Adapting to meet reader expectations
Telegraph Media Group (TMG) is the publisher of The Daily Telegraph, a newspaper trusted by readers for over 165 years. “We have a loyal, long-standing customer base,” reflects Group Technology Director Dylan Jacques. “The quality journalism game has been through lots of different paradigms, with industry-wide structural decline in print circulation. However, we have adapted through an ad-funded website model and then transitioned again into subscriptions.”
Newsletters have become an important vehicle for TMG to convert casual readers into paid subscribers. “Newsletters are primarily top-of-funnel content sampling,” Jacques explains. “They’re designed to drive people to the website and into a paid subscription model once they're familiar with our content.”
Tailoring newsletter content to readers’ interests has traditionally required editors to manually curate articles – a time-consuming process that limited The Telegraph’s ability to scale its newsletter offerings. They wanted to employ artificial intelligence (AI) and machine learning (ML) to automate the content curation process, while delivering personalized content. This would ease the burden on journalists, who in turn can focus on producing high-quality editorial content.
Rethinking newsletter content curation
In 2021– 2022, TMG partnered with the Google News Initiative (GNI) to rethink their newsletter personalization and production strategy. They had about 40 digital newsletters in production, all requiring hands-on curation by journalists. “We couldn’t scale them without incurring a disproportionately higher human cost,” Jacques notes. “We were keen to draw on Google’s experience in data-driven use cases and optimization to lower costs and increase performance.”
With support from GNI, TMG embarked on a year-long process to use AI and ML to automate newsletter article curation. “AI wouldn’t write the content, that would remain with our highly skilled journalists. But it would understand what that customer is interested in and match them with relevant articles,” Jacques says.
