There is probably no other content metric that is as important, and yet so complex, as the video view. Since the middle of the last century, when TV started to take over the living room, to the rise of TikTok in the last few years, the same question has been asked by everyone from programme makers to schedulers to advertisers to on-screen talent – is anyone actually watching this?
The ways in which we’ve answered this question have changed dramatically, and as they have changed, they’ve upended the billion-dollar industries that make and distribute moving images. So here’s a short summary of the way we’ve counted views over the years. And just as an aside, if you’ve read this far, most social video services would count this as a ‘view’. Hold that thought.
Why should I care?
Because just as software is eating the world, video is eating online content. Because of the rise of TikTok, all the major social platforms (with the notable exception of Twitter, very much the crazy country cousin of the current media landscape) are turning into streaming video platforms. And this has made social media less, well, social. We spend a lot more time passively watching video on these platforms than we do actively creating it.
UK content regulator OFCOM’s annual report on children’s media lives showed that the spaces they use to chat with friends and the spaces they use to consume content are separating. As a result, the content they watch is more from professional producers and creators, and less from friends and peers.
What’s the big picture?
It all starts with a black box, the size of a couple of hardback books, that kickstarted our current era of attention measurement technology. The Nielsen Audimeter was a metal box sent out to a sample of households across the US from the late 1940s and fixed to their TVs. It contained a 16mm film canister that captured little points of light, recording the channels you watched on your TV and when you watched them. When finished, it would deposit a quarter to encourage the owner to send the film to Nielsen for developing, so the points of light could be turned into data, and that data could be turned into TV ratings. These were just small samples, though – at the height of TV’s dominance in the second half of the 20th century, the number of ‘Nielsen Families’ sampled to create TV ratings was just a few thousand. The vast majority of us were ghosts – unmeasured and uncounted.
Then in the early 2000s, the internet came into our homes – then YouTube, iPhones, VOD platforms and social video. All these platforms returned data about every second we watched and every play button we clicked. We are no longer ghosts, and the way we are measured has in turn completely changed the media industries.
TV Ratings used to be analysed by schedulers and media planners to decide what future shows to commission – you’d only know much later if your hunches, based on small samples and broad demographics, were correct. Now, algorithms create custom playlists for every single one of us, based on real time aggregation of billions of points of data, and creators use that data to quickly iterate and optimise their video formats.
So what does this all mean?
Video has ended up dominating social media platforms for a couple of reasons, but the main one is money. Because the creative gold standard of advertising was – for years – the 30 second TV spot, the amounts you can charge for video advertising slots has always been way more than other forms like search or display ads. So after Facebook and Google had taken the display ad budgets from newspapers and magazines, they turned their attention to getting some of the much bigger budgets spent on TV ads. But this raised an interesting problem – when you have so much data about user behaviours around video, what should you count as a view?
During the decades when TV was king, it was in the broadcasters’ interests to agree on an independent standard metric, working with companies like Nielsen to produce TV ratings that advertisers would trust. But the new digital platforms of the last 20 years wanted to mark their own homework, each of them using different metrics to measure a ‘view’. This has led to a huge diversity of what a view is – on most social platforms it’s now only a couple of seconds, whilst Netflix’s most important metric is how many people complete a whole season within the first month it is released.
So we’ve gone from an era when the amount of data was tiny, but everyone believed in the same analysis, to one where the data is almost infinite, but it can be interpreted into many many different stories. That’s because the stories differ depending on what decision you’re making – if you’re optimising an algorithm, you want to know if someone stops swiping long enough to watch something; if you’re commissioning a hugely expensive bingeable series, you want to know that people won’t drop out after the second episode; if you’re selling advertising, you want the smallest amount of attention that your advertisers will accept to represent a real person watching; and if you’re a creator, you want as much data as you can to know where your format’s strong and weak points are.
Tell me something I don’t know
We might feel uneasy about how much we are tracked by our devices now, but that wasn’t always the case. In 2004, a proposed change to Nielsen’s TV ratings technology was opposed by a ‘Don’t Count Us Out’ campaign, arguing that the new technology would undercount black audiences, and calling for Nielsen to make sure they included them in their measurement. But the campaign wasn’t as grassroots as it seemed – the campaign was funded by Rupert Murdoch’s News Corp, who thought the new metrics would massively impact the ratings for their right-wing Fox News Network.
In the golden era of TV from the 1970s-90s, writers and directors often wouldn’t get any data about TV ratings. In this clip from a documentary about the cult British sitcom Blackadder, writer and director Richard Curtis tells the story of how he used to walk the streets of London looking in people’s windows as the show was broadcast, as it was the only way he knew if people were watching and laughing.
Give me some numbers
The COVID pandemic led to a huge rise in video viewing. According to OFCOM, UK viewers total daily video consumption on all devices went from 4 hrs 52 mins in 2019 to 5 hrs 41 mins in 2020, although this slipped back half an hour to 5 hrs 16 mins in 2021.
Of the 30 most watched US TV broadcasts of all time, 21 of them are Superbowls, and none of the others were broadcast in the 21st century.
Ok, I want to know more
This great post from Hootsuite goes into detail about the way the YouTube algorithm has changed over the years
And this fantastic post from Gibson Biddle, former VP of Product at Netflix, gives a detailed account of how Netflix has used data to personalise content for users, from the days they sent physical discs through the post to the current VOD platform.
This Slate article goes deep into Facebook’s ‘pivot to video’ in 2014, and how they deceived publishers and advertisers about the amount of video viewing on their platform.
And Freddie De Boer tells the story of what happens when metrics drive video makers to focus on numbers and revenue more than actual content.