Focused on India’s Hindi-language market, Amar Ujala publishes around 2,500 stories each day on its website, amarujala.com.
In recent years, the publisher’s major challenge has been to balance and direct that wide range of content with the wants and needs of the millions of users coming to their website.
“Every user has different schedules, different expectations for a product or a brand. We were facing this challenge, and engagement was declining,” Shekhawat told participants at WAN-IFRA’s AI Summit in Bengaluru in July.
Building a meaningful bond with users through personalisation
Amar Ujala’s plan was to strengthen personalisation to build engagement, and this is a key reason they have introduced AI across departments.
“You cannot have a meaningful bond with your user until your product is fully personalised for the end user. It has to be one-to-one personalisation,” Shekhawat said.
The difficulty is that not only is every user different in their wants, but their needs and expectations change with time as well, he said. “Their expectations are different in January and in the middle of the year. It becomes almost impossible to engage them and keep them coming back to the platform.”
To address this challenge, Amar Ujala began experimenting with AI and devised a few central goals to help them see whether they were headed in the right direction. The first one, Shekhawat said, was that they wanted to increase their paid digital subscribers.
Growing paid subscribers
“We have a very small number of paid subscribers, but we want to grow them and nurture them. Subscription is more of a retention business rather than acquisition. Our data shows if you are able to engage users more, make them come back again and again, in some way or another they will stick with you,” Shekhawat said.
A major issue for Amar Ujala was addressing casual users, many of whom might be coming to the site for the first time.
“If someone is coming regularly, then we know their persona, but someone who is coming for the first time, we don’t know anything about them,” he said.
In these cases, the publisher aims to build a profile through a variety of other information, such as getting the user’s location through the onboarding journey, or if that is not available then they look at what devices and browsers the person is using, and try to capture these granular data points and add them to their system, Shekhawat noted.
If Amar Ujala can succeed in engaging users when they are coming to their site for the first time, “the probability of them coming back again is strengthened,” he said.
They have also divided their users into segments, which then trigger several action plans.
“We have that capability now where we can actually understand that OK, with this set of users, we just show them the ads. This set of users we show a paywall. This set of users we can give a data wall, because they will give us first-party data,” Shekhawat said.
Determining ‘shelf-life’ for content
Amar Ujala also tries to find out users’ short-term and long-term interests, through different filters. Some filters are designed to remove content that is not trending, for instance.
Making adjustments for the “shelf-life” of a particular piece of content is also important. Shekhawat pointed to sporting events and the mix of different kinds of content that can go along with them.
For example, he said, “when a match is going on, there will be match commentary, but once the match is over the value of that particular piece of content is over. But the match analysis, that will last for two days.”
The shelf-life for a news article can, likewise, be very important. To help determine this time frame, Amar Ujala gave its journalists a way to suggest a timeframe for their articles within the internal CMS.
He noted that they started with the homepage and moved on to other areas, such as their navigation, which is now personalised as well.
So far, they are seeing positive results. “If we take one month’s data, 50 percent of users are returning,” he said.
Paying digital subscriber figures have also doubled, though he said they are still fairly small, but the results have been encouraging, and they are working to keep those numbers growing.
“On the whole, subscription is very important for us. This particular part of the funnel is very, very small right now, but in the long run, we will have to work hard to increase it,” he said.
“We have with everything in place now for promotional and direct campaigns, so it can help increase the revenues. We have been working very closely with our data and marketing teams,” he added.
Next up: Expanding video efforts
While some smaller news publishers use AI to boost their content volume by generating articles on topics like local sports or real estate, Amar Ujala isn’t interested in using AI in that manner.
“We don’t want to create more content; we want to free up our team’s time to focus on leveraging the content they are already doing,” Shekhawat said.
Every story has a number of possibilities for leveraging the content, he said. For example, “Can we quickly create a video and add to it? Can we use some AI tool to add some maps to it? Our authors want to do that, but they don’t have time. Time is a big currency that AI can help us with.”
Video will be a key area moving forward.
“We are working very aggressively on videos. It is one of the areas where the personalisation engine is not plugged in,” Shekhawat said.