How India Today Uses AI to Predict Audience Engagement

India Today developed Audipulse, an AI-powered audience prediction engine, to forecast story performance and optimal publishing times. Created through the WAN-IFRA Newsroom AI Catalyst program with OpenAI, the tool achieved a 64% prediction precision rate during a 15-day pilot, outperforming a 52% baseline established by human editors, according to WAN-IFRA.

The system aims to move editorial decision-making from retrospective analysis to predictive forecasting. Bal Krishna, who leads the Fact Check team at India Today, stated that Audipulse was designed to assist editors using real data rather than assumptions.

India Today, founded in 1975, operates across television, digital, print, and video. Its digital newsrooms cover business, sports, entertainment, politics, and breaking news. The organization joined the 2025 Newsroom AI Catalyst accelerator, a partnership between WAN-IFRA and OpenAI, to support this AI initiative.

How does Audipulse predict engagement?

The engine combines draft headlines for the following day with analytics from the previous day. It processes data from Chartbeat and Google Analytics, focusing on clicks, time spent, story topics, and content formats including interactive pieces, videos, and text stories, according to WAN-IFRA.

Based on these inputs, the model recommends specific publishing times and formats. The system continuously retrains by comparing its predicted outcomes against actual performance data.

To protect sensitive data, India Today deployed the system on-premises using local GPU infrastructure. This decision followed concerns about sending Comscore and Google Analytics data to external cloud environments, according to Krishna.

Why did contextual data improve accuracy?

The pilot revealed that performance data alone was insufficient for precise predictions. The team found that adding contextual taxonomies—specifically categories like Bollywood, cricket, and elections—increased prediction precision by 11 percentage points, according to WAN-IFRA.

Krishna noted that while AI is efficient at identifying trends in large datasets, reaching definitive conclusions without AI assistance is extremely difficult.

“AI is very efficient at analysing data and identifying trends. Even if ample data on audience behaviour is collected by an organisation, it is extremely difficult to reach a definitive conclusion without the help of AI.”Bal Krishna

What are the limitations of the AI system?

Despite the precision gains, the data-driven approach struggled to capture the deeper context of specific stories. Krishna stated that the system requires continuous monitoring, additional data, and manual refinement to improve outputs, which necessitates dedicated resources.

Editorial staff initially expressed skepticism regarding the predictive recommendations. This resistance persisted until the newsroom demonstrated side-by-side results during the testing phase, according to WAN-IFRA.

To manage operational costs, the team implemented overnight batch retraining to control cloud expenses.

What happens next for the project?

Audipulse remains in the development and testing phase. India Today plans to extend the system’s capabilities to include push alerts and video thumbnails, according to WAN-IFRA.

The development team is also building an explainability layer. This feature will show editors the specific factors that influenced a particular prediction. As a next step, the organization is exploring a 30-day A/B test to further validate the system’s impact on editorial decision-making.

Krishna explained that the project seeks to determine if predictive systems can support editors in a digital environment where algorithms increasingly dictate how audiences consume news.

“In today’s digital space, where people do not actively choose the source of news, and instead become passive consumers of whatever the algorithm pushes, it is important to know what your audience expects from you to keep them loyal.”Bal Krishna