Youtube's Paradox of the Long Tail

There has been a lot of talk recently about YouTube and monetization. Fred Wilson suggested some $150M+ in annual revenue. Others like Jason Calcanis disagree with some of Mr. Wilson's assumptions but he does acknowledge that Youtube could do $20M and TJ suggests that even a minor home page tweak could net $5M in annual revenue.

I don't dispute that Youtube has a lot of value, but I wonder if overtime some of their value will erode. As an aggregator, Youtube has benefited greatly from the long tail of video production. However, that same taste for diversity means that audiences are fractured into small niches. In other mediums like blogs, small targeted audiences are great for advertising. However, in video advertising has always required demographic audience studies to sell advertising. This may be possible around the short head of the most popular videos but I suspect it will be difficult to get a critical mass of survey data about any clip in the long tail in a meaningful time frame to capture the majority of its views. Contextual advertising could also work in video, but it currently requires some type of human categorization beyond simple tags like "funny" or "weird," which makes applying it to the long tail too expensive. Thus, the long tail paradox is that while diversity may help build and audience, when the content can't be easily contextualized long tail audiences will be more difficult to target and therefore more difficult to monetize through advertising.

One of the over-arching trends of the Internet has been the increasing efficiency of advertising. Thus, I'd also question how much longer the value of demographic based advertising will hold. I can't remember exactly where (maybe Battelle's The Search) but I distinctly remember reading that TV advertising, which is demographic based, is far away the least efficient form of advertising. Furthermore, what demographic based advertising is good at is branding. Yet, as barriers to entry fall in new markets and the long tail phenomenon spreads, branding will become less valuable as choices proliferate. Or maybe it will become more valuable precisely because product selection is so diverse, but I don't think so. The bottom line is that my decision to watch a short clip of someone falling off of a skateboard does not say much about my consumption preferences. It certainly says a lot less about me than when I search for "cycling computers," or read a blog about the Tour de France, which both clearly demonstrate my interest in bicycles.

Overtime, advertisers will allocate their budgets more efficiently. This means that less dollars will be spent on audiences that can only targeted by demographics and the value of Youtube's inventory will decrease. In order for Youtube to be a billion dollar business in the medium to long term, I think that they will need find ways to place contextual ads or get more information from their users to target them as individuals. I don't think that cookies tracking, which video clips a user watches will be sufficient to provide enough data about individuals without understanding the video's content at a deeper level than basic tagging. Furthermore, I doubt that Youtube users' tolerance for marketing surveys is very high. In the future, video search or advanced forms of tagging may help Youtube place more targeted ads but right now I don't believe either are sufficient to create meaningful contextual ads. Thus, I find it pretty easy to agree with Jason Calcanis's description of Youtube selling ads at junk CPM rates and I think these rates could even fall over time.