Among the many challenges facing the news media industry is the fragmentation of audiences. With multiple formats now serving audience needs, meeting users on their terms is possible, but news media companies are challenged with ensuring their technical environment can support those various formats.  

During this week’s Webinar, David Caswell, executive product manager at the BBC, joined Ariane Bernard, INMA Smart Data Initiative lead, to look at some of the ways publishers can approach automation and Artificial Intelligence to meet the diverse needs of their audiences.

Caswell set the scene by explaining how much the media ecosystem has been disrupted over the past few decades, noting that “we used to have this idea of the audience as this single population or this very cohesive population that bought the newspaper or tuned into the channel.”

Now, publishers face a different reality: Older audiences may still have a print or broadcast habit, while younger audiences have “lots of different kinds of people with lots of different consumption behaviours.”

That means a change in how media is distributed, but it also changes how media is produced.

Caswell pointed to generative AI and ChatGPT, which have changed how news articles can be written, broadcast, and published. Synthetic voices and avatars are becoming sophisticated enough to be used by reputable news media companies and affordable enough to be accessible to virtually everyone.

But the disruption doesn’t end with production; it is also changing workflows within companies. Caswell defined three categories of AI use in the production process:

1. Augmenting existing workflows to produce existing content items more effectively.
2. Creating new augmented workflows to produce new variations of existing content items (audio, video, articles, etc.) that require more technical or analytical guidance.
3. Producing entirely new media experiences, which is something the BBC is currently experimenting with: “For example, [the BBC has] a programme called object-based media or flexible media that is essentially a system of content production that produces dynamic experiences that change for the user in the situation that the user is in,” he explained. 

Adapting to AI

Caswell noted that many news media companies have implemented AI into their newsflows, using it for everything from automated tagging and transcription to editing and optimising images. “AI shows up all over the place in many newsrooms now,” he said.

He shared examples of enabling new augmented workflows through natural language processing, automatic assembly of image galleries, and generating different versions of stories for other regions or localities.

While much of what the new media experiences will look like is yet unknown, Caswell said “a couple of assumptions” can already be made about the future: “One is that stories are not going away. Stories are this deeply human thing. So stories are going to remain at the centre of this, but then we’ll have all these new ways to work with stories, communicate stories, interact with stories in terms of formats, and so on.”

Much of the task will be about making stories more relevant and accessible, which is critical for connecting with an increasingly fragmented audience. The stories being told may be the same, but how they’re told and distributed will change from one group to the next.

Even though the tools to do that are still in their early stages, Caswell said they would be central to enabling the technology needed to earn audiences’ attention in the future and address the needs of a changing ecosystem.

Maintaining flexibility

Caswell pointed out that two simultaneous disruptive revolutions are taking place — one in distribution and now a new one in production — in a rapidly changing media ecosystem.

“The solution is essentially flexibility of the experience. We want to have different ways of communicating stories,” he said. “We’re not just producing and communicating stories as text, but we can do them as summaries, we can do them as image galleries or graphical experiences. We can do them as video experiences of different kinds and audio experiences and on and on. So essentially, it’s about having a portfolio of formats for every story.”

This allows for customisation for differing audiences, something he said the BBC has done, for example, when covering elections in the U.K.

“In the U.K., we have four nations that have quite different media interests,” Caswell said. “We’ve been able to customise and adapt our stories to the nation for Scotland and England and Wales and Northern Ireland. That’s quite important.”

New technology allows for another critical component — contexualisation, which demographics can do.   

“Beyond locality or location, we can add different degrees of explanation or context to different audiences. So, for example, for a younger audience who hasn’t experienced or been exposed to a story that’s been running for decades, we might want to give them a little more context.”

That can also be used to meet the different reading interests of rural audiences or different age groups. It opens up seemingly limitless options for customising, targeting, and engaging audiences.

“There’s all these different ways to recommend stories and customising the way that we recommend stories to the particular audience segment that’s also quite important,” he said.

Still, this kind of flexibility depends on the proper infrastructure — and that starts with data.

“Starting with a sign-in process where if you have a significant portion of your user base signed in, then you have a way to work with that user and understand what that user values and therefore serve that user,” he said.

Storing the data from that interaction allows you to access it quickly and serve the customer better. Then, he suggested, have a mechanism in place such as a segmentation service “where you can use the behaviour and the understanding in that data from your users to divide the audience up into segments.”

At the BBC, this is referred to as the single customer view.

“Then you need ways to work with the content itself,” he added. “This is about setting up the content in a modular way where you can do different things with different modules for different segments.”

The BBC has a new modular content management system called Optimo that allows them to set up content in different ways to create new user experiences. About 5% of the BBC journalists have been migrated onto this system, with complete migration expected within the year.

This modular approach allows them to be more flexible and use pieces of content in different formats, Caswell said. It allows the BBC to generate new modules that haven’t previously existed and test them with audience segments to see how they do.

The final piece of the data-driven infrastructure is an architecture that allows the content to be served when and where it’s needed.

“We have this ongoing challenge with fragmented audiences that essentially comes from the appearance of the Internet and our ability to adapt to it,” he said, adding that it may be possible to use new and the current disruption around media production to address some of the challenges created by the disruption in media distribution.

“That’s easy to say, hard to do, but a lot of the work that we’ve done … points to ways that we can reach audiences that we’re not currently reaching and do that easier using tools from generative AI.”

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .