Videos with known AD promoted views

We are receiving a lot of messages about a supposedly "Youtube giving back views" as cause for some view spikes. No, Youtube does not do that, if they did, you would see likes also being rewarded back, or likes compatible with the views. It is a simple math: a lot of views, too little likes means ads, period, get over it.
Hide disclaimer crap

Please read this at least once, it will save you some time understanding how it works, and if its accurate

Is this data trustworthy?
In short, the estimated data DEPENDS. the data derived from Youtube Charts Yes.

The estimated data uses View per Like and View behaviour (bumps on views without likes following) to estimate ads. Since viewership follows a predictable path, it is also somewhat easy to notice the presence of ad views, but calculating how much depends in a lot of factors.
To prevent showing inflated ad view estimates, this site is conservative and will always round up numbers to less views, and will not detect small campaigns. Because of this, the data is not fully trustworthy, but the error margin is for LESS calculated ad views, and usually only 20% variation.
Youtube Chart numbers are trustworthy because they come straight from Youtube: Their charts do not count ads, so a basic "public view counter" minus "views on the chart" will reveal the ad views. You can check Youtube Charts here.
For more on how the data is calculated, how to see ad patterns and the math behind it, check this article.
Videos that get high views (above around 5 million per week) will usually show on Youtube charts and be very accurate, the less acurate data is for videos with less than 5 million views per week, because they won't chart on the top 100 Youtube charts.

I saw an advertisement from a MV and it is not listed here
Like mentioned above, one can only detect big campaigns that usually purchase millions of views. Most smaller companies, and sometimes even big companies, buy small number of ads or spread the ads in several weeks, making it discerning them from normal views through either behaviour statistics or monitoring Youtube charts pretty pointless. This list only shows the big numbers.
If you care that much that the video be listed here, you can mail us, but the best we will be able to do is probably just list it without any number.

What about small campaigns, bot views or third party views?
Smaller campaigns, or a small number of views (relative to the millions that videos usually have) are pretty hard or impossible to discern and are bound to go unnoticed. Like mentioned above, this list shows the big and obvious ones. We can know only what we can find, otherwise we would be inventing numbers.

Why bother with listing adviews if only big numbers can even be detected?
So people understand that VIEWS MEANS NOTHING. It might be impossible to have accurate ad-views data, but we can at least have enough to show the numbers are highly manipulated, and this is just the tip of the iceberg, since there are a lot of ways to gain views that are not actual people viewing and cannot be detected with precision. Trust your instincts, like what you want, and don't let numbers rule your music style.

Someone said Youtube give back views / retain views, that's why there is a view spike!
Someone said Bogfoot existed, it didn't make it exist. No, Youtube do not "give back" or "retain" views other than, rarely, on the first day of major videos when in fact a heavy cache causes the illusion the view counter stopped (and a more stringent audit of views to prevent bots takes place, speeding down view counting). Studios purchase ads and use them whenever they see fit, it is not new and everybody does it.

Remember: Youtube have no system in place to increase views (give back, validade old views, etc), only systems to monitor bots, repeats and invalid views, which can cause a sudden slow down or drop on views. Spikes always mean bots or ads.
Which limitations this data have (in regards to the estimated data, not Youtube Chart data)?
  • All math behind this analysis is pretty conservative. In most cases, actual ad views are higher. Accuracy rate is around 80%;
  • Videos with likes disabled are even harder to predict, and might be ignored altogether;
  • Monitoring started only on early 2019, a few months after Youtube Trueview gained traction with Music Videos;
  • Fanbase and boting do not influence this math because purchased ads far surpasses the number of views created by bots. You can see that, for instance, in videos of groups which don't purchase ads but have high view numbers like Red Velvet or BTS;
  • It can take up to 48 hours after a video is released for the system flag and calculate the ads;
  • Videos with small number of views are harder to evaluate because they don't chart on Youtube, expect even less accuracy (thse are videos with less than 5 million views per week);
  • Music Videos are created to be advertisements, be GLAD your favourite artist have funds to invest in ad-views, it means they are doing great.

Data is reliable up to 80%. If you take notice only of videos that charted on Youtube, it is nearly 95%.
You can disable this disclaimer permanently by disabling disclaimers on your user settings (logged users)
From the 9359 videos we track, there were 2.143.300.000 guaranteed bot/ad views (since 2019), from a total of 123.995.377.518 views - That's 1.73% of "fake" views
Views confirmed and guaranteed to be ads by checking Youtube Charts: 963.000.000
Check last 30 days top videos

Release Artist Music Video Views Estimated ads YC ads* % ads Graph
2021-02-16 Kang Daniel (강다니엘) Paranoia 23.9 m ~17.0 m ~2.0 m ~71.2%
 ↳ On YC of Feb 18th excess of 2M views. (ongoing)
2021-02-15 Chung ha (청하) Bicycle 21.3 m ~12.0 m ~56.3%
 ↳ (ongoing)
2021-02-07 Kim Wooseok (김우석) Sugar 10.4 m ~7.0 m ~67.1%
2021-02-05 Aespa (에스파) Forever 23.8 m ~3.0 m ~2.0 m ~12.6%
 ↳ Charted on Feb 11th with 9.7M, counter was 12M
2021-02-03 Purple Kiss (퍼플키스) Can We Talk Again 3.9 m ~2.0 m ~51.6%
2021-02-02 CIX (씨아이엑스) Cinema 10.9 m ~5.0 m ~2.0 m ~45.8%
 ↳ Never charted. Excess 2M on Feb 11th
2021-01-26 Dreamcatcher (드림캐쳐) Odd Eye 27.5 m ~15.0 m ~1.5 m ~54.6%
 ↳ Didn't chart on YC at Jan 28th, 1.5M excess views
2021-01-25 Golden Child (골든차일드) Burn It 5.0 m ~2.5 m ~50.5%
2021-01-19 Oneus (원어스) No Diggity 17.0 m ~10.0 m ~1.0 m ~58.8%
 ↳ Had 7M on Jan 15th, but did't chart at 1M excess
2021-01-19 CRAVITY (크래비티) My Turn 15.3 m ~7.0 m ~1.0 m ~45.8%
 ↳ Didn't chart on Jan 28th, 1M excess views
2021-01-18 AB6IX (에이비식스) Stay Young 12.8 m ~10.0 m ~78.0%
2021-01-11 (G)I-dle ((여자)아이들) Hwaa 61.0 m ~32.0 m ~25.0 m ~52.5%
 ↳ Charted with 10.8M on Jan 14th, and 7.9M on Jan 25th. By 25th had 43.5M on public counter: 25M ads. 7M adds in one day in Feb 12th
2021-01-11 Victon (빅톤) What I Said 15.4 m ~5.0 m ~32.4%
2021-01-11 Treasure (트레저) My Treasure 23.5 m ~4.0 m ~4.0 m ~17.0%
 ↳ Never charted, 4M excess views in the first 2 weeks
2021-01-11 T1419 Asurabalbalta 28.6 m ~22.0 m ~3.0 m ~77.0%
 ↳ Never charted, excess of 3M by Jan 21st (ongoing)
2021-01-07 MCND (엠시엔디) Crush 14.6 m ~0.5 m ~3.4%
2021-01-06 Jeong Se-woon (정세운) In the Dark 2.3 m ~1.5 m ~66.4%
2020-12-10 Ghost9 (고스트나인) Wall 10.5 m ~10.0 m ~1.0 m ~95.5%
 ↳ Didn't chart on dec 18th YC, excess 1M views
2020-12-09 Jamie (제이미) 5 Christmas Languages 4.2 m ~2.5 m ~59.6%
 ↳ Bump in views a week after release
2020-12-07 The Boyz (더보이즈) Christmassy! 4.0 m ~1.0 m ~25.3%
 ↳ Bump in views a week after release (no likes to check)
2020-12-07 IZ*ONE (아이즈원) Panorama 41.4 m ~15.0 m ~10.0 m ~36.2%
 ↳ Never charted. Excess 10M views on first two weeks
2020-12-04 NCT (엔시티) Resonance (NCT 2020) 70.1 m ~26.0 m ~26.0 m ~37.1%
 ↳ On Dec 10th had 11.7M, public counter at 33M. On dec 18th didn't chart, 5M excess views
2020-11-30 GOT7 (갓세븐) Last Piece 33.6 m ~9.0 m ~9.0 m ~26.8%
 ↳ On dec 3rd had 8.6M on YC, public counter was 18M
2020-11-30 Kai (카이) Mmmh 57.9 m ~3.0 m ~4.0 m ~5.2%
 ↳ On Dec 3rd had 12.6M, public counter at 16M. Had 8M between 4th and 10th of dec but didn't chart, excess 1M
2020-11-27 NCT U (엔시티) Work It 41.2 m ~15.0 m ~15.0 m ~36.4%
 ↳ On dec 3rd had 9.6M, public counter 26M

* - Youtube Charts (YC) Ads are manually calculated by subtracting total views with the sum of views in Youtube Charts, since those don't include ads, whenever possible. If the total views are higher than the last place of a Youtube Chart week, and that video didn't Chart, the excess views are also ads.

Estimated ads are calculated automatically by the system. As described below:

Toggles promotion on: While less than 60 days old, More than 1 day old, more than 2 million views, more than 400.000 view/day, Average VPL < current VPL, Average VPL > Expected VPL and Current VPL > 60 (this last its a sure mark of ad promotion during debut).

Toggles end of promotion: Gets under 400.000 views/day, current VPL decreases from 60 or current VPL is smaller than 1/2 last VPL (sudden drop)

Calculated promotion: Uses VPL table for first 30 days, from that calculates how many views are ADs from (1 - (average VPL / expected VPL))%

Maximum expected VPL: Starts at 10 on the first day, then climbs to around 30 on the first month.
Videos might have a normal View/Like rate after a few months of the AD campaign, that is normal. We detect the AD campaigns as they happen (view/like increasing and past normal), so just like they can get back to normal after the campaign ends, we cannot say which videos older than when we started monitoring (~ September 2019) were guaranteed to have used True View.

Excessions to the rule

Some videos have very high view/likes and are not accountable to ad promotion, they are true viral videos. There are a few that are obvious, here are they:

  • As one - For the night - 85m views, 1615 view/like - Got viral because of the mature content, notice the extreme high "dislikes" (38%, the most disliked K-pop video ever)
  • Kisum - Make your own style - 347t views, 1543 view/like - Usually ads run for millions, but for some reason this ad only run for some hundred thousands. At the time, A Monster X Add was running witht he same campaing, they probably diverted resources to that
  • Sung Si-kyung - We make a good pair - Old video, its normal to have high VPL
  • SG Wannabe - Christmas Story - A weird one. Low view count, low likes, but old. probably all compound to a one-of-a-kind anomaly that doesn't seem to be ad related
  • Kim Bum-soo - Passing - Old video, its normal to have high VPL
Videos with low view count or too old enter statistical noise that are hard to understand and are not commented here.