Overview
Helsingin Sanomat, Finland's leading newspaper, wanted to introduce personalised news recommendations to improve reader engagement.
Over a century of editorial control over news placement, introducing an algorithm raised concerns about transparency and the role of editors. Paired with another designer, I helped bridge the gap between newsroom editors and data scientists, turning a technology-driven initiative into a solution that journalists could understand, trust, and control.
- Role
- Senior Service and Product Designer
- Timeline
- 2018
Project highlights
Human-first AI
- Co-creating with newsroom editors to understand their workflows, fears and concerns
- Enabling a common dictionary between journalists and data scientists
- Enabling transparency through a performance dashboard to make algorithmic recommendations visible
Impact
- 10% increase in front page article average CTR (vs non-personalized front page)
- Newsroom adoption with full journalistic control over algorithmic content delivery with editorial configurations
Awards
INMA Global Media Awards – Best Idea to Encourage Reader Engagement – 2022
The challenge
Helsingin Sanomat was moving from manually curated news placement towards algorithm-based recommendations. Editors were concerned about losing visibility and control over which stories reached readers. A technically successful recommendation engine would have little value if the newsroom couldn't trust it.
The opportunity: How could we make AI recommendations transparent and give editors meaningful control over the results?
Starting with people
We worked closely with editors to understand their daily decisions, expectations, and concerns about personalisation. While algorithms could adapt recommendations faster than manual curation, editors brought something equally important: journalistic judgement and responsibility for the newspaper's overall quality.
Our goal was to find a way for both to work together, rather than replacing editorial decisions.
Making the algorithm transparent
We designed a dashboard that gave editors visibility into article performance. By combining CTR, impressions, and article freshness, editors could evaluate stories beyond popularity, understand the overall news offering, and make informed decisions.
Together with editorial controls, the dashboard made the recommendation engine transparent and manageable without compromising editorial judgement.
Outcome & impact
+10% average article CTR on the personalised front page compared with the non-personalised version.
Newsroom adoption with editorial control over algorithmic recommendations and a dashboard to better inform editorial decision-making.
INMA Global Media Award 2022 for Best Idea to Encourage Reader Engagement.