twitter
Videos uploaded to Twitter will now automatically play as users scroll through their timelines, following a change to the microblogging platform.

The move sees Twitter follow in the footsteps of Facebook in introducing autoplay video and like on Facebook, the videos will be muted by default. When a user clicks on the content, it will expand to full-screen with sound. The new feature applies to all video, GIFs and Vines uploaded via tweet.

Twitter explained the changes on their blog, revealing some impressive-looking statistics to sell the virtues of autoplay video. These stats, gained through what Twitter describes as “extensive testing” were:

  • Users were 250% more likely to prefer autoplay videos over other viewing methods (including click-to-play and video preview thumbnails).
  • Users had a 14% lift in video recall, when comparing autoplay to other video formats.
  • Brands saw a 700% increase in completions of Promoted Videos via autoplay.

On the face of it, those are some very impressive figures. Armed with that data, marketers should rush out and fill their brand’s Twitter feed with video content, right? Well, not so fast. As ever, the devil is in the detail.

Do users really like autoplay?

facebook autoplay video

Let’s start with some of those claims by Twitter, that users were 250% more likely to prefer autoplay videos. Does that stat really stack up?

Satista, an American survey group, asked focus groups about their views on autoplay videos on Facebook and Instagram in September 2014. It found that 53% of respondents disliked autoplay videos, with just 15% expressing positive opinions on autoplay.  Just under one fifth, 19%, had no opinion, while 13% said they’ve never come across autoplay videos online.

Mobile users are more likely to dislike autoplay, which is why the ComScore claim referred to earlier about Facebook overtaking YouTube only applied to YouTube. Many mobile users have actually disabled autoplay, to protect themselves from excess data use charges and preserve the battery life of their smartphone.

The stats behind the stats

facebook video

The statistics were produced following experimentation at Twitter HQ. What we don’t know is how the results from Twitter’s sample tests will compare with the results from the real world outside of the Twitter bubble.

To add some context to these statistics and claims, we can take a look at some of the data from Facebook, which has been running autoplay video since late 2013 in the US, and mid-2014 in the UK.

The introduction of autoplay resulted in a bold claim being made by ComScore – that Facebook had overtaken YouTube as the king of video for desktop users, by number of videos consumed.

But there was a strong caveat to that claim. YouTube only counts a view when the user initiates a view (in other words, when they hit the play button). Facebook, on the other hand, counts a view as any occasion where a user has dwelled on a video for at least three seconds, irrespective of whether they initiated the video play. Facebook also uses a statistic known as ‘reach’, which counts every session that a video appeared in a news feed, even if the user completely ignored the content. The two measurements make for an unfair comparison. Facebook is counting passive viewers in the same way that YouTube is counting active engagement and video play initiations.

And we can see the problem with this problem ourselves at Stickyeyes with our own video series, Digital Minute.

Digital Minute is hosted on a number of channels, including Facebook, but the primary channel that we use is Wistia, which is what powers the video in our Twitter feed and on our blog.

digital minute wistia screen

The above graph demonstrates the viewing statistics for an edition of Digital Minute from January 2015. The episode (viewable here) discusses Google’s warnings about mobile friendliness.

The blue line is what we are interested in, as this represents the proportion of users who are still engaged with the video at any given time point. The average figure for this video is 78% engagement, indicating that the average viewer consumers 78% of the content, or 56.16 seconds of a 1:11 video.

But when we compare the same metrics, on the same video, to Facebook, we see a very different picture.

digital minute facebook screen

The Facebook video in question has a comparable number of video views (using the Facebook definition of a user dwelling for three seconds) as it does on the Wistia platform, but the engagement measurement looks significantly different. Where the viewer drop-off rate in Wistia is very gradual, Facebook sees a very pronounced drop-off in the opening 5-10 seconds of the video. The engagement rate then plateaus around the 15-20 second mark.

Facebook reports that only 23.5% of viewers reach the 30-second point of the video, while the average view time for this same video is just 24 seconds, or 33.3% of the overall video.

It raises a lot of questions about the value of branded video on social media platforms, and how effective they are at engaging audiences – something that Twitter claims to be addressing with the way in which advertisers will be charged for video views.

Twitter claims that it is “delivering a new standard for how brands will be charged for a video view”, by only charging advertisers for a video view when a video is 100% in-view on the user’s device, and has been watched for at least 3 seconds. If a video is only 99% viewable, or if the user doesn’t dwell on the video for more than three seconds, then an advertiser won’t pay. It’s a bold move, and responds to one of the biggest demands from brands; namely, that advertisers want assurances that their videos are actually being viewed and that they are paying for engaged viewers, rather than for passive consumption of their content.

So how do advertisers make autoplay work?

Autoplay isn’t going away any time soon, so the onus is on brands to work with it.

Social media is a place for criminally short attention spans, and that problem is even more pronounced when a user has a video thrust upon them, playing without them asking it to play. If you’re going to get the attention of that audience, you need to make an instant impact.

That is undoubtedly going to influence how brands go about making video for Facebook. We have already seen a growth of micro-video content, in the form of Vine, and in order to retain viewers on autoplay channels, we should expect to see more of this in the future. If content can make an impact immediately, users are more likely to remain on that content.

We have seen examples in Trueview advertising of brand content that is specifically designed purely to retain the audience by rewarding the viewer. This video for Google Android actually ‘dares’ the user to skip the ad, whilst offering more and more entertainment in return.

Android: Skip Ad from NOMINT on Vimeo.

The American insurer GEICO has used a similar tactic with its ‘unskippable’ Trueview ad campaign, which again tempts the viewer to skip, right at the point where the more rewarding content is provided.

It could well be that autoplay increases the trend for this style of content.

Platform choice is also going to be a major consideration, particularly for those brands that have more considered, lengthy video content as part of their strategy. Promoting such content on Facebook may prove to be misguided if Facebook continues to serve, and charge promoters for, videos that users fail to engage with.

Want to know how to engage your audiences through online video? Speak to one of our specialists to see how we can help