Yuvutu Blog

In-Video Advertising

(C) yuvutu.com 2008

Adult webmasters who run free tube sites face exorbitant hosting costs, and need to find revenues streams to cover this expense. To remain a free tube site, the obvious solution is advertising, and there has been much hype about the high rewards of in-video advertising. However, adult webmasters be warned, the CPMs in the adult industry for in-video advertising are far lower than mainstream internet closer to 10 cents rather than 10 dollars.

In video advertising

In video advertising comes in 3 main groups:

• Pre-roll
• Mid-roll
• Post-roll

Each has its advantages and disadvantages. The main disadvantage of post-roll is that only a proportion of videos are viewed until the end by the viewer and hence quite often the ad is served not served. Its main advantage is that it poses no irritation to the viewer who gets an uninterrupted viewing experience.

Pre-rolls on the other hand are served to 100% of the video views, but will inevitably annoy the viewer who may decide to switch to another free tube site that does not serve pre-rolls ads. The executives at advertising networks may counter-argue that the pre-roll is only shown in the few seconds that the video is buffering. But as broadband improves, this is a shrinking window and requires the viewer to have lightning reactions in order to read the ad, register an interest and click on the displayed ad. It is therefore of little value to the advertiser as well the viewer.

Mid-rolls are similar to both pre-rolls and post rolls, in that not all ads will get served (more than post rolls but less than pre-rolls) and will cause annoyance to the viewer by blocking out part of the screen.

Ideally to avoid losing your visitors surfing away to another porn site, post-rolls should be served. But what are the economics of post rolls? And can the revenue support the cost of serving the video?

Post roll efficacy

To answer these questions, yuvutu analysed a sample of 65000 video views of 2600 different videos and measured the number of videos that were viewed right to the end of the video – we will call this the video completion rate. We then plotted the completion rate against the length of the video and obtained the following graph:

As can be seen from the graph about 50% of 30 second and only 35% of one minute videos are viewed till the end. The rate continues to drop fast as the length of videos increases: only 25% view 2 minutes videos will the end and 8% for a 10 minute video.

The post-roll CPM for a long video is the same for a short video, so from a revenue point of view, it makes sense to host shorter videos than long ones. Except that it is reasonable to assume that people prefer long one to short ones, and if your tubesite contains only short videos, will your visitors seek out a competitor site with longer videos?

Ideal video length

Using our historical video data, we looked at the number of views per video against their length and concluded that there is a strong correlation. As can be seen in the chart below, people seem indifferent whether the video is 15 seconds or 60 seconds long, however for videos greater than 60 seconds the longer the video, the more people will watch it. A 10 minute video will get more than double (2.25 x) the views that a 2 minute video is likely to get. For videos videos greater than 10 minutes, the gain in number of views diminishes rapidly.

Despite a clear user preference for longer videos, there is still a strong appetite for videos less than 2 minutes long.

Economics of video hosting

Given the above information, what is the ideal length from the webmaster’s point of view? Obviously the longer the video, the greater the hosting cost. Below are the calculations that determine the break-even point for length of video:

Assumptions

• Video file size = 35 KB per second of video
• In-video revenue = 12 cents CPM per video play
• Hosting cost = $15 per Mbps per month
• Let’s assume S is length of the video in seconds

$15 per Mbps per month = $15 per 60 x 60 x 24 x 30.5 Mb = $15 per 2 635 200 Mb
= $15 per 2 635 200 / 8  MB = $15 per 329400 MB (=4.6 cents per GB)

Hosting cost = File size in MB x  $15 per 329400 MB

= 35/1024 MB x S x $15 per 329400MB
=  0.035 x 8 Mb x S x $15/329400MB
= $0.00000156 x S

Breakeven point CPM = 12 cents per thousand views = $0.00012 per view

$0.00012 = $0.00000156 x S
S= 77 seconds

Hence if you share the same assumptions above and you try to breakeven on advertising revenue alone, you better make sure your average video length is less than 77 seconds. If you try to break-even on post-roll advertising alone, you need to factor in the fact that the post rolls will only be served for a percentage of views. For 77 seconds that percentage is 28%. Of course if the viewer has not watched till the end then there is probably a bandwidth saving – BUT a user with a 2 Mbps broadband connection will download the a 77 second video within about 10 seconds – so if it takes him or her 10 seconds or more to decide to navigate away then there is no hosting cost saving. It is this kind of complication that makes the calculation of the exact post-roll breakeven point a very complex calculation. Without going into more detail on the calculation, yuvutu breaks-even on 40 second videos assuming a post roll revenue CPM of 12 cents alone

40 seconds is probably too short for a webmaster who wants to attract sustainable traffic. Hence other advertising on the page will be necessary to cover the full hosting cost.