Schibsted News Media has a reputation for consistently delivering excellence through innovative product strategies. Notably, the group is renowned for its fearless transition to a digital-first publishing model, making early investments in video, and nimble experimentation with paywalls and new revenue models.

Most recently, Schibsted turned to audio as a way to deepen engagement with its journalism, fit into new routines, and attract younger audiences.

The audio opportunity 

Karl Oskar Teien, head of product at Schibsted, explained why he thinks audio poses such a unique promising opportunity for news publishers: “Our biggest competitors are not just other media companies; it is anyone distributing their content on a screen,” he said.  

In a recent blog post, Teien explained how Schibsted has strategically pinpointed audio as the format where it can reach and grow its audience effectively, away from dopamine-inducing social media apps.

“We know that tech and streaming giants dominate users’ visual attention, and it seems unlikely that news publishers will turn the tide anytime soon,” he said.

Despite competition for visual attention, news dominates a disproportionate share of listening, surpassing other categories. In the U.S., 40% of spoken-word audio time is spent listening to news or information.

Schibsted took advantage of this trend early, recognising the importance of building audio functionality into its core products, where the company could experiment, analyse, and monetise more effectively.

Teien said Schibsted’s primary goal is to “tap into users’ sonic attention,” thereby taking advantage of the less crowded waters offered by audio.  

In 2022, Schibsted looked to extend the success of its audio strategy by adding the option to listen to individual articles.

Inspired by the success of localised AI voices deployed by South African media group Media24, Schibsted partnered with BeyondWords.

The importance of natural-sounding speech

When considering AI audio articles, Teien said his team looked specifically at quality and listener engagement. 

To enable functionality across all articles, Schibsted needed confidence that the voice wouldn’t make critical pronunciation mistakes. It assessed AI voice partners on their domain experience and ability to offer quality that would remain “the best in the industry over time.”

Since 2018, BeyondWords has worked with leading product teams to deliver natural-sounding AI voices.

Our focus on delivering high-quality voices in the domains of news, corporate, and financial research, combined with our publisher-first product development ethos, has made us the first choice of many newsrooms.

Working with Schibsted Product Manager Lena Beate Pedersen, the BeyondWords team developed a custom AI voice based on Schibsted podcast host, Anne Lindholm.

Teien said the team wanted “familiarity with a voice that users already know.” In fact, studies show that resonance and comprehension improve when listening to a voice with a familiar accent or voice.

The new voice, internally dubbed “Aftenposten Robot” at Schibsted, was piloted on Aftenposten Junior skole in late 2022.

Pederson explained in a blog post that 92% of teachers in Norway indicated some children in their classrooms struggled to read. Some schools even suggested that, without audio functionality, they wouldn’t subscribe to the Aftenposten Junior skole product.

Her team recognised the significance of this trend and its future effect amongst subscribers.

Early listening insights

After a successful pilot on Aftenposten Junior skole, the custom AI voice was deployed to Schibsted’s flagship masthead, Aftenposten.

The Schibsted team observed data and feedback from its audience, convincing them of the efficacy of audio versions of their articles:

1. Higher completion rates

Firstly, Schibsted observed that those who clicked to listen were likely to finish more of the article. On average, audio articles on Aftenposten have a 58% completion rate. In a time when 13% of news readers cancel subscriptions because of too little time to read, this was extremely compelling.

Furthermore, feedback from focus groups signalled the audience was choosing to listen when “they really want to get the details in the story.”

Schibsted recognised this improved engagement with its stories means better value to subscribers, resulting in fewer cancellations.

2. AI audio vs human-read

Secondly, Pederson, Teien, and the Schibsted team wanted to test how the AI voice compared to human-read audio.

After comparing the engagement of the AI voice vs human-read audio, it concluded that, although completion rates were slightly higher for human-read audio, the AI voice provided the required scale and efficiency. And after a few months live on Aftenposten, audio article usage is on par with podcasts.

More AI audio in the works

Based on these early data points, several more custom AI voices have been commissioned across Schibsted brands in Norway and Sweden.

To further grow engagement, Schibsted plans to invest more time optimising its products for listening, utilising playlists, and other features.

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