How SPH Media Uses AI to Rethink Story Framing and User Needs

Singapore’s Chinese-language broadsheet Lianhe Zaobao is restructuring its editorial approach to story framing after an analysis revealed that its content often fails to align with specific audience needs. The newsroom, part of SPH Media, is utilizing a custom generative AI model to move away from a volume-based traffic strategy toward one focused on the quality of reader attention.

The shift follows the newsroom’s participation in the User Needs Bootcamp organized by WAN-IFRA. As part of the program, SPH Media’s data scientists developed a custom GPT model to analyze 6,006 articles produced over a single month. This baseline analysis was designed to identify whether the editorial output matched the actual needs of the audience.

Addressing Content Imbalance

The AI-driven analysis identified a common industry pattern: a heavy overproduction of stories categorized as “Update Me” articles. Tan Lee Chin, Digital Product Editor at SPH Media, noted that many publishers allocate a disproportionate amount of resources to these standard updates, often at the expense of other content types that could have a deeper impact on readers.

Specifically, the data suggested that emotional stories were an underserved category. The newsroom segmented its audience into general readers, registered users, and subscribers to determine if different groups required different types of information.

The findings indicated that subscribers, in particular, would benefit from a higher allocation of resources toward actionable articles. This realization has prompted the newsroom to shift its primary metrics of success, moving beyond reach and pageviews to focus on engagement and attention.

The Challenge of Story Framing

One of the most significant discoveries occurred during blind tests where human reviewers verified the AI’s classifications. In several instances, the custom GPT assigned multiple user needs to a single article, and in some cases, human editors also struggled to identify a dominant need for a story.

This classification difficulty revealed a fundamental issue with how stories were being framed. Tan Lee Chin explained that the newsroom frequently attempted to pack all possible angles into a single article rather than building a story around a clearly defined user need.

When we try to write an article, we pack all angles in it. That becomes very difficult for us to identify user needs.

Tan Lee Chin, Digital Product Editor, SPH Media

The newsroom is now evaluating whether articles should be restricted to a single primary user need or if a strategy should be adopted where a dominant need is assigned for analytical purposes while secondary needs are permitted editorially.

Implementing a Shared Newsroom Language

To make the user needs framework functional for daily operations, SPH Media is working to translate the framework into Chinese and align it with the specific brand identity of Lianhe Zaobao. The goal is to provide newsroom leaders and editors with a standardized language they can use when assigning stories to journalists.

The newsroom plans to hold sessions later in May 2026 to review past coverage and determine which editorial approaches best reflect the brand. By establishing a shared internal model rather than adopting a standard external framework, the organization aims to make collaboration across different sections more consistent.

Phased Rollout and Testing

The implementation of the new framework will follow a gradual, three-step process:

• Defining a specific model tailored to Lianhe Zaobao.
• Piloting the framework within two selected news sections that have been identified as currently underserving audience needs.
• Establishing specific metrics to evaluate the results of these trials.

The newsroom intends to track engagement-focused indicators to verify the effectiveness of the framework. Tan Lee Chin emphasized that the rollout is being handled cautiously, noting that the true challenges of the transition will only become apparent once the framework moves from theoretical discussion into day-to-day practice.