat map, so it needs a starting point. These are the study consensus positions, and they should be treated as an arbitrary but reasonable place to begin rather than as findings that apply to you.
| Format | Starting window | Days to try first |
|---|---|---|
| Long-form, weekday | Early afternoon, ahead of the evening viewing block | Wednesday, Thursday |
| Long-form, weekend | Mid-morning | Saturday, Sunday |
| Shorts | Late afternoon into the evening | Any weekday; midweek slightly favoured |
Note what the table does not contain: a minute. Anyone giving you 3:47 p.m. is selling precision that the underlying data cannot support. Note also that the windows are stated in your audience's time, not yours.
How to test a posting time properly
This is the part almost nobody does, and working through it explains why the public advice is such a mess. Suppose you want to know whether Wednesday at 2 p.m. beats Sunday at 10 a.m. for your channel. To find out, you would need to hold everything else constant while varying only the slot.
Everything else includes: topic, because topic demand swamps every other variable; thumbnail and title, because packaging drives the click; video length and format; where you were in a series; whether the algorithm was already pushing a previous video; the season; and your subscriber count, which is moving the whole time. Two videos are never a comparison. Ten videos per slot, matched on topic and format, is roughly the point at which the noise starts to average out — and by the time you have published twenty matched videos, six months have passed and your channel is a different channel.
Which is why the honest answer is: you probably cannot test this cleanly, so do not spend much on it. Pick a window from your own heat map, keep it, and put the energy into the variables you can test cleanly. Thumbnails are the obvious one, because YouTube gives you a native three-way test and the comparison runs on the same video at the same moment, which eliminates every confound listed above in one move. Our thumbnail A/B testing guide covers how to run those so the result means something.
What consistency is actually doing
"Post consistently" survives every algorithm change, and it is worth being precise about why, because the usual explanation — that the algorithm rewards regularity — is not a mechanism YouTube has described.
Consistency works on the audience side. A viewer who knows roughly when your videos appear develops a habit around them. Subscribers who see you in their feed on a predictable rhythm are less likely to drift. And an upload schedule you can actually sustain produces more videos, which is the only reliable way to give the recommendation system more chances to find an audience for you. SocialPilot's 2026 analysis suggests a sustainable mix of one to two long-form videos plus three to five Shorts a week, with the median active channel publishing around twelve videos a month — a useful sanity check if your plan is more ambitious than that.
The corollary matters more than the rule. A schedule you miss is worse than a looser one you keep, and the day is a far stronger habit cue than the hour. Nobody has ever unsubscribed because a video landed at 3 p.m. instead of 2 p.m. Plenty have drifted because the Thursday video stopped appearing on Thursdays.
If the timing question is really a views question
Most people searching for the best time to post are not curious about scheduling. They have a video that underperformed and they are looking for the variable that explains it. Timing is an appealing candidate because it is cheap to change and requires admitting nothing.
The diagnostic is quick. Open the video and look at impressions and click-through rate together. If impressions were served and few people clicked, the hour was not the problem — the packaging was, and the fix is the thumbnail and the title. If impressions themselves never materialised, the problem is upstream of packaging: topic demand, or a video that failed its early test on retention. Our guides to what counts as a good CTR and writing titles that get clicked deal with the first case; the first thirty seconds deals with the second.
Ranking those by leverage, the publish hour comes last by a distance. A thumbnail that lifts CTR from 4% to 6% is a 50% increase in clicks from the same impressions, compounding for as long as the video is served — an effect no posting time has ever plausibly matched. That is the argument for spending your optimisation effort on packaging, and it is why Thumblore exists: generating and comparing thumbnail variants is fast enough that you can afford to make several and let the data pick, rather than shipping the first one and wondering afterwards whether the hour was wrong.
So the answer to the question, in full: publish inside the window your own Audience tab shows you, an hour or two before your audience's peak, on a day you can keep to, with long-form and Shorts on separate clocks. Then stop thinking about it. The hour is a small, real, quickly exhausted optimisation, and everything meaningful about how a video performs is decided by what is inside it and by the thumbnail and title that persuade anyone to open it in the first place.