If the $23k average for the top 1k is correct, then only the top 250 or so actually make $23k a month or more. At that line are some pretty big channels, like Husky Starcraft and big bands like Kesha and Muse. In which case, this doesn't seem like a lot of money.
That is an awesome plot. That combined with a combined with a tenancy plot to determine what sort of churn there is in the top 500 and its mix would really illustrate what sort of 'opportunity' YouTube is.
Note that syndication is very popular with the top channels, so you have top creators like RWJ owning 3 of the top 1000 channels, VEVO owning more than 50, same with machinima.com, Demand Media and others which I might have missed.
So I'm not sure what law the distribution follows, but the top players get a TON more money than the rest.
The top 211 players make more than the rest 4789.
EDIT:
It also means that the top player universalmusicgroup makes $485,707/month, and the 1000th player playboy makes only $7,880/month (assuming payouts are proportional to the views, which isn't perfectly correct)
The way people normally plot these kinds of distributions is on a log/log graph (to be clear: where both axes are log'd, not where only one of the axes is doubly-log'd).
That's not true, only if the distribution in both directions is exponential, which is not the case here. The horizontal scale is just the rank, which is linear.
Interesting; FWIW, I am going based on the comment "the channels follow a power law" from the grandparent post. I was under the (possibly mistaken) impression that these are normally plotted using log-log; would you be willing to explain what I am getting wrong? (I am totally happy to believe I'm missing something btw, and I certainly did not anticipate the person I responded to deleting their post...).
(edit: Further, looking at the graphs, it certainly seems like they would do well on a log-log chart; also, I would have expected the behavior that was described in the deleted post from a distribution of this sort: specifically, where further introductions of logarithms to the same axis did not affect the result, as they were asymptotically identical.)
The parent is obviously wrong about power law (like e^x). It's not even e^(e^x), it's something that grows more rapidly, like x^x, or x factorial. I'm too lazy to figure it out.
Rank can be considered linear or logarithmic. It can be argued that the difference between being 1st or 10th is orders of magnitude bigger than the difference between being 100th and 110th.
In fact I think a log scale for rank usually makes the most sense, especially when starting from 1st.
The parent is correct, the claimed relationship being y ~ rank^a for some a<0. You would want to see a linear relationship on the log log plot. Another choice would be to examine the complmentary CDF P(X > x) and look for power law tails.
I don't really have time to do this because I'm supposed to be working on a paper about, wait for it, power laws, but I went ahead and re-plotted the data using log-log axes. Sure enough, it's linear as would be expected for power law type behavior:
I'm glad to see more attention paid to this. The idea of the "youtube celebrity" has been around a while, but to my mind that's less interesting than the idea of people making a living off of youtube. And the fascinating thing is just how many people are now gainfully employed making youtube content. This is a phenomenon that has crept up rather rapidly and hasn't received very much attention in traditional media channels or even in business channels. I find it odd that traditional television networks and movie studios have basically been trying their hardest to pretend that youtube doesn't exist, or that it is merely a repository of funny cat videos.
There's a real revolution in video entertainment going on right now. And youtube is still growing at a phenomenal rate. The recent example of the explosive popularity of the Gangnam Style music video is a perfect case in point. It's not just "viral", it's a cultural phenomenon, and it's taking place on youtube whereas in the past you'd expect people to be exposed to popular music videos through television.
If I were in the traditional TV business I would be all over this. I'd be snapping up talent off youtube and I'd be setting up deals and producing content for the internet, etc. But it looks like this is still a case of two separate worlds. I suspect the "oh shit" moment will come when a more traditional format TV show or movie published on youtube becomes wildly successful, and it's really only a matter of time until that happens.
Long tail risks makes avg/var/standard distribution stuff useless. Best measure here would be the median income (50th percentile - or the 50th person out of 100).
Same thing with wealth distribution and taxes - averages are pointless since 1 rich guy + 99 poor guys will have an average of ~1/100 the rich guy (which is still a lot of money) - even though the poor guys have $0 in assets (usually negative equity because of debt).
The median of that distribution would be $0 - hence providing a better representation of actual probabilistic income/taxes.
I'm absolutely with you; I would love to see more data on this.
That said, while we definitely exist in a "hits" driven culture, I expect such hits would be boundary crossing, and I see no reason to presume a middle class can't exist in any given boundary. (Or, maybe, it could grow toward that.) I'm reminded of the recent post on Pandora talking about "middle class musicians".
What is the distribution like among the top 1000 channels?
Presumably the top of the top 1000 channels far outproduce the bottom, just like the top 1000 far outproduce the rest of youtube. Given such a distribution, what are the top few channels earning? 100,000 a month? 1,000,000 a month? Even more?
Other interesting questions: How much of that is profit? Relatedly, does it cost a prohibitive amount to have a highly successful youtube channel? Does amount spent on production each month correlate with amount earned, or do consumers care much more about the content than the production quality?
What is the median income of the top 1000 channels? What's the overall median income of active youtube partners? What's the median income of channels produced (primarily) by a single person?
Some of these questions should be answerable by looking at available stats, like views for the top couple of channels versus the lower end of the top 1000. (Are the actual channels in the top 1000, preferable ranked, available somewhere?) I'll see if I can answer any more of them in the morning...
There are literal kids (teenagers) making 6 figures a month from their Youtube channels with absolute 0 production costs. The best example would be a guy called Pewdiepie[1], he makes gaming videos and does something like 5 million views per day (which is probably around $10k revenue assuming he has a good deal...). He isn't alone, there's quite a few others approaching his level of success.
Something a lot of people seem to forget too is for every video they put out their old videos still get views. It's like investments, the money they make compounds over time because for every new video that's more and more views. So as their viewership grows, so does their previous video views and so do their potential views for the future. Even if someone "only" does 10,000,000 views from new videos in a month they can still be doing substantial revenue from old videos. This is how some of the bigger channels work, they produce incredibly popular videos every couple of months and fill the time with smaller videos.
A good example of that would be Captain Sparklez [2], a guy that makes video game music and commentaries, his music videos can do anywhere from 10 million to 70 million views and he then produces game play commentaries between his bigger hits.
There is also the fact that this is only ad revenue. I would wager that people with good marketing skills make much more from side gigs than from direct ad revenue.
From what I've noticed on many of the top YouTube channels I subscribe to, many have merch to sell and other side projects that earn them a little more money.
Not to mention that several producers have multiple channels, where they use the success of a single channel to make their subsequent channels just as successful.
His list is based on top subscribers over all time, these people are fast rising to the top but are relatively new. Pewdiepie has been making videos for about a year (as someone popular) and is currently #18 on the most subscribed, the channels in the list compiled by rorrr are accounts that are ~5 years old.
Pewdiepie currently averages around 4,000,000 views per day, that's 120,000,000 per month which even at a really poor $1 CPM is still over $100k; at $2 CPM (or more) he'd be making well into 200k+
That's odd. I recall several channel owners claiming million dollar revenues, eg. Jason Njorku from irokotv claimed 1m in annual revenue on this channel for Nigerian movies:
The article talks about monthly revenue. And even if the average was in the middle (ignoring the "power law" mentioned above) then you would have a lot of channels making over $83k which equals over 1m annually.
What do you mean? $23k is the average for the top 1,000. The top 10 (and maybe even 100) (which includes RWJ) are making millions, this article doesn't dispute that.
What strikes me is how little money is being made here. $2 CPM rates will not support the production quality we all expect in our entertainment. The dream of the internet supporting a large creative class self publishing is still far off.
Well, what I do not understand is the 2$ eCPM, that is thrown in in the comments.
If the average 23k$ hold true for the Top 1000 and if the 2$ eCPM holds true, why did the 5.000th channel make an estimated 57,371.70$? (28685850 AdViews * 2$ eCPM / 1000)
I recently got an email from youtube suggesting that I monetize my channel as they'd noticed that one of my videos was becoming quite popular. It's had about 500 (five hundred) views after 4 years! How much would I make?
If you got the average eCPM that these numbers indicate ($2) then you'd make about $1 for 500 views; could be more, could be less (CPM = cost per thousand)
If the $23k average for the top 1k is correct, then only the top 250 or so actually make $23k a month or more. At that line are some pretty big channels, like Husky Starcraft and big bands like Kesha and Muse. In which case, this doesn't seem like a lot of money.