Does Buying YouTube Views Work? What Purchased Traffic Can and Cannot Move

The pitch is always some version of the same sentence: the algorithm favours videos that already have traction, so buy the traction and the rest follows. It is a coherent theory. It is also testable, and the last two years have made it much easier to test, because YouTube has published enough about how views, impressions and earnings are counted that you can work out exactly what a purchased view can and cannot touch.

The short answer is that it moves one number and only one number, and in August 2026 YouTube redefined that number in a way that made it cheaper than it has ever been. Everything downstream — distribution, earnings, Partner Programme eligibility, the analytics you would use to decide what to make next — either ignores the purchase or is actively damaged by it.

This piece works through the mechanism rather than the morality. What you are buying, which reports it lands in, what the research says happens to bought views over the following days, what the policy says when enforcement arrives, and what the money would do instead if you spent it on the part of the funnel that is genuinely purchasable.

What a view has meant since 24 August 2026

The single most important fact about buying views in 2026 is that the view itself was devalued last month, by YouTube, on purpose.

From 24 August 2026, the public view count on a long-form video, a Short or a live stream increments from the first frame of playback. There is no minimum duration. The stricter, older definition — a play where the viewer stayed past the opening seconds or acted on the video — survives under the name engaged views, and lives in YouTube Analytics under Advanced mode. The change was not applied retroactively, so older videos keep their old numbers.

YouTube was unusually direct about the consequence: earnings and Partner Programme eligibility continue to run on engaged views and engaged watch hours, not on the public number. Two counters now exist because they do two different jobs. The public one is a display metric. The engaged one is the one attached to money.

A seller of views can only ever deliver against the first of those. Whatever the delivery method — bots, incentivised clicks on a traffic exchange, an autoplaying player embedded on a piracy site — the product is a playback event, and a playback event is precisely what YouTube has just declared to be worth the least. The market is selling the metric on the day its price was cut.

The number that decides distribution cannot be bought

Views are not the input to the recommendation system. They are its output. What the system actually spends on you is impressions, and impressions have a specific definition that rules purchased traffic out by construction.

An impression is counted when your thumbnail is displayed on YouTube — the home feed, search results, suggested videos, the Shorts feed and comparable surfaces — for more than one second and with at least half of it visible. YouTube's own documentation on impressions and click-through rate is explicit that this covers impressions on YouTube only, not on external sites or apps, and that some placements are excluded from the CTR metric altogether.

Follow that through. Traffic delivered by a third-party service does not arrive through a YouTube surface. It arrives as external or direct traffic, from a player somebody else loaded. It therefore generates no impression, which means it cannot raise your click-through rate — and cannot lower it either, contrary to a widely repeated claim that bought views tank your CTR by inflating the denominator. The denominator never sees them. The purchase happens entirely outside the funnel it is supposed to influence.

Metric	Can purchased traffic move it?	What actually happens
Public view count	Yes, temporarily	Rises, then is corrected downward in batch when validation catches up
Impressions	No	Only counted on YouTube surfaces; external players are excluded by definition
Impressions CTR	No	Neither numerator nor denominator is touched by off-platform traffic
Engaged views	Effectively no	Requires sustained, intentional playback — the thing cheap delivery cannot fake at scale
Average view duration	Yes, downward	Short synthetic sessions drag the average and distort the retention curve
Valid public watch hours	No	Invalid traffic is excluded from eligibility metrics
AdSense earnings	Yes, downward	Invalid traffic revenue is reversed, before or after disbursement

The honest version of the sales pitch, then, is that you are paying for a number on the watch page that no system inside YouTube treats as evidence of anything. If you want a fuller account of how impressions and CTR interact — including why a falling CTR is often a sign a video is winning — we covered it in what a good CTR actually is.

Corrections are a schedule, not a verdict

The most useful research on this subject is not about buyers at all. It is about how routinely YouTube revises its own numbers.

A 2024 paper in Scientific Reports by Maria Castaldo, Paolo Frasca, Tommaso Venturini and Floriana Gargiulo, Fake views removal and popularity on YouTube, tracked around a thousand French channels for eighteen months and watched the public view counts move. Corrections affected the large majority of channels and more than 78% of the videos in the corpus. They were not applied continuously as views accumulated but in batches, generally around five in the afternoon. And they tended to arrive late in a video's life, after it had already collected most of its audience — with later corrections concentrated on the more popular videos.

Two things follow, and they point in opposite directions for different readers. If you have never bought anything and your count ticks downward, that is the ordinary operation of a validation system that touches almost every video, not a signal that you have been flagged. If you have bought something, the finding to sit with is the batching: the correction does not arrive as you buy, which is exactly why the purchase looks like it worked for a day or two.

The other paper worth knowing is Dhruv Kuchhal and Frank Li's A View into YouTube View Fraud, presented at the ACM Web Conference in 2022. They spent nine months studying a large view-fraud operation that used a piracy streaming site to force unsolicited YouTube videos to play as pre-rolls, and observed hundreds of thousands of videos across tens of thousands of channels caught up in it. Their central finding about outcomes is blunt: videos gained views in the short term, and the majority went on to lose a substantial portion of them, sometimes within a week.

What the policy says, in YouTube's words

Mechanism aside, this is also a rules question, and the rules are not ambiguous. YouTube's fake engagement policy prohibits anything that artificially inflates views, likes, comments or subscribers. The prohibited list includes using third-party services to increase engagement, knowingly using a service to inflate live traffic — viewbotting — content whose only purpose is to trade metrics, such as sub4sub, and even linking to an artificial traffic provider in a promotional or supportive context. That last clause catches the creator who did not buy anything and simply made a video recommending the service.

The stated consequences escalate. Traffic found to be artificial is not counted. Violations can produce strikes, and three strikes inside 90 days terminates the channel. Content and channels that do not follow the policy may be removed outright.

It is worth being equally precise about what is not prohibited, because fear of this policy causes creators to stop doing normal things. Asking viewers to like, comment, subscribe or share is explicitly fine. What the spam, deceptive practices and scams policies target is coordination and compulsion: groups agreeing to like each other's uploads en masse, high-volume repetitive comments driving traffic elsewhere, rewards offered in exchange for engagement, community features abused to force interaction. The line is between asking an audience and manufacturing one.

Invalid traffic and the bill that arrives afterwards

The financial half of this runs on a separate system with its own vocabulary. Invalid traffic, in Google's definition, is interaction with ads that does not come from genuine users or users with genuine interest. It is measured against ad impressions, not video views, which is why YouTube Analytics does not show it and why creators are so often surprised by it.

YouTube's guidance on invalid traffic describes a settlement process rather than a punishment. Earnings are normally adjusted before disbursement to strip out revenue from invalid traffic. If the traffic is identified after payment has been calculated or made, the amount is offset against your current or future balance and appears as a separate line item in AdSense. Where YouTube has broader concerns about the accuracy of a channel's metrics, it may slow, pause or alter the counting itself.

Two details in that documentation matter more than the rest. The first is that invalid traffic is not necessarily your doing — a video can attract it without the creator generating or driving it, and the adjustment still happens. The second is the instruction that creators consistently get wrong under pressure: deleting your channel, deleting or delisting content, or switching off monetisation does not immediately remove restrictions, and can reduce your overall revenue. Panic-deleting the affected videos is the one response guaranteed to make the position worse.

The watch-hours version of the trap

Most view-buying is really watch-hour buying: the goal is the Partner Programme threshold rather than vanity. This is where the purchase fails most completely, because eligibility has always been specified in terms of valid public watch hours, and validity is the whole point of the filter. Invalid traffic does not count toward it.

The instructive comparison is legitimate paid promotion. Views from a Google Ads campaign are real people, disclosed and paid for at market rates — and the watch time they generate still does not count toward the 4,000-hour eligibility threshold. If YouTube excludes watch time it sold you itself, the idea that it will accept watch time from an anonymous panel is not a serious proposition. The current and forthcoming thresholds are laid out in our guide to Partner Programme requirements.

The damage that outlasts the views

Suppose the worst case never happens: no strike, no termination, the traffic quietly evaporates over a fortnight and the AdSense adjustment is small. Something has still been destroyed, and it is the thing that is hardest to rebuild.

A video's analytics are the only instrument you have for deciding what to make next. Average view duration, the retention curve, the traffic source split, the audience geography, the returning viewer rate — every one of those is computed over whoever showed up. Mix synthetic sessions into that population and the numbers describe a fictional audience. You cannot tell whether the retention cliff at ninety seconds is a scripting problem or an artefact of the purchase. You cannot compare the video to your last ten uploads, because it is no longer the same kind of measurement. And any test you run on it is contaminated at the source.

That last point is the expensive one. Packaging decisions are made by comparison — this thumbnail against that one, this video against your own recent baseline. A contaminated video is excluded from the comparison set for its entire life, which means the money bought a number and cost you a data point. Our post on diagnosing a view drop is built almost entirely around reading those comparisons; it does not work on a video whose traffic you cannot account for.

There is a commercial version of the same problem. Sponsors increasingly audit audience quality before signing, looking at geography that does not match the content, engagement that does not scale with views, and growth curves with steps in them. A channel with purchased history has a shape that shows up in exactly the review that decides a four-figure deal, and the creator explaining that the odd spike was a one-off experiment is not in a strong negotiating position. What brands look for, and what they pay, is covered in our piece on how sponsorship deals actually get priced.

Sub4sub, pods and the versions that feel harmless

The softer end of this market involves no money at all. Subscriber exchanges, comment pods, Discord servers where twenty creators agree to watch each other's uploads on publication day — the participants are real people, which makes the whole thing feel categorically different from buying bots.

Policy-wise it is not. Sub4sub is named in the fake engagement policy; coordinated mass engagement is named in the spam policies. But the more interesting objection is technical. A bot that watches nothing tells the recommendation system almost nothing. A real person who watches your video out of obligation tells it something specific and wrong: that people with their tastes and their watch history are interested in your content. The system takes that at face value and looks for more of them. Exchanged engagement does not just fail to help — it teaches the model to show your work to an audience that was never yours, and that error persists after the pod disbands.

It also front-loads the audience least likely to stay. A subscriber who joined as part of a trade will not open the notification, which drags the ratios that genuinely do inform distribution. The legitimate alternative is not mysterious: the mechanics of building an audience without leverage are set out in growing a channel from nothing.

Reading the seller's promises

The services are careful with language, and most of the reassuring phrases describe delivery mechanics rather than outcomes. It is worth knowing what each one is actually claiming.

The promise	What it describes
"High retention views"	Sessions instructed to stay longer. It is a delivery parameter, not a guarantee that validation accepts them.
"Drip feed" or "gradual delivery"	Spreading the delivery over days to look organic. An admission that the raw version is detectable.
"Real accounts, not bots"	Often means incentivised humans on an exchange — which the fake engagement policy covers just as squarely.
"No-drop guarantee" or "refill"	The seller expects the views to be removed, and has priced re-delivery into the package rather than preventing the removal.
"100% safe, no strikes"	A promise about someone else's enforcement decision, made by a party with no visibility into it and no liability for it.

The refill guarantee is the tell. A vendor confident that the views survive would not need a policy for replacing the ones that do not.

If you already bought views

Plenty of people arrive at this question after the fact, often having tried a cheap package early on. The recovery order is short, and most of it is restraint.

1. Stop the delivery. Cancel any drip feed or subscription before anything else; continuing delivery while under review is the one action that reliably compounds the problem.
2. Do not delete the videos or the channel. YouTube's invalid traffic guidance says directly that deleting content, delisting it or disabling monetisation will not lift restrictions and can cost you revenue.
3. Expect the count to fall, possibly weeks later, in one batch. That is the validation system working, not a new penalty.
4. Treat the affected videos as unreadable. Exclude them from your baselines rather than trying to reason about their retention curves.
5. Rebuild the baseline from the next ten clean uploads, comparing like with like — same format, similar length, same traffic source.
6. If the Partner Programme was the goal, recalculate against engaged watch hours in Advanced mode, which is where eligibility is actually assessed.

One thing not to do: mass-deleting old videos to erase the episode. Nothing in the documentation suggests it helps, and the archive is usually the most productive asset on a small channel.

What the money buys instead

The uncomfortable part of the argument is that the impulse behind buying views is correct. Early distribution is a real constraint, and telling a creator with 200 subscribers to simply make better videos is not advice. So here is the honest ranking of what you can legitimately purchase.

Advertising is the one real answer for paid reach. A Google Ads video campaign puts your video in front of a targetable audience, it is disclosed, and the resulting views are counted as public views. The caveats are equally real: the watch time does not count toward eligibility, the traffic is not a substitute for organic pull, and it is only worth running when the video converts a viewer into a subscriber or a customer reliably enough to justify the cost per view.

Everything else worth spending on lives upstream of the click. Impressions are allocated by the system, but the fraction of impressions that turn into views is yours, and it is the only multiplier in the funnel you fully control. A video that converts at 6% rather than 3% earns twice the views from identical distribution — and unlike bought traffic, those views arrive attached to real people whose watch behaviour then earns the next round of impressions. The compounding runs in the right direction.

That is a packaging problem, and it is solvable in an afternoon. Make three candidate thumbnails instead of one, run them through Test & Compare and let the platform pick, rather than betting the video on your own taste. The reason most creators do not is production cost: the second and third candidate take as long as the first. That specific bottleneck is what Thumblore removes — you describe the video and get finished thumbnails back in a couple of minutes, which makes a three-way test an ordinary part of publishing instead of a lost evening. The free thumbnail preview tool will show you the candidates at the sizes viewers actually see them, which is smaller than you think.

So: does it work?

It works at exactly one thing, which is making a number on the watch page temporarily larger. It does not produce impressions, because impressions are only counted on YouTube's surfaces. It does not produce engaged views, which is what earnings and eligibility are computed from. It does not produce valid public watch hours, since YouTube excludes even the watch time from advertising it sells you. And the research suggests the inflated number itself is impermanent: corrections reach the large majority of videos, arriving in batches, usually late.

Against that, the costs are concrete. A policy whose stated remedies include strikes and termination. An AdSense balance that can be debited after payment. Analytics that no longer describe a real audience, on the videos you would most want to learn from. The social-proof theory underneath the purchase — that a big number attracts real viewers — also assumes people are browsing view counts, when what they are actually doing is scanning a wall of thumbnails and choosing one.

Which is the useful way to end this. The constraint that makes buying views feel necessary is real — new channels genuinely are starved of distribution — but the lever that resolves it is the click rate, not the view count. That is the number you can change this afternoon, on videos you have already published, without paying anyone. If you want the mechanics rather than the exhortation, start with what a good CTR actually is and how to diagnose a view drop; when the answer turns out to be the artwork, Thumblore is built to make the second and third version cost almost nothing.