What is the Optimal 360 Video Length for Social Media?

Brief Case Study

[Insert mandatory paragraph about VR being a new medium, this is the far west, trial and error, we are all learning, new paradigm of filmmaking, blah, blah]

How long should our 360 videos be? I think it’s safe to say that I’m not the only one pondering this question. Actually, it’s a question that pops into my mind more often than not while editing (and also while not editing).

Tired of being a “victim” I decided to take matters into my own hands and make a quick case study of around 40 projects we’ve worked on during the past 6 months and which are hosted on YouTube -many thanks to those clients that shared their analytics with us-.

Please be aware that most of the sample used for this post was mainly comprised of commercial projects for brands -with a few exceptions-. Also, I will not focus on viewer demographics other than the platform and OS they used to consume the content.

Out of those 40 projects, I isolated a few variables and picked out 7different projects plus 4 more to be used as the control group.

How much of my content is viewed?

The first thing that caught my attention as I went over the stats, was the Average View Duration of the whole playlist.

Average View Duration: 1:07 minutes

In all honesty, this was a surprise to me. Usually, when the topic is discussed among the VR circles or with clients, the consensus is around an optimal playtime of 3 minutes tops.

So I went digging deeper and contrasting video duration (and theme of the clip) to percentage viewed.

I drew an imaginary line at 2 minutes and picked two different videos:

The first one was a drone clip we made for a BTL action for a beer company:
Total duration 1:26
Average Watch Time: 1:21
95% Average Percentage Viewed
The second clip was a static shot (no cuts) for a different beer company:
Total duration 1:16
Average Watch Time: 1:16
100% Average Percentage Viewed

I could argue that the missing 5% on the former belongs to the last seconds of the video (5 second company logo on outro).

On the other side of my imaginary 2 minute mark, I analysed 5 of our most viewed projects:
Vicky X (a local troublemaker-slash-sex-bomb female celebrity)
Total duration 5:45, Average: 1:08
20% Average Percentage Viewed
Wedding Video 
Total duration 3:56
Average Watch Time: 1:07
28% Average Percentage Viewed
Chaman’s Demo Reel 
Total duration 2:40
Average Watch Time: 1:21
51% Average Percentage Viewed
Music Festival 1 
Total duration 2:18
Average Watch Time: 1:13
53% Average Percentage Viewed
Music Festival 2
Total duration 2:17
Average Watch Time: 1:04
47% Average Percentage Viewed

As you can see, there’s an emergent, and almost obvious, trend on these numbers: there’s an inversely proportional correlation between duration and how much of the video is viewed.

Note that balance between both variables is achieved between 70 and 80 seconds.

So, based on the <2 minute analytics, as long as we stayed close to the 1 minute mark we are safe. One theory I’m pondering is that if the viewer is past the 1 minute mark and believes that the video is about to end, they will give you those extra 10 to 20 seconds to close the story up. Meaning that if we are shooting for a 1 minute total duration but due to the content’s characteristics you can’t avoid getting past the mark, generate some tension/anticipation on the mark to buy a few more seconds of attention. This hypothesis correlates well with the graph above but still more data is needed to make a conclusion.

Here’s another way of looking at these same numbers only now, I’m correlating duration with how much of the piece was not watched.

The reason why I decided to represent the dataset in this other way, is to illustrate that even when the correlation is not logarithmic (1st pic), when presented in this other way, it sure does show that the longer the total duration, the increasingly more of your content is not watched.

Platforms and OS

I have no way to discriminate between users watching the content in through HMDs or on a traditional screen. I bet a one digit percentage of Android and Windows users (but mostly Android users) are using HMDs, so I am assuming that almost all of the content was streamed through a regular screen.

Here’s the breakdown:

Mobile: 56% (Average View Duration: 0:48)
Computer: 39% (Average View Duration: 0:49)
Tablet: 4.6% (Average View Duration: 0:49)
TV: 0.2% (Average View Duration: 0:20)
Game Console: 0.1% (Average View Duration: 0:17)

My hypothesis for the big drop in Avg View Duration on Smart TVs and consoles is due to their inability to properly display 360 video.

OS 
Android: 48% 
Average View Duration: 0:44, Average Percentage Viewed: 36%
Windows: 31%
Average View Duration: 0:47, Average Percentage Viewed: 38%
iOS: 12%
Average View Duration: 1:03, Average Percentage Viewed: 49%)
Mac: 7,5%
Average View Duration: 1:00, Average Percentage Viewed: 44%)

One last thing that caught my attention, was that Apple users tend to stick longer when watching the same content when compared to users of other brands. Perhaps we should reconsider the “no VR for Mac until Apple gets its stuff together” policy.

Final Words

This post is in no way a scientific study of consumer behavior and its conclusions were drawn using a limited dataset with a limited number of variables. Still, I think it does illustrate the relationship between video length and engagement with 360 content and at least personally, it provides me with some bearings about how to structure the content we produce to optimize the experience and the bang-for-buck of our clients.

I’d love to hear about your experience and thoughts about this topic!