Analytics In Live Video: The Challenges And Opportunities

The success of live video can be difficult to evaluate


There are many reasons for the explosion of video in social. For one, smartphone cameras have rapidly improved to a point where even amateur video content is of a high quality. Also, improvements in mobile data have meant that video has become more and more viable on the billions of mobile devices worldwide. When these developments are coupled with the comparatively strong engagement rates seen for video content, it’s easy to see why the likes of Facebook have made it central to their strategies.

Increases in connectivity have led to the natural next iteration of online content - live video. Anyone who uses Facebook’s mobile app will not only be aware of the News Feed becoming essentially a video library, but also that it occasionally notifies users when a friend’s live video begins (Facebook-owned Instagram also does this). Google has been similarly interested in live video, reflected in YouTube’s clear effort to popularize its live feature, and brands are rushing to tackle the emerging trend head-on and find ways to exploit it.

The problems with live video, particularly for brands, are numerous. For one, it is a lot more risky an output than pre-produced video; presenters can make mistakes, messages can’t be retrospectively tailored, and there is no way of having any quality control before the content is seen by its immediate audience. On top of all of this, it is difficult to administer the same level of analytical scrutiny to live video given that you can only really assess the data in real-time. This is where the social media giants are looking to improve to make live a more enticing prospect for businesses.

Livestream: 80% would rather watch a live stream from a brand than read a blog, while 82% prefer a live stream to other social posts

A lot of the progress being made in live video analytics as a result of the examination of surveillance footage. The number of security cameras in the US alone is estimated to be around 62 million, resulting in a truly mind-blowing amount of footage amassed every day. To draw any large-scale analysis from traffic footage, for example, would be near-impossible without the help of artificial intelligence like video recognition software. This technology can be used to draw real-time analysis of traffic flows, the immediate highlighting of any problems on the road, and vehicle identification and location, among other things. It’s easy to see, then, why law enforcement agencies are keen to develop the technology, for example.

For businesses, though, the analytical capabilities for live videos are still fairly simple, particularly on the mediums on which they are most likely to be published. Of course, there are companies offering software that can provide detailed analysis of live video output, but on the key social media channels these features have been somewhat underdeveloped.

In April of last year, Facebook added two new metrics to their live video analytics software as a result of feedback from publishers. One was Peak Concurrent Viewers: ‘The highest number of viewers who were watching the video while it was live.’ The other was Viewers During Live Broadcast: ‘A visual representation of the number of viewers during each moment of the live broadcast.’ Both are aimed at allowing publishers to assess their content mid-stream, rather than retrospectively. When offered alongside information like unique viewers, total views, 10-second views, and average % completion, publishers can build a reasonably detailed picture of how well their live content is performing.

Find out more about analytics and the use of data in marketing at the Big Data Innovation Summit this September 11 & 12 in Boston.

Live video is still something of a curiosity for most brands. Many have broadcast live events and many run regular live shows as part of their digital publishing schedule, but there is a sense that the medium is something most brands are still figuring out. There have been high-profile mistakes on the medium, such as Buzzfeed’s awkward non-interview with Barack Obama when the live feed stalled before the president entered the room. Gawker ran a similarly awkward event when they had one staffer go through each of his internet tabs, none of which were particularly interesting, over a 20-minute slog.

It’s difficult to know if analytics can help brands avoid missteps like these; ultimately, they come down to a lack of planning. What further analytics for live video can offer, though, is an insight into what worked about particular content at particular times during it. The more brands can get information on which specific elements of their live content users engaged with - through analytics rather than speculation - the more live content will improve and bring value to the audience, beyond that of a novelty. Live video clearly offers strong engagement figures, otherwise both brands and social media giants wouldn’t be pushing it quite so strongly. What remains to be seen, though, is if analytics can affect a piece of content’s efficacy during its broadcast, or whether retrospective analysis is all digital publishing can utilize. 

Content small

Read next:

How Content Creation Builds Your Reputation And Brand Equity