How Does YouTube Make Money?
We looked at Hulu and most of all, at YouTube…The models use public domain information and make some of our own assumptions as to underlying economics. Before you read this, you should probably read this post.
The dynamics of the 2 businesses are different in that YouTube serves an enormous volume (reportedly 1 billion plus downloads a day) of small files (around 2.75 minutes on average) at a low ad-inventory fill ratio - only 3-4% of videos carry advertising at all - and low CPMs (around $10, if that). Hulu serves a far smaller volume (about 3-4 million per day) of larger files (around 27.5 minutes), with dramatically higher ad inventory served (we have assumed 80% in the models) with higher CPMs ( $15 — 20).
Regarding content creation, we assume YouTube, in effect, pays nothing for its content (recent mainstream deals will change that going forward of course), whereas Hulu has to pay 70 to 72% of its income to the content providers. Aggregation is assumed to be hosting costs, upload and transcode costs, and all staff and SG&A costs. As can be seen, they are fairly small compared to video distribution costs as volumes grow.
The costs of distribution we use are based on Amazon S3 and Akamai pricing; we reckon Hulu’s paying roughly the S3 pricing, and YouTube is paying 50% of that due to its use of Google data centres and extensive peering. Interestingly, the widely-quoted figure of $1 million a day in bandwidth appears to be accurate; our estimates using two different methods converged on a bandwidth bill of $30 million a month.
The main business dynamics revolve around average ad value vs cost per video served — the higher the CPM and Ad inventory % served, the better the revenues (favours Hulu), the bigger the file size and payment to content producers, the worse the costs (favours YouTube). The percentage of outgoing streams that carry any advertising is crucial; whether it’s a macro-advertising play like Hulu (comparatively few but high-value) or a micro-advertising play like Google Ads (low value but extremely high volume) is secondary, as both stand or fall on how much of their content actually carries ads.
We estimate that Hulu is marginally lossmaking today, losing around $9m this year on revenue of $52m, and YouTube will lose about $91m on revenues of $118m. We conclude that, for YouTube to break even, it needs to either drive down its transport costs to 33% of major CDN equivalent or radically increase the percentage of streams that carry advertising. Major change factors which could perturb this include the impact of high-quality (HQ FLV or H.264) video on YouTube’s bandwidth budget, the relationship of both competitors to their parent companies, and their relations with content providers.
Our bandwidth estimates are based on Adobe’s documentation of the Flash video protocol and the actual size and other details of YouTube videos, which give a bitrate of 384Kbps; providing high-quality video could increase this by as much as a factor of five (assuming H.264 High Predictive, 320×240, frame rate 36). Obviously this would pose a significant challenge to YouTube’s profitability.
YouTube is increasingly making deals with content providers to share ad revenue with them in return for the use of their copyrights. This is of course a hit to revenue, but it can be confidently assumed that YouTube will make sure that if anything carries ads, it will be the videos they are paying for. Depending on the distribution of traffic among the videos, it may be that the 3-4% ad ratio is enough to fund the long tail of free material, if the hits all carry advertising; a sort of balanced portfolio approach. (They may also choose to advertise preferentially on high-quality videos.)
Hulu, being owned by content providers, has the feature that its profitability is a function of their tax policy - the owners could choose to set the price of their content high enough to secure the whole of Hulu’s potential profits as sales, or alternatively they could keep it low or even free and take any surplus after distribution costs as a dividend.
However, the fundamental issue is the same even if they choose to obfuscate the business model; funding online video from advertising is all about how many ads per stream you serve up. If YouTube could sign some more peering agreements and keep moving things to cheap sources of power, increasing the ad ratio from 3 to 4% would be enough to reach breakeven. Essentially, it’s a traditional media sales desk - first of all, you’ve got to fill ad space. You can choose whether you want a margin-first strategy, like traditional advertising, or a volume strategy, like Google Ads; but either way you must sell.