On Sunday, YouTube broadcast the live jump by Felix Baumgartner for the Red Bull Stratos event. Soon afterwards, multiple media outlets were quick to call it the largest webcast ever and said it broke records, simply because YouTube’s player has a counter on it showing how many users were watching the stream. What they didn’t mention was that at peak, all videos on YouTube was down for many users, the live stream was encoded at a low bitrate and YouTube didn’t even broadcast the video on their network, but relied on third-party CDN Akamai instead. Seems nothing has changed from years ago when I wrote a similar posts asking, “Does Anyone Care About The Business Of Live Events, Or Just The Traffic?”
[Updated 9:36pm: YouTube has just posted to their blog that "more than 8 million concurrent livestreams" of the event took place, but their numbers don't match the ones that we have so far seen from Akamai. And they didn't say "unique" streams. If they are going to put out vague numbers, they should at least put out the methodology on how they count streams as many times I could load the player, but not the video. So are they counting player loads or streams starting? Also, they say it was the "most concurrent views ever on YouTube", but didn't say "delivered by" YouTube. And unlike the media, no where did they imply it was the largest video event ever on the web, so I will give them credit for that.]
While the YouTube player showed a counter that went up and down during the event, it’s not a true indication of exactly how many simultaneous users were watching the stream. As an example, I watched the stream on three devices, all at the same time, but I am only one user. Is Google counting me as one unique user, or three unique users? Am I counted as one simultaneous stream, or three simultaneous streams? None of the numbers you see in the player are ever exact and many times, they are way off. Content owners who do live events across multiple CDNs will tell you that many times, the number of users they see via the tracking technology in their player and the number tracked via the content distribution network, rarely match up.
For the event, all streams I traced were coming back to Akamai which was clearly involved in the event and when I looked at Akamai’s reporting dashboard during the event, their “live stream” count never went above 3M. And that’s 3M for all of Akamai’s customers on their network, not just YouTube’s. It is possible that both YouTube and Akamai were both doing distribution of the video for the event, but people I spoke to who traced the stream in Europe and Asia all saw it being delivered by Akamai as well. One thing that was clear from this webcast is that Google still can’t seem to handle large-scale live events on their network. For the recent Presidential Debate and PSY concert, both of them crashed on YouTube and doing trace routes during those events, didn’t show any CDN provider like Akamai being involved.
For this event, the live stream that Akamai hosted worked without any hiccups that I saw or that were reported, but the traffic in general to YouTube’s website caused a lot of 502 errors with the site being down for me and many others. While the media is so quick to want to call something the biggest ever, Google has never provided their methodology for how they count “simultaneous” streams and I’m sure they never will. As we have seen from past YouTube events, like the Olympics, they have never given out an exact number, but rather said “more than”, so I don’t expect them to break out the Red Bull webcast with any really detailed numbers, with methodology.
Far too many people in the media and in the industry want to judge the success of a webcast solely by the number of people who watched the event, rather than using that number as just one of many data points that need to be viewed together to judge the success of an event. But the biggest thing the media is missing, that no one has discussed yet, is who made money from this event? Why is the media so quick to want to talk stream numbers after an event, but then never looks at the bigger picture of what live events mean for content owners and publishers who are trying to monetize content? I didn’t get any ads during the event, so it does not look like YouTube monetized the stream and the way YouTube is throwing money at content owners, I’d be willing to bet that Red Bull didn’t even have to pay YouTube for the live stream, since YouTube was a sponsor of the event. The media should be focusing on what the business model was for this webcast and not simply how many streams it delivered.
So what is the largest record for a live event on the web, in terms of simultaneous numbers? The answer is, no one truly knows. None of the data provided after live events is usually very detailed, validated by a third-party and many large-scale webcasts take place with inflated numbers. The largest event I know of, based on the data I was given at the time and from research I did, was the 2009 Presidential inauguration. At the time, the media reported that Akamai did 7.7M streams of the event, but once again, got the data wrong. Akamai did 7.7M live streams of all their customers combined that day, of which 3.8M were the Obama inauguration, as verified by Akamai directly. Between Akamai, Limelight, Highwinds and a few smaller CDNs, all vendors combined, I estimate they did between 7-8M live simultaneous streams of the 2009 inauguration. But no one knows the exact number. And talking about business models, for all of the bandwidth Akamai did for those 3.8M streams, the company told me back in 2009 that they made less than $100,000 on the business. So for those that think there is a lot of money in distributing live events on the web, there isn’t.
Webcasting events live on the web has been going on for more than 15 years now and it’s time the media stops getting all giddy with bandwidth numbers and instead, starts asking the questions of how this medium can be monetized, when content owners will start to make money from live events and what changes need to take place in the market so that webcasts can be profitable events for content owners, as opposed to simply a way for someone to show off meaningless stream count numbers. In 2007 I wrote a post entitled “Webcasting Large Entertainment Events Still Unprofitable“, and five years later, as an industry, we’re still not seeing signs or discussions taking place about the business side of webcasting big events. The media needs to focus on the business, not the technology. Without a business model behind it, the technology is useless if content owners can’t cover their costs of using or deploying it.
And for those members of the media that having been picking up on the webcasting story over the last few hours, make sure you question the data. I’ve done a few interviews already where some have said to me that”YouTube said” they did 8 million streams. YouTube has not said anything of the sort. The numbers you are quoting come from articles in the Huffington Post and WSJ and none of them have quoted or said they have spoken to anyone at YouTube.
More Data On Large Webcast Events: