Is Online Video Growth Really Slowing Down?

The headline on data that Nielsen released last week: growth in online video time spent is outpacing the growth in the number of users.   Growth in total minutes for the most recent period, August 2010-August 2011 (9%) is more than double the growth for users (4%) and higher than the growth for streams (7%).  Thus the average stream is longer than it was a year ago.  The chart is shown below.

This trend in rising length of stream is apparent from data in Nielsen’s January press release…I pointed it out here and noted that the finding makes sense; the audience for online video is becoming saturated while the upside for time spent is tremendous.

According to Nielsen’s June data, time spent for the average online video viewer is 4 hours/30 minutes per month.   Compare that to average time spent per TV viewer at upwards of 30 hours per week and it’s clear that online video time spent has nowhere to go but up particularly as consumers increasingly watch long form content over-the-top on their TV screens.

That said, what seems odd about the chart, not explicitly called-out in the accompanying text, is that growth for all the online video metrics has slowed severely over the past 12 months.   I’d expect growth in users to plateau (as per the chart) and I’d expect growth in number of streams per user to slow as people watch longer streams.  But it’s surprising that total time spent with online video is growing at less than 10%, year over year.

It’s particularly surprising since the base for time spent per user is so low.   To get a sense of how low it is, take a look at Nielsen’s June press release and do a little back-of-the-envelope on time spent per stream.  It’s only 2 minutes/40 seconds, on an average of 101.5 streams per viewer per month.  So, according to this information, the average viewer of online video watches somewhere around 3½ streams per day at less than 3 minutes apiece.   You would think, given Netflix, Hulu, Amazon and other long-form streaming options, that total time spent would be growing at more than 10% off this base.

Another interesting back-of-the envelope:  calculate the average streams per user and time per stream for the three competitors for which the press release provides monthly user, streams and time spent data.    For YouTube (with 108M users): 81.7 streams at 1:55 (mins/secs) apiece.   For Hulu (13.5M users):  46.7 streams at 4:45 apiece.    And for Netflix (8.0M users) 24.1 streams at 21:18 apiece.    The total is tilted toward the YouTube number with its enormous user base, but the Hulu number is surprisingly short and even the Netflix number, at less than half an hour per stream, makes me wonder about what might not be included in the data.

The shape of the Nielsen growth curve is contrary to the one below, for example, from Cisco data cited in ReelSEO earlier this year.   This predicts that bandwidth demands for streaming video will not decelerate but in fact accelerate over the coming years.

I suspect the Nielsen time spent growth curve is affected by a particular aspect of their methodology; they monitor online video activity on their respondents’ computers.    The way I read it:

Online video that people access through games consoles, Internet-connected TV’s, Roku boxes, Apple TV and any other device that circumvents the computer  and ports online video directly to the TV screen are not included.   This is probably not a substantial piece of the puzzle on a user basis or on a streams basis.   But the absence of this piece may understate the growth of time spent per stream with the increasing consumption of longer streams concentrated in non-PC devices.

This isn’t a knock on Nielsen’s methodology – it measures no more and no less than it says it does.  But if we’re really going to understand the total picture of online video behavior we’ll need to get a handle on the whole ball of wax no matter what devices the video streams flow through.

What’s Driving Online Video Growth?

Nielsen and Comscore online video data look like they come from different planets, as ReelSEO points out.    The most dramatic difference is in hours per viewer per month; 4.7 hours for Nielsen as of January 2011 versus 14.2 for Comscore, December 2010.    Yes, both are for U.S.

But there’s another headscratch, aside from how different the numbers are.   Each source tells a totally different story about where the growth in online video is coming from.   Completely different answers to the question:  more users or more usage?

Take these Jan 2011 Nielsen numbers for example.  They seem to say that the audience for online video is saturated and growth comes almost entirely from dramatically more usage from the same people.  I calculated a number from those provided that suggests: growth in time spent comes from more videos per person; not so much from increased time per stream:

Nielsen Online Video Data: January 2011

Now Comscore, for Dec 2010, from their 2010 Digital Year In Review: 

Comscore Online Video Data: December 2010

Granted the measures are apples and oranges.   Still, the Comscore data seems to say that online video users are increasingly dramatically while usage growth is more modest.   A completely different story.

With no bias toward sources or methodologies, the Nielsen growth story is the one I tend to believe.   I would think, at this point in its evolution, most everyone who’s going to watch online videos is already doing so (not counting new users coming in as kids mature).    I would guess that growth is coming from changing behavior among those who’ve already caught the habit.

Clearly there’s tremendous usage upside for online video.   If you think about the higher (Comscore) number for online video consumption, 14.2 hours a month, and compare it to consumption of TV, at some 35 hours a week.   Given that comparison you’d sort of expect usage to be the dynamic driving online video growth.

The question for either story is the detail behind it:  

  • If new users are pouring into online video, as per Comscore, who are those users; what are their demographics or other characteristics?  
  • If the same users are consuming dramatically more streams as per Nielsen, how are there habits changing?   What types of video content are making up the difference?   

And why do both sources show the average length of stream to be so short (both under 5 minutes) and growing relatively modestly compared to the other metrics?  Wouldn’t we expect, given the growth of Netflix, for this measure to be growing dramatically?   

I would love to break this information down into user segments.   Because I would bet there are segments of consumers for whom average length of stream is much longer than the average and segments for whom this metric is growing more sharply than the others.   When those segments start to drive the total sample average, when average length of stream really starts to grow, that’s when disruption of the TV business will be under way.