At Amedia, our mission is to ensure that strong local journalism continues to play an essential role in people’s daily lives. With more than a million people visiting our sites every day, we know we face a growing challenge: connecting each reader with the content that matters most to them.

In 2023, we launched a personalisation initiative aimed at addressing this challenge. Our ambition was clear: to make it easier for each user to discover content that is relevant, engaging, and trustworthy, without compromising editorial integrity.

The starting point

Although our newsrooms produce a wide range of high-quality journalism, only about 50% of users were seeing even half of what was available. To us, this indicated a distribution problem, not a content problem.

We set out to improve content discovery across our digital platforms, with the ultimate goal of strengthening user loyalty. Our data has shown that daily usage is the strongest predictor of long-term subscription retention. We believed personalisation could help drive this.

Our approach

Rather than relying solely on complex AI systems, we took a practical and transparent approach. We developed a personalisation engine based on user behaviour, content metadata, and editorial input. This engine allows us to present tailored news experiences across our network of over 100 news brands.

Some of the features we have tested include:

• Segmented top lists: We group subscribers into segments (such as age or location), then display popular stories within each segment to users who haven’t yet seen them.
• Boost button: Editors can flag stories as particularly important, helping ensure these are widely distributed, regardless of algorithmic predictions.
• News in short: Short summaries presented to non-subscribers to encourage engagement and return visits.
• Sales winners: Stories that tend to drive subscriptions are prioritised for relevant users.
• Geographic relevance: We show more locally relevant stories to users based on location.

Throughout the process, we have focused on editorial control and transparency. Personalisation at Amedia is a set of tools to support our editors, not a black box that dictates the front page.

What we’ve learned

One key insight is that simple, rule-based solutions can sometimes outperform more advanced machine learning models. For example, our segmented top lists — based on straightforward filtering logic — delivered better results than more complex predictive systems.

By testing features across different titles and audiences, we’ve been able to iterate quickly. In one case, personalised teasers performed 81% better than manually placed ones. In another, automated sorting of part of the front page resulted in 12% more clicks, compared to manually curated versions.

Results so far

From October 2023 to October 2024, we’ve saw:

•  A 4.4% year-over-year increase in daily usage among subscribers.
•  A 14.5% increase in the number of stories that reach significant audience thresholds.
• Growth in digital-only subscriptions from around 430,000 to 465,000. 
• Users are exposed to more journalism per visit, not less.

These results are encouraging, but we see them as a starting point. Currently, personalised content makes up a relatively small portion of our front pages. Based on early results and newsroom feedback, we plan to expand the use of personalisation carefully and responsibly.

Looking ahead

Our personalisation efforts are part of a broader strategy to improve user experience and editorial reach. We believe better distribution helps good journalism travel further — especially when guided by the values of transparency, local relevance, and trust.

We will continue to refine our tools, listen to our editors, and learn from our readers. Our goal remains the same: to make sure the journalism we produce reaches the people it is made for.

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