Audience engagement: What broadcasters can learn from YouTube and Twitch

In reality, the tools and data-led approach that makes online platforms so effective can be harnessed by savvy rights owners, broadcasters and OTT players to personalise the experience of their viewers, and bring about a generational leap in engagement, according to an expert panel composed of Tom Dvorak, Co-founder XRoadmedia, Roland Sars, CEO of Media Distillery and Ed Abis, Managing Director, DIZPLAI.

Big data requires smart automations

While the promise of big data is significant, at least some degree of automation is essential in order to harness it, because the new ways of honing a content offering are likely to involve data-heavy processes. It’s a move from thinking about “recommended” content to a more holistic “personalisation” approach. The more you know about your audience, the more successfully content can be tailored to their wants and needs.

Tom Dvorak, co-founder of XRoadmedia described this as: “getting a little bit away from the mindset of recommendations, because I think that’s old school.

“I think recommendations is really the past. I think it’s time to think about content discovery and personalisation, and make sure that the user is at the centre of your delivery,” he said.

Making this a reality requires not just a certain level of knowledge about your viewer, but a more advanced level of insight into the content being distributed. In the traditional model, a viewer’s experience of discovering content relies on its title and a series of metatags.

These are key words linked to a piece of content that might signal its genre, key themes, its director and so on. However, a traditional approach will never dig deep enough into a content library to represent all the connective tissue inside it. Tools such as Media Distillery use artificial intelligence to analyse content on a much deeper level, claimed Sars.

Read more Artificial intelligence in broadcasting

“We constantly and in real time analyse based on speech, based on faces, based on overlay detection, based on all kinds of elements that are happening in the screen, whether sounds or visuals,” said Media Distillery CEO Roland Sars.

Facial recognition is used to recognise actors and other figures in videos. OCR registers any written text, logos seen on-screen are analysed in speech is deciphered to work out themes present in the content.

Sars called this “deep metadata”. Using this process Media Distillery typically finds an 82% increase in items discovered in searches, largely thanks to an additional 22 “topics” linked to each programme “on average”.

“It’s a combination of understanding the customer, and then understanding the content,” said Sars. Viewers can make much more specific searches, allowing niche and special interest content to surface easily when it is wanted.

“I think this is the key ingredient that YouTube does very well because they understand [customer and content] very well. And that’s why their recommendations are often quite meaningful and you keep hanging in their platform once you’re there. Because on the right you have all kinds of meaningful recommendations relevant to the content that you’re looking for, or relevant to content that you’ve been looking for in the past.”

Automation in a live setting

This data-hungry approach can be used in the creation of the content, before or even during its production too.

One company that specialises in this is DIZPLAI. The company works with broadcasters, sports associations and advertisers to make user engagement part of an event or broadcast.

This will often start with a social media call to action and a hashtag, inviting viewers to participate.

“We worked with a show in the UK for ITV called the Martin Lewis Money Show,” said Managing Director of DIZPLAI Ed Abis.

“The idea was it was going to run for a couple of weeks and it’s still running do this day. The audience could literally share hat they wanted throughout the week, on any topic, on any matter. And the producers would trawl through it.”

The potential problem when attempting this user generated content-led approach is the sheer amount of correspondence likely to come in.

“The trouble with that is we have thousands and thousands of bits of content. — the producer would literally not have enough hours in the day to go through to see what people are saying. There are ways we’ve tried to help them — an old-fashioned word cloud helps us because we can see the topics that are popping up, so the producers can hone in. All of this is effectively metadata,” said Abis.

DIZPLAI then creates visuals of the user-supplied responses, which can then become part of the programme itself. It marries some of the character of a classic phone-in with that of modern online platforms like Twitch, where the interaction between host and viewer is paramount.

Finding the balance

However, in either case — of using a data-driven approach to content delivery or content creation — a sense of balance and proportionality is important. It should be driven more by the viewer’s desires than those of the platform holder.

“The consumer really quickly understands if it’s something just pushed because they want you to watch it, and that’s not the best UX anymore,” said Media Distillery’s Roland Sars.

“My personal experiences is that with Netflix, they just keep on pushing the big shows they have, not because they think you like it, but just they want you to watch it. That is not going to work in the coming years anymore, especially with more and more video platforms. UX is getting more and more important for the selection of what video platform to choose.”

The key is that the viewer should feel in control, that personalised content is an empowering force, not one attempting to manipulate their choices, as Tom Dvorak explained.

“It’s coming back to control — how much do they want to be in control versus how much the user wants them to be in control. And I think there’s a little bit of mismatch at the moment,” he said.

The goal is to demonstrate the value of the personal information each user effectively hands over through the quality of experience they receive. At this IBC2022 panel discussion all three speakers agreed that when the balance is right a person is only going to be too happy to hand over this data — when it clearly being used to empower, rather than exploit, them.

Co-founder XRoadmedia Tom Dvorak, CEO of Media Distillery Roland Sars and Managing Director DIZPLAI Ed Abis sat down to talk about personalisation and engagement in content at an IBC2022 panel discussion chaired by Ian Nock, Founder of Fairmile West Consulting.

Read more Paramount on Personalisation, FAST and the ad value chain