Maximize reader revenue from personalized, user-controlled news applications.
Traditional publishers who seek to deliver fact-checked, quality journalism online face strong competition. Global tech giants have massive resources to build complex, personalized digital applications. These raise customer expectations of all digital news tools, as well as the expectation that the media will be available across all platforms, especially mobile.
To survive in this environment, traditional publishers need to present consumers with a convincing case for their direct revenue-based business models. And to maximize the willingness to pay, they need to provide a convenient and customizable, frictionless, functional experience that makes the most of new technological tools. They also need to make it easy for consumers to access their tools seamlessly across all platforms and devices, with easy payment options that nonetheless allow for different levels of access and use.
The NewsButler project aimed to research and develop a demonstrator incorporating an intelligent theme and article recommender and an editorial optimization engine. This engine would act like a ‘digital butler’ helping readers and editorial teams to serve every reader with a personal, user-controlled news experience, as well as assisting readers to intelligently explore new subjects. In addition, the project sought to explore new channels of news distribution and interaction, and it investigated new monetization models for quality content based on better aligning pricing with consumption and levels of engagement.
The project led to three key results.
1. Recommendation engine for readers and editorial teams
The project partners began by designing a personalized content recommender system aimed at readers. This engine suggests themes and articles, including articles both within and outside the user’s subscribed themes. The article recommender combines content embeddings, popularity and recency with user profiles in a hybrid engine to optimize accuracy and diversity (to address the phenomenon of ‘filter bubbles’). The theme recommender extends the widely used Bayesian Personalized Ranking recommendation framework to consider readers’ content consumption. The partners also designed a pipeline to recommend suitable themes to the editorial team. These themes are based on reading behavior and combine an interpretable bi-clustering algorithm with a post-processing step. This filters the most relevant bi-clusters and translates them into content themes.
2. Insights into subscription models
The project partners observed high adoption potential and willingness to pay in several use cases. The first was in users extending their subscription from one title to multiple titles (bundles). Secondly, there was high interest in family subscriptions, particularly among ‘hard news’ brands (Knack, Trends). Thirdly, there was relatively lower but still relevant interest for family subscriptions for lifestyle brands. An imec.icon research project | project results NEWSBUTLER Maximize reader revenue from personalized, user-controlled news applications SMART INFOTAINMENT
3. Results relating to new channels for news distribution
The project also aimed to identify how news can be delivered to users through various channels of their preference in a variety of ways. It explored ambient (omnichannel) solutions. This showed a clear interest among consumers to receive news which is bundled together and sent to them, either through conversation bots or via audio (voice-based) user interfaces.
An intelligent, personalized news engine for consumer engagement in publishing.
NewsButler is an imec.icon research project funded by imec and Agentschap Innoveren & Ondernemen.
It started on 01.09.2018 and ran until 30.04.2021.