YouTube Views

What does the recommendation system actually measure?

Time, not tallies. The two surfaces that send a video most of its traffic — the home feed and the suggested column beside the player — are built on expected watch time: YouTube estimates how long a particular viewer will stay if it puts your thumbnail in front of them, and it ranks candidates on that estimate. Click-through rate feeds the estimate and average view duration feeds it, and the two are always read together because either one alone is trivial to inflate.

There is a second layer above it that gets far less attention. YouTube is not only asking whether someone watches your video; it is asking what happens afterwards — whether the viewer carries on watching on the platform or closes the tab. A video that hands the viewer straight into the next one is worth more to the system than a video with identical retention that ends the session, which is why recommendations tend to favour material that sits inside a subject people work through in a sitting.

The counter under the player is an output of all this rather than an input to it. It records what happened; it does not tell the system what to do next. That distinction is the most useful thing to hold on to on this page, because it explains both what buying views does well and what it leaves untouched, and every section below is downstream of it.

Do bought views help a video rank?

Only indirectly, and less than on any other platform. YouTube's ranking is built on expected watch time — click-through rate and average view duration read together — because either one alone is trivial to game. A view that arrives and leaves immediately lowers the average it joins, which is the opposite of what the buyer wanted.

What a view count does do is change human behaviour. A video showing 40 views and one showing 4,000 get clicked at different rates from the same thumbnail, and the second gets the benefit of the doubt in the first ten seconds. That is a real effect, and it is the honest reason to buy.

The effect is largest exactly where the click decision is hardest. Someone scanning a suggested column sees four pieces of information about each candidate — thumbnail, title, channel name and view count — and one of them is a number that reads as a verdict other people already reached. Search results present the same four. In both places your video is standing next to alternatives, and a count that says "people watched this" does work that no amount of title-writing does on its own.

What follows is a rule of proportion rather than a rule of size. Views bought against a video that holds attention give real viewers a reason to click something that will then keep them, so the count borrows credibility and the content pays it back. Views bought against a video nobody stays for raise the ceiling on a room that is already emptying. The same order produces two different results, and the video decides which one.

What counts as a view in the first place?

A play that a person started and stayed with. YouTube has never published the exact threshold, and the figure the industry has settled on is roughly 30 seconds of genuine playback from a viewer who did something deliberate to begin it. A muted autoplay in a background tab is treated differently from a press of the play button, and a page that reloads itself ten times does not produce ten views.

The counter also pauses to check itself. On videos gathering views quickly the public number famously sticks near 301 while validation catches up, and the modern equivalents of that pause happen at larger numbers without being as conspicuous. Studio's realtime card and the public counter disagree during those windows by design: one figure is provisional and the other has been checked.

Which is why a delivery is best read over days rather than hours. The number visible an hour in is the part that has cleared so far, the analytics figure and the public figure converge later, and a shortfall is only a shortfall once validation has finished with it. The refill window described below is written around that timing rather than around a screenshot.

Will this get me to monetisation?

Read this part before ordering, because it is where most money is wasted. The Partner Programme needs 1,000 subscribers plus 4,000 valid public watch hours in 12 months — or 1,000 subscribers plus 10 million Shorts views in 90 days. Those are two separate routes and the watch-hours one has strict conditions.

Watch hours count only from public, long-form videos. Shorts do not count toward the 4,000 hours. Streams you later delete do not count. Private or unlisted uploads do not count. Buying views against any of those produces a number on the screen and nothing in the monetisation tab — open YouTube Studio and check which requirement you are actually short on before you spend anything.

The arithmetic is worth doing before the order rather than after it. Four thousand hours is 240,000 minutes. A video whose average view duration is four minutes needs about 60,000 genuine views to produce them; one averaging 40 seconds needs closer to 360,000. Average view duration, not video length — a 20-minute upload watched for two minutes contributes two minutes. Studio prints that figure for every video, so the calculation takes a minute and it tells you whether the route you are on is a short walk or a long one.

That is also why YouTube watch hours is a separate product from views. Views are counted per play; hours are counted in minutes retained, and only from public long-form uploads inside the rolling window. If the bar you are trying to fill is the hours bar, ordering against the hours is the direct route and buying views is the indirect one.

Which requirement am I actually short on?

Open YouTube Studio, go to Earn, and read the two bars. One counts subscribers, one counts watch hours, and they move independently. Buying against the bar that is already full is the single most common way money is wasted on this page, and it is entirely avoidable by looking first.

If subscribers are the gap, views will not close it — subscribers is the separate count. If watch hours are the gap, check that the videos carrying your hours are public and long-form before ordering anything, because those are the only ones the counter accepts.

The window matters as much as the number. The 4,000 hours are counted across the previous 12 months on a rolling basis, so hours earned 13 months ago drop off the back as new ones arrive at the front, and a channel that has been quiet can watch the bar fall while doing nothing at all. Subscribers do not expire that way. Reading both bars on the same afternoon tells you which of the two clocks you are actually racing.

What about Shorts?

Shorts are a separate economy inside the same platform. They have their own monetisation route — 1,000 subscribers plus 10 million Shorts views in 90 days — and their views do not count toward the 4,000 watch hours the long-form route needs.

That makes the decision unusually clear. If your channel is Shorts-first, buying long-form views does nothing for your qualification, and vice versa. Pick the route your content actually fits and buy against that requirement only.

The two formats are also counted differently at the moment of playback. A Short registers a view when it starts playing in the feed, with no click decision in front of it, so the number climbs from scroll traffic a long-form video would never see. A long-form view requires somebody to have chosen the video first. Two counters wearing the same name, produced by two different behaviours, and only one of them accumulates the minutes the hours bar reads.

So if the channel is Shorts-first, Shorts views is the matching delivery and the 90-day route is the one in play — 10 million inside a rolling window, a volume that in practice only a feed hit produces. For most channels the useful version of this is simpler: decide which format your content genuinely is, and buy against the bar that format fills.

How fast should views arrive?

Gradually. YouTube compares a video's traffic sources and pacing against what is normal for that channel, and a video that goes from 200 to 50,000 in an hour on a channel that averages 800 is the pattern that invites review. Delivery here is paced for that reason rather than dumped.

The same logic applies to the channel as a whole. Views should track your own history, not your ambitions — a single video wildly out of line with everything around it draws exactly the scrutiny nobody wants, from viewers, from sponsors, and from YouTube itself.

It helps to know what the system sees while it happens. Studio labels every view by traffic source: browse features, suggested videos, search, external sites, and direct or unknown. An ordinary upload draws a mixture, weighted toward browse and suggested if the channel has an audience and toward search if it answers a question. A delivery arrives on the external and direct side of that split, so the mixture stays plausible when the delivered volume is a share of the video's traffic rather than all of it.

Which sets the pace in practical terms. Match the daily arrival to something the video could have earned on its best day — a few multiples of it, not a hundred — and let the order run over the number of days that requires. A video already climbing can absorb a faster delivery because the shape of its curve already bends upward; a video sitting flat is better served by a slice a day, which lands as a change in slope rather than as a step.

How many views should I buy?

Enough to change the decision a stranger makes, which is usually a smaller number than people order. A video sitting at 200 views and one at 5,000 are read very differently; 5,000 and 40,000 are read almost identically by anyone who is not already a fan of the channel. The return curve flattens quickly, and most of the value is bought in the first few thousand.

Two anchors make the choice concrete. The first is your own median: a video that lands at two or three times the median of your last ten uploads reads as one that travelled, while forty times the median reads as one that was helped. The second is the channel's ceiling — if your best upload ever reached 12,000, a new one at 90,000 raises the question rather than answering it.

The other half of the sizing is what the count has to live beside. Likes, comments and the retention graph all sit under the same player, so a sensible order is one those numbers can plausibly accompany. In practice that means a smaller view order with likes arriving alongside it rather than a larger one on its own — less money for a result that holds together when someone looks twice.

What does it cost and what is required?

The per-thousand rate falls as the order grows, and the full ladder is visible above before you commit. All we need is the public video URL — no password, no channel access, no app authorisation.

The video has to stay public and stay up. Deleting it, switching it to private or unlisted, or changing the channel handle mid-order all remove the target, and a refill cannot be issued against a video that no longer exists.

The falling rate is a cost structure rather than a discount. The fixed work behind an order — placing it, confirming the URL resolves, watching the delivery, holding the refill window open — is the same for 1,000 views as for 100,000, so it lands lighter per unit on the larger order. Pace is the second input: a delivery spread across a week occupies the queue longer than one that finishes in a day, and the tiers reflect that in both directions, which is why the cheapest configuration is not automatically the one you want.

The figure worth comparing between suppliers is not the price per thousand delivered but the price per thousand still counted a month later. A cheap order that sheds a fifth of its views during validation costs more per retained view than a dearer one that holds, and it costs the average view duration on top. The 30-day refill is what converts the second number back into the first.

Do the views stay?

Most do. YouTube runs its own validation and periodically discards views it judges invalid, which is normal platform behaviour and happens to organic traffic too. The 30-day refill window covers a shortfall at no cost.

Anyone promising that no view will ever be removed is describing a process they do not control. The honest promise is the refill, not permanence.

The mechanism is worth picturing, because it explains the timing. Validation runs in passes rather than continuously, so a video can sit at its delivered number for a fortnight and then adjust in a single step when a sweep reaches it. The figure to watch is the one in Studio's analytics rather than the public counter on the watch page, since the two update on different schedules and the analytics number is the settled one.

Claiming a refill is deliberately plain: the order number and the current count are enough, because the delivered quantity is on record and the comparison is done at our end. Most reports arrive in the second or third week, which is when a sweep usually catches up with a batch, and the window is set at 30 days for exactly that reason.

What decides whether the video stays up?

What is in it. YouTube acts on videos for their content and on channels for their record — the community guidelines, copyright claims and strikes, and the reused-content and authenticity rules that govern monetisation. Incoming traffic sits outside that list for a structural reason worth stating plainly: anybody can send traffic at any public video, including traffic the owner never arranged, so arriving views cannot be treated as evidence of what the owner did.

Where the platform does act is on the views themselves. Its validation systems discount plays they judge invalid and the number on the video adjusts to match. That is a correction to a counter rather than a mark against a channel, and it is the same process that trims organic traffic arriving from bot-heavy referrers.

The surface that deserves more care than either is monetisation eligibility, which is reviewed by people as well as by systems. What a review reads well is a channel whose numbers agree with one another over time — views, watch time, subscribers and engagement all pointing the same way. Keeping them in proportion is the practical step that makes that review uneventful, which is the only outcome anybody wants from it.

Are cheap views different from expensive ones?

Yes, and on YouTube the difference matters more than on any other platform, because the metric being measured is duration rather than a count. Cheap views are delivered fast and leave immediately; they raise the counter and drag the average view duration down with them.

Views that hold for part of the video cost more because they are harder to source, and they are the only kind that do not damage the number YouTube actually ranks on. If a page sells "1 million views for a few dollars", the arithmetic tells you what is behind it before the description does.

What the higher price buys is specific rather than vague. It buys duration — a play that stays through part of the video instead of a fraction of a second. It buys sourcing, meaning traffic that arrives from somewhere plausible rather than from one address a thousand times. And it buys pacing, because a delivery that can be spread across days costs more to run than one dumped in an hour. Those three are the whole of the difference, and all three are visible afterwards in the video's own analytics.

Which makes the sensible approach a small test first. Order a modest quantity against one video, then read three numbers a week later: how many views survived validation, what happened to the average view duration, and where the traffic-source split moved. That tells you more about a supplier than any description on any page, and it costs the price of the smallest tier to find out.

Should I buy for an old video or a new one?

An older video that already retains well is often the better target. Its retention curve is proven, its metadata has settled, and YouTube has had time to work out who it is for — a lift in views there arrives on a video the system already knows how to distribute.

A new upload is the riskier choice, because the first days are when YouTube is testing it and when unusual traffic is most visible against the channel's baseline. If you buy for a new video, keep it small and let it arrive over days rather than hours.

There is a particular kind of older video worth looking for. Open the analytics for the last 12 months, sort by average view duration rather than by views, and find the upload that holds people longest while getting the least traffic. That video is the one the system has under-distributed, its retention is already proven, and it is the one where an additional thousand views has somewhere to go — a video that keeps its audience feeds the session, and the session is what the recommendation surfaces are built to protect.

What should the video itself look like before the order?

Finished. The money buys attention arriving at a page, and everything deciding what that attention does next is set beforehand: the thumbnail that earns the click, the title that agrees with it, and the first thirty seconds that decide whether the retention graph starts flat or falls away.

The opening is where the leverage is concentrated. Studio's retention curve almost always shows its steepest drop in the first fifteen to thirty seconds, because that is when viewers decide whether the video is what the thumbnail promised. An opening that delivers on the title immediately keeps a measurably larger share of arrivals than one that spends the window on an intro animation and a greeting.

The furniture around the video is worth a few minutes too: chapters so a long upload can be entered where it is useful, an end screen pointing at the next video so the session continues, a pinned comment saying what the video is for, and a description carrying the links a viewer will look for. Each of them raises the share of arriving traffic that turns into something, which is precisely the traffic this order is buying.

Do views bring subscribers with them?

A few, sometimes, and not reliably enough to plan around. The conversion from viewer to subscriber on a normal channel is somewhere around one in a hundred, and bought views convert at close to nothing because there is no person behind them deciding they want more.

What views do is put the video in a position where real viewers arrive — a higher count makes the thumbnail worth clicking, and clicks from real people convert at the normal rate. The subscribers come from the content once the traffic is there, which is the same order of operations that works organically.

The place that decision gets made is worth knowing, because it changes what you optimise. Subscribe clicks cluster at the end of a video and immediately after it, when someone has just been given something they valued and wants the next one. That means retention drives subscribers far more directly than the view count does: a video that keeps half its audience to the end converts several times better per view than one that keeps a tenth, at the same traffic.

Will this affect my ad revenue or CPM?

Views that do not watch produce no watchable ad impressions, so they add nothing to revenue. That is the neutral case. The case worth avoiding is a large volume arriving fast on a channel that normally earns steadily, because the shape of the traffic changes and YouTube's systems compare a video's sources against the channel's own history.

For a monetised channel the sensible order is small and gradual, or not at all. For a channel that is not monetised yet there is no revenue to disturb, which is why views are a more reasonable purchase before the threshold than after it.

There is a reporting effect worth expecting so it does not surprise you. RPM divides your revenue by every view the video received, monetised or not, so views that show no ad enlarge the denominator and print a lower RPM against the same money. CPM behaves differently because it is calculated only on the impressions that actually ran. Nothing has been lost when RPM dips after a delivery; a different set of views is being divided into an unchanged amount of revenue.

Which suggests the timing most channels find comfortable. Buy before the threshold, when the counter is doing credibility work and there is no revenue figure to reinterpret. After the threshold, keep orders proportionate and spread, and read RPM across the quarter rather than the week — that is the horizon on which the number means anything.

How do I keep the channel looking consistent?

Let the numbers around the video agree with it. A video at 50,000 views with 12 likes and no comments reads as bought to a viewer before it reads that way to any algorithm — and the like count is right there under the player.

If you are lifting a video's views, lift the likes with it in proportion; somewhere around 2 to 4 per hundred views is the ordinary range on YouTube. Keep the rise gradual and keep uploading through it, because a channel that publishes nothing for a month while one video climbs is a pattern that explains itself.

Comments follow a wider band and a lower one — roughly one per two or three hundred views on a channel where people talk, far less on one where they do not — so they are worth adding sparingly if at all, and worth answering when they arrive. YouTube likes is the count that does the most work for the least money here, because it sits directly beneath the player and it is the number a viewer checks first when a view count surprises them.

At channel level, the thing to preserve is the relationship between videos. One upload at ten times the median with everything else unchanged is conspicuous; the same lift spread across the two or three videos you would most like people to find looks like a channel that gained traction. Keep publishing on your usual schedule while it happens, and the pattern explains itself in the ordinary way.

Do views work differently on a live stream or a premiere?

Yes, because a stream carries two counters. During the broadcast YouTube shows concurrent viewers, the number watching at that moment, and once it ends the archive carries an ordinary view count like any other video. They measure different things and only the second one is still there tomorrow.

The hours a stream earns count toward the 4,000 only while the recording stays public. Deleting the archive after the broadcast, or letting an unsaved stream expire, removes those hours along with it. Streamers who want the hours to count keep the archive up and public, which is also what lets the stream keep collecting views for months afterwards.

For a premiere the moment that matters is the countdown, when the chat is open and early arrivals are deciding whether the room feels worth waiting in. A count that reads as attended pulls more of the waiting audience through the door. After the premiere ends the video behaves exactly like any other upload, with the same retention graph and the same rules, so anything ordered against it should be sized as if it were one.

How do I know whether it worked?

Write down four numbers before you order and read them again afterwards. Views is the obvious one. The other three are impressions click-through rate, average view duration and the traffic-source split, all of them in the advanced view of a video's analytics, and together they tell you what the count actually did.

The reading is straightforward. If click-through rate on browse and suggested traffic rises in the days after the delivery, the count is doing its job and strangers are choosing your thumbnail more often than before. If average view duration holds roughly where it was, the delivery has not diluted the metric the system ranks on. If subscribers per thousand views stays near your usual rate, the real traffic arriving alongside is behaving normally.

Give it a fortnight before deciding anything. Impressions and suggested traffic move on a lag, validation finishes inside the same window, and a single day of data after a delivery is mostly noise. The comparison that carries meaning is the video's own thirty-day shape set against the thirty-day shape of the uploads either side of it.

Where does buying views sit next to the other ways of growing a video?

Underneath them. A better thumbnail costs an afternoon and changes the click rate on every impression the video will ever receive. A collaboration puts you in front of an audience that already watches your subject. YouTube's own ad platform will place the video in front of a chosen audience at a known price, which is the paid route with the most control over who arrives.

Every one of those ends at the same moment: somebody looking at a thumbnail, a title and a view count, deciding in about a second. That is the moment this purchase is aimed at, which is why it works as a layer under a plan rather than as the plan itself. Spend on traffic while the count reads as empty and you pay for impressions that do not convert; spend on the count with nothing bringing people past it and there is nobody there to read it.

The sequence most channels find workable runs: fix the thumbnail and the first thirty seconds, lift the count on the video worth lifting, add likes so the numbers agree, then point traffic at it from wherever your audience already is. Each step lowers the cost of the next, and the count is the one that can be finished this afternoon. The YouTube service overview lays out the rest of what a visitor sees once they arrive.

What the view count carries, and what the video keeps

The count carries the first second. It is one of four things a stranger reads before choosing your video over the one beneath it, it is what a sponsor filters on before opening anything, and it is what makes a thumbnail worth a click on a channel nobody has heard of yet. On a platform where most traffic arrives through a comparison between candidates, winning that comparison is most of the job.

What the video keeps is everything measured after the click. The audience-retention graph is drawn by the people who stayed — visible to you in Studio and to any sponsor who asks for a screenshot, showing precisely where attention went. A high count sitting on top of a curve that holds tells a strong story, and a smaller number with a flat line tells a better one than a large number with a curve that collapses at ten seconds. The curve is written by the video, and it is the part the purchase hands back to you.

The same division applies to eligibility. Views move the counter, while the hours bar reads minutes retained from public long-form uploads and the subscriber bar reads a separate decision people make at the end of a video — YouTube subscribers is where that count is ordered, and it moves on its own terms. Buy the count for the comparison it wins, keep the video worth winning it with, and every number under the player ends up describing the same thing.