YouTube's Potential Revenue

I heard last week from a pretty good source that YouTube is serving 100 million videos per day. Say what you will about YouTube's content (unlicensed, kids falling of skateboards, etc), that's a huge number. And it got me thinking about how much revenue could be extracted from such an audience.

Let's say that advertisers will pay on average a $15cpm for a ten second pre-roll ad in front of licensed content and high quality user generated content (lisa nova, etc). And let's say that 60% of the videos being served on YouTube are unlicensed content that could be licensed with the right business deal. And let's say that another 20% of the videos being served on YouTube are user generated content that is high quality. That leaves 20% of the videos being served that are not monetizable. I realize these are unsubstantiated assumptions, but my point is not to be accurate, it's to make a point.

If you make those assumptions, then YouTube could be generating $440 million in annual revenue at their current volume of videos served. $213 million of that revenue would be passed on to content owners assuming a 65/35 rev share in favor of the content owners. And another $70 million would go to the creators of high quality user generated content. That would leave YouTube with net revenue of $150 million at it's current run rate.

Here's the analysis. This is, unfortunately, a screen shot of a google spreadsheet. What I'd really like to do here is embed the google spreasheet into this page so you could all play around with my assumptions because I am sure they are wrong in some meaningful ways.

I also want to dig into the theoretical revenues to a creator of high quality user generated content. Take Lisa Nova, who to date has produced 16 videos, and generated a total of 2 million views on YouTube. Using the same assumptions as I used in the above analysis, Lisa would have made about $20,000 to date from her 16 videos, not a bad take for 3 months work. Certainly not the kind of economics that the major studios want to see from web video, but not bad for a do it yourselfer like Lisa. Here's the spreadsheet I ran for Lisa Nova.

I realize that there are many people (including possibly YouTube management) who think introducing a 10 second pre-roll will negatively impact the viewing activity. It could certainly cause the audience to move elswhere in search of ad free video content. It could also reduce the amount of views.

But I don't think so for a couple of reasons. First, 10 seconds is so short. It's not even worth doing the work of fast forwarding to bypass a 10 second video. Second, if users are allowed to tag ads they don't like and favorite ads they do like (kind of like the way personalized internet radio works), then the ads will become more relevant and will start to be seen as content in itself.

But that's really a digression from my main point. 100 million videos served per day is a huge audience that ought to be worth a lot of money to YouTube and the content creators whose content is being viewed on YouTube. Instead of filing DMCA requests to take down the video, content owners should be cutting deals to run the same content with a 10 second pre-roll and sit back and collect checks.

I think its going to happen. Probably soon.

September 7, 2006 Venture Capital and Technology | Comments (42)

Comments

"Say what you will about YouTube's content (unlicensed, kids falling of skateboards, etc), that's a huge number."

A huge number of shit is still, well, shit.

It's important to remember that those 100 million videos per day don't represent 100 million unique viewers today. My guess is there's something like fewer than 5 million repeat, unique viewers of YouTube shit every day. The other 5.995 billion people on Earth have better things to do with their time. Like pick their noses.

But I digress. We could also say "NBC reaches millions of viewers a day through its sitcoms." But they're still shit.

Besides, Google has more money than YouTube; why do you think YouTube will succeed?

Posted by: Dave | Sep 7, 2006 7:14:46 AM

if the investment bankers on wall st. weren't already salivating about a youtube IPO, they are now. it's an interesting financial model.

Posted by: mark evans | Sep 7, 2006 7:47:04 AM

I wonder how the user gen video makers will survive revenue.

20K isn't bad if Lisa Nova doesn't have to pay anyone. But will camera people, actors and other production people volunteer if they know Lisa is getting paid? 20K for producing 16 short videos doesn't go far if you have to pay your crew.

Posted by: Erik | Sep 7, 2006 8:32:19 AM

For a web embeddable spreadsheet, I'd recommend WikiCalc.

As far a YouTube monetization, I think the reason that it's so popular is because it's free and because it requires no client installation.

I watch YouTube videos, and other videos (like ThatVideoSite.com) pretty much every day. Most of the videos I watch were recommended to me by someone, but I will occasionally browse. I was also a big fan of iFilm back in the day (circa 2001-2002) precisely because it was free and the content is good. I like the interesting shorts, like The 1K Project, or Ryan vs Dorkman, or Diet Coke and Mentos Fountains instead of the watching how many people can hurt themselves doing stupid things.

I can stand short advertisements before the movie, but I rarely actually watch them. If the advertising starts to interfere with the movie watching (ie, takes more than a minute, requires subscribing, or AACK costs money), I would certainly start going somewhere else.

I think the movies are a good example of how not to advertise. Because of all the junk played before the movie starts, I now arrive late to the showings, so I don't have to be told how wonderful the world would be if we all drank Coke or ate peanut caramel nougat chocolate energy bars.

Maybe advertising isn't the right way to monetize YouTube. If we have reliable micro-payments, how much would I pay to see one of the videos above? 5 cents? That may be too much, but I would certainly pay 2 cents for it. But I don't want to have to subscribe to the video site and manage another account. If I could use some sort of single account that allowed me to authorize, in a single password, a 2 cent fee, then watch the video, and then to be able to watch the video again for no additional fee, that would be the best.

Posted by: Steve Betts | Sep 7, 2006 8:52:14 AM

Talking to ad agencies, the problem with YouTube is that brands do not want to take the chances of being linked to irresponsible, anarchic or politically incorrect content.

This is what makes the video indexing arena interesting. If they can scan videos against a list of words and phrases, they'll be able to more or less eliminate the chances of brands being besmirched.

Posted by: Ian Delaney | Sep 7, 2006 8:54:58 AM

An interesting analysis, Fred. I have a feeling that your CPM rate is likely too high, and the proportion of monetizeable videos is also likely too high -- but definitely food for thought.

Posted by: Mathew Ingram | Sep 7, 2006 9:01:30 AM

Points well made, and I cannot resist (and do this kind of modelling for a living), I need to dig into the assumptions offered, to wit:

- 100mm streams/day is publicly-printed figure, so your "good source" might be the NY Times, or directly from YouTube. :)

- 80% sell-through and $15 CPM's are the two most senstive variables to your top-line assumptions, and both strike me as very, very high, unless 100% of inventory was being sold directly by a mature sales team to brand-focused advertisers (both facotrs are years from reality).

- 100% sell-through (via 3d-party ad networks) *might* yield $5CPM's (after splits), and realistic, directly-sold CPM's are likely closer to $10 in any meaningful volumes (which YouTube certainly has), and might represent 15-20% of total inventory -- even 12 months from now.

- In terms of revenue-sharing with content providers, the most widely-available, analagous "3d-Party Splits" are Google AdSense, whose splits are not disclosed at the site level, but the blended rate GOOG pays to 3d-party sites is ~75%. Revver pays out 50/50 to its video content producers, so your 65/35 assumption is a good one (strategically and competitively) for YouTube, assuming they seek to lure producers away from Revver/etc.

- Finally: The "10-Second Pre-Roll" debate is a false one, and strikes me as unnecessarily intrusive. It also assumes "One Stream, One Ad Impression," which also over-simplifies. More mature online video sites (eg, Heavy.com) already "wrap" their video windows with ad units that display in the form of "Flash Curtains," and maintain persistent states "around" the video streams. Postroller is a burgeoning online video ad network that does (you guessed it) post-roll video. I could seen a world where each stream produces (unobtrusively) the equivalent of two ad impressions.

Forgive the length, but I enjoy this kind of thing, and how models can help to tame complexity (to a certain extent).

Posted by: Chris | Sep 7, 2006 9:03:14 AM

Your analysis seems sound assuming 100 million views. Trouble is, start putting ads in front of eveything good and people will watch far fewer videos. As other alluded to, part of YouTube's appeal is it serves up vids dang fast with no players to download. I can jump around and view lots of stuff in a short period. Anything that impedes that will mean I watch fewer vids. Still money to be made, maybe not as much, though.

Posted by: greg | Sep 7, 2006 9:23:33 AM

It's an interesting analysis, Fred, but frankly, I think the most important assumption is the one that's missing - the proportion of the userbase that will evaporate as soon as commercials are put in front of it.

Posted by: Rob Hyndman | Sep 7, 2006 11:22:14 AM

You really have to ask yourself, how likely are you to surf around YouTube if you have to wait an additional 10 seconds before watching a video. That is like asking people to use dial-up again, giving up broadband speed. This is not progress, and not likely a model they will embrace. However, your numbers do point to the potential of the site.

Posted by: Dan Blank | Sep 7, 2006 12:31:24 PM

YouTube is a brilliant winner and I hope everyone makes out like bandits.

But I think a little context is in order:

Pre-Katie Couric's exit of course, NBC's smash hit most-sturdy franchise "The Today Show" generates about $500MM revenues/year -- about what Fred thinks YouTube can do. The show lasts three hours per day (180 minutes) and reaches about 7 million viewers per broadcast, for a total of 1.26 billion "viewer minutes" per day (180 minutes x 7 million viewers.)

Its a crude analogy, but here goes:

At 100MM "videos served per day", YouTube is generating (in my guess) about 250MM "viewer minutes" per day (assuming each "video" is about 2.5 minutes of viewing).

So for YouTube to generate annual revenues roughly equivelent to The Today Show, a "viewer minute" on YouTube would have to be worth 5X what a viewer minute is worth on "The Today Show". And that I just can't imagine. (Remember: YouTube generates only 1/5 as many "viewer minutes" as does "The Today Show.")

Also consider:

Ad spots on "The Today Show" are 30 seconds long, and full screen, broadcast quality visuals, as often as not now displayed in high quality sound on big high quality flat screen TVs and the like. Ad spots on YouTube are small poor quality, etc.

And most important, at only one ad spot per video, YouTube can at best display 100MM "ad views" per day -- again, they only get one view per ad spot and one ad spot per video served -- whereas "The Today Show" displays 672MM "ad views" per day (16 minutes or 32 ad spots per hour X 3 hours X 7 million viewers.)

So for Fred's analysis to be right, the small low quality "ad view" on YouTube has to be worth 6.72X what an "ad view" on The Today Show is worth.

Not yet anyway.

Posted by: steve | Sep 7, 2006 1:09:35 PM

I write about YouTube. I use it on my sites. I watch it for entertainment.

10 second prerolls would really suck and I've already started looking for alternate sources of video because of that possibility.

10 seconds is a long ass time once the Internet has rewired your nervous system.

I could handle a 2 to 3 second preroll without complaint.

If you want in-video ads, which I think makes total sense for YouTube, check out Revver's end of video ads. I found them totally not annoying though purists will disagree. I think they're a potentially profitable compromise.

I have problems with Revver that Youtube seems to have solved, for example, on my OSX/Firefox combo Revver displays with a black screen while Firefox displays with a still picture.

That's a huge difference.

In any case, as a daily user, I'll be really upset if they add prerolls.

Posted by: Clyde Smith | Sep 7, 2006 1:57:34 PM

It would be interesting to see and compare this to how much Yahoo makes on their News stories delivered via the web. They air ads such as pepto bismal before these stories and I don't find them too invasive.

If a number could be reached, this could be a benchmark to estimate the value of the advertising placement.

Overall, I do think it would hurt viewership on the website; however, probably not enough to say it is not worthwile to try out. Hell if it is a funny video people are going to watch it regardless of the 10 sec advertising in front of it...at least I would.

Posted by: Andy | Sep 7, 2006 2:12:58 PM

Pre or post roll ads: this idea is an old one and goes back to a slew of Web1.0 video sharing sites.

The problem is that there's absolutely no demand right now from serious advertisers for user-generated content. Maybe that will change in the future, but I don't think so.

Posted by: Chris D | Sep 7, 2006 2:54:20 PM

Pre or post roll ads: this idea is an old one and goes back to a slew of Web1.0 video sharing sites.

The problem is that there's absolutely no demand right now from serious advertisers for user-generated content. Maybe that will change in the future, but I don't think so.

Posted by: Chris D | Sep 7, 2006 2:54:20 PM

Thanks for getting the party started, Fred.

I think the model is a good starting place. I agree with many comments that cpm is too high, 100% sell through isn't possible yet, and 80% of saleable inventory is too high, and 1 ad / video is probably not the mix YT will optimize for. Rev splits seem about right, though...

Today Show comparison and Content provider economics were also interesting.

But we should also consider,

-- this company has 35 people (none of whom make Couric-like wages and BW is getting cheaper)

-- it's still early (100MM vid views is a month old number)

-- once monetization becomes possible, higher quality content will appear

-- this has global potential

-- there are network effects; the more content, the more users

-- and secondary network effects (as we see in Google adsense today) the more traffic and the better your monetization, the more you can pay (overpay) for content or distribution (syndication)

-- I think Fred's main point is that YT, despite it's lack of obvious bus. model today, could be a powerhouse is dead-on!

Advertisers will follow eyeballs.

Posted by: Jeff Schrock | Sep 7, 2006 4:56:30 PM

There has to be a better way than a 10 second intro. As for a $15 CMP on that kind of attention, I'd say its nrealistic. People pay similar rates for banner ads and banners are a lesser form of interruption marketing...

Rob is right though...people will leave and fast. So again...there has to be a better way than 10 second spots.

Posted by: Mike McDerment | Sep 7, 2006 5:53:51 PM

What does 100 million count?

Does it count the video I watch the first 10 seconds of before changing away?

I wonder percentage videos that are started play through to the end. I wonder what percentage of videos are terminated in less than 15 seconds.

If I'm playing around in YT looking for something fun (channel surfing as it were), a few seconds on each video to see if it grabs me, then 10 seconds pre-roll ads are a total buzz kill.

Posted by: Erik | Sep 7, 2006 5:56:17 PM

This tidbit came YouTube's peering coordinator, Colin, at NANOG in May 2006... When they were streaming about 40M videos per day. At that time, they were purchasing about 30 Megabits/second/per month of transit bandwidth. At $15 per meg, YouTube's bandwidt costs back then were probably running $450K per month. With 100M videos per day, their bandwidth costs must be exceeding $1Million a month now. Since bandwidth is the highest variable cost for anyone distributing video over the Internet, these costs must be factored into any business model and will likely be one of the key drivers to determine whether or not the business can succeed.

Posted by: Jeff Turner | Sep 7, 2006 6:01:02 PM

Sorry, I posted above and then I realized that I forgot an important point.

If YT is going to sell ad inventory, then they are going to have to guarantee that their content is "clean", with no copyright issues. Otherwise they will loose their - already wobbly - Safe Harbor provisioning under the DMCA since they will be profiting from copyright infringement. That means they will have to:

- Disregard all current content since they have not been screened (except for obvious well know content)
- manually review all submitted content to make sure they are safe

This would create a huge operation burden on YT. They say they get 40k videos/day. How many people would you need to screen all of that content? How would that huge headcount change the operating financials of YT? Plus content would no longer be available "immediately" after submission and a time lag would occur - similar to what happens in Google Video. People have been saying that this immediacy of publishing has been a plus for YT, so they will loose that.

Fred, what's ironic here is that the LisaNova Content Creator you trot out as an example of quality user-generated content actually has some copyrighted background music. I can only presume that she did not get clearance from the publishers to use that music.

I believe YouTube is already aware of this and has been forced to change revenue models away from video ads. I've read that Chad has explicitly said that he doesn't want to do pre/post-roll advertisments, presumably because they came to the same conclusion as myself.

Basically, if YouTube starts to roll in the cash, they will be immediately pounded with lawsuits. You may see opportunity for Content Owners to get publicity/distribution, but that's not for you to decide.

Posted by: Chris D | Sep 7, 2006 9:03:19 PM

I think you've made two miscalculations.

1: A 10 second pre-roll will kill YouTube. Stone dead. The audience simply will not sit through it - there are too many alternatives.

2: The amount of highly viewed YT content that cannot be monetised because it contains unlicensed copyrighted material is much closer to 90%

Posted by: Mr Angry | Sep 7, 2006 11:56:57 PM

Great post and awesome comments.

Not much talk of expenses, though.

I'd bet they are running about $20-25 million in this year for bandwidth/hardware/people. That's a pretty good profit margin, based on Fred's numbers.

Posted by: Rick | Sep 8, 2006 12:19:58 AM

The answer is not youtube, but Revver. Revver already does what you are stating and splits the revenue with the craeter 50/50

Posted by: Bob | Sep 8, 2006 2:10:19 PM

"Let's say that advertisers will pay on average a $15cpm for a ten second pre-roll ad"

I stopped reading right there. 15 CPM... not going to happen, i don't care if it IS video, you're not going to sell 100 million slots a day on the internet for 15 CPM with limited revelancy and revenue ties...

1.5 ... that sounds more reasonable. MUCH more reasonable.

Posted by: Nii A. | Sep 8, 2006 5:48:41 PM

Thanks for putting together this analysis. I agree with the other posters that a 10 second clip might not be realistic since it will dramatically change the experience for YouTube's users.

YouTube might want to look into a watermark, banner ad, or something else that doesn't detract from the experience. A huge part of its appeal (for me anyway) is the quick, instant access to videos. I feel the effects of interface friction pretty quickly and feel users would quickly migrate to Google video or another site. I'm not certain what brand loyalty YouTube has.

I actually made a tool (InstaCalc) to deal with exactly the need to share and play with this type of analysis. I've put in your numbers:

YouTube analysis on InstaCalc

You can also embed a mini version in any webpage, which would look like this:

YouTube Analysis (Mini-mode)

For those who don't agree with Fred's CPM numbers, you can easily plug in new values in the pages above.

Posted by: Kalid | Sep 8, 2006 11:02:27 PM

A VC

Musings of a VC in NYC