“AI will be to digital what digital was to traditional news media,” said  Puneet Jain, CEO – HT Digital.

“AI is simply a tool – a means to an end. Our newsroom and business objectives remain the same as they were pre-AI. Now, the focus is on using AI to accelerate these goals,” he added.

Jain spoke about harnessing data analytics and AI to build products, scale business, increase engagement and better monetisation models at WAN-IFRA’s recent Digital Media India conference. 

Data and AI across digital media verticals

At HT, the typical process of content generation is divided into three steps: 

Content:

“Almost all of these steps can be improved and automated with AI, especially with Gen AI in play,” Jain said. 

Based on early experiments, HT Digital has automated one area of the generation stage and three areas of the augmentation stage. 

• Generation: This process begins with story ideation, followed by aggregation and copy-editing. It is then assigned a Quality Read Score (automated). The team can choose to make it richer by introducing elements around data visualisation, video, audio, etc. 

“It’s only been a couple of months since we deployed the Quality Read Score, and we are already seeing a 30 percent upliftment in the stories that pass these quality checks, in terms of reach it can create, versus stories that cannot,” he noted.

• Augmentation: It is then augmented through summary, translation, and headlines (automated).
• Distribution: Finally, it is ready for distribution on the website or on traditional and social media, in platform-adaptive formats. 

“Most of this automation is in the middle layer, which has to do with locking productivity, and generally saving the writer’s time,” he added.

User experience:

HT Digital is using AI to build personalised user journeys in the subscription funnel. The team has automated the following processes for content and beyond-content areas:

• Recommendations
• User-based interventions – logins, registrations
• Contextual ads

“Publishers must be able to create rule-based linear journeys for various cohorts of users according to their needs. Typically, a best-in-class news media provider would have about 2000 cohorts. That number is lower for India,” Jain said.

HT registers about 200 million monthly users on its various websites and language platforms, and about 20 million users on a single day, as per Jain.

“With AI, we can literally create 20 million journeys for 20 million users. It has helped us move from a linear model to a non-linear model with crisscrosses, aimed at maximising the user engagement and experience in a self-learning automated fashion. We can never achieve this sort of progress in a rule-based model,” he said.

The company has registered a 25-35 percent increase in user reach, and a 20 percent increase in page views, post automation.

Effective ad monetisation:

A few ad monetisation processes the team has automated are around pricing and fulfilment. 

Leveraging an in-house “anomaly detection” algorithm, the team can receive system-generated recommendations for adjusting pricing promptly, bypassing the need to rely solely on manual alerts.

They have also automated real-time ad placement, enabling them to seamlessly integrate contextual ads based on the content of the story.

Based on these two automations, the team has already seen an improvement in pricing. 

“Generally, pricing has been under pressure with the growing supply of inventory in the digital ad media. At HT, we have been able to grow our pricing over the last year,” Jain pointed out.

Currently, the most significant commercial utilisation of AI lies in digital ad monetisation.

“The scalability of AI presents an opportunity for publishers to offer dynamic ad formats, tailored in real time for different cohorts or create cohorts in real-time,” he explained. 

“This approach enables us to deliver equal value to advertisers and direct sales, while enhancing the experience for users who view ads as less intrusive.”