Hermes

Helsingin Sanomat was eager to explore algorithm-based news recommendations, while providing readers with all the information they need about the world and without giving an inch on journalistic integrity. They wanted to challenge their editorial processes built with over 100 years of experience and tradition.

There were two sides to the challenge:

• Build a successful technical solution that provides readers with a personalized experience.
• Enable trust in the technical solutions by building transparency between the machine and journalists.

We started by understanding the preconceptions of personalized content HS journalists may have. We wanted their voice, worries, and excitement to be the crucial drivers in the project. They were our users, and from the early stages, we understood that the success of the project strongly depended on the adaptability of our system to their daily work needs.

While a news editor, doing manual prioritization of news articles, would not be competitive in speed compared to an AI at delivering personalized content to readers, the value of a newspaper is simply beyond swiftness. It is about the high-quality and carefully curated content that makes the job of news editors an art on its own.

Parallel to our algorithm development, we created a dashboard to bring transparency over our recommendation engine and measure editorial decision-making impact. Our dashboard, compared to existing commercial tools, aimed at understanding the performance of any published article as well as the overall picture of the newspaper within a window of time, and not just limited to top performer articles. We enabled a new level of data-informed journalistic decision-making through visualizations of new approaches to performance.

By creating a solution that fundamentally diverges from commercial tools, we better-equipped journalists. News editors nowadays have a broader and greater toolset that highly supports and enhances the newspaper development.

The dashboard is now an integral part of the newsroom and editorial process, offering unique data and perspectives not available through commercial tools. The project provided valuable insights that the newspaper continues to explore.

• 10% increase in front page pageviews per user (vs non-personalized front page in Ilta-Sanomat news service).
• 10% increase in front page article average CTR (vs non-personalized front page in Helsingin Sanomat news service).
• 98% increase in push notifications CTR (vs non-personalized push notifications in Ilta-Sanomat news service apps).
• Full journalistic control over algorithmic content delivery with editorial configurations (e.g. selecting full-exposure articles).

INMA Global Media Awards – Best Idea to Encourage Reader Engagement – 2022

• Co-creation with data scientists and developers
• Concept design
• User experience & prototyping
• Interviews & testing with news editors
• Storytelling materials and diagrams for communication with stakeholders

About HS

Helsingin-Sanomat is one of Finland’s major newspapers, offering the country’s most popular mobile and web services. The company’s website serves vast amounts of data to readers, with approximately 31 million visits per week.