Longer & More Popular (no, this is not about porn)

Last week the numbers for online video consumption in the U.S. came out from comScore (basically, the internet-video version of Nielsen, but site-specific). I found out about it through Marc Hustvedt’s great online video resource, tubefilter.tv.

Two key takeaways:
– 6.3 BILLION viewing sessions. Everybody is watching internet video, and watching more and more of it.
– The average video duration is 5.4 minutes. It’s been climbing steadily from December 2007.

Think about that second stat for a second. If you’re coming from TV land, 5.4 minutes doesn’t sound like much screentime. But if you’re coming from web video land, this is huge.

When my partners and I started The Burg way back in 2006, comScore wasn’t even around, but online ‘common sense’ was. This common sense told us, nay, SCREAMED at us, that a minute and a half was an ideal video length, and anything longer than 3 minutes was suicide. This frightened us, as a typical episode of The Burg was anywhere from 14 to 22 minutes long. So, we played it safe. We began to create shorts of anywhere from 30 seconds to 5 minutes (yes, even our shorts were longer than most people’s ‘longs’.) We interspersed our normal eps with our shorts. We fully anticipated looking at our viewcount and having much less views for our longer eps.

The exact opposite happened. Our views actually went down every time we posted a short. Why? We don’t know exactly. But after talking to fans, we have a pretty good idea: they were outside the chronology of the story. They were little jokes, standalone scenes, things that didn’t fit in the tightly plotted and structured episodes of The Burg. And so, people didn’t care as much. I knew right then that we had something good. People were hooked on the structure and the story and didn’t mind the length.

Ever since then, I’ve rigidly maintained that length should not be the top deciding factor when you’re creating your content. I’ve been mocked for this, as there are many creators who believe otherwise. In the early days, there were a lot of 90-second episodic thrillers. For me, even when well produced, the story jolted and jittered, because 90 seconds of a thriller is enough to get you to a cliffhanger, but usually seems to stop short of great character development. When working with other online portals, I’ve had to cut 5 minute shorts to under 3 minutes, and in the process, lost some of the best moments because they just didn’t fit. (This is not to say you can’t make a great episode in 90 seconds – of course you can. I just think you shouldn’t have to.)

I get why this happens in TV. You have a fixed inventory of time. 22 minutes for your sitcom once commercials are factored in. It becomes a surgical process (and to a degree, it should with all content). But on the internet, there is no such restriction. Yet content creators and programmers decided to all limit themselves to one anyway. It seems cynical, arbitrary, and a big underestimation of viewers’ tastes.

Well, it seems that common sense was wrong in this case. People are now, officially, measurably, watching longer and longer video. And 5.4 minutes is the average. Meaning, many people are consuming video that’s much, much longer.

As Hustvedt states, “If the same trajectory were to be taken forward a few years, which is probably a conservative estimate given the current market, we’d expect to see average online video duration at 10.4 minutes by 2014.”

Which means, by my shoddy estimate, people are going to be ready for The Burg by… let’s see… May 2017. Hm. Oh well. Better 11 years early than too late.

Advertisements

2 Comments on “Longer & More Popular (no, this is not about porn)”

  1. GraphsAreFun says:

    It might be more informative to look at the median, as opposed to the mean. Or better yet, a histogram. Based on The Burg and assorted anecdata, I would predict that a histogram of viewing durations would show a large peak corresponding to shorts (e.g. 1-4 min clips on YouTube*) and another, smaller bump to the right corresponding to lower frequency, longer duration views (e.g. network shows and, to a lesser extent, web-specific content).

    This graph would represent less of a “long-tail” situation than a “two overlapping populations” situation. In which case, the goal can be restated as adding more data points (aka viewers) to the righthand population, which is not the same as simply increasing total content viewing duration.

    In order to predict future viewing patterns, it would also make sense to look at the growth of the righthand peak over time, rather than the mean, especially if the total number of hits has increased in both peaks. (This seems likely, as there is more content on YouTube and similar sites, and more people have easy access and are inclined to use it.) Even if both peaks increased by the same percentage, the sheer number of short-duration views would swamp the increase in long-duration views. This kind of analysis might unmask real, but slightly subtler, viewing trends in a way that would strengthen your case.

    [*This also doesn’t take into account videos on YouTube that are spliced up into many 5-9 min segments, which people watch consecutively.]

  2. GraphsAreFun says:

    Oh, also: In terms of future predictions, you need to factor in that the increase in accessibility to video on the interwebs isn’t linear (Moore’s Law of Streaming?). Another point in your favor.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s