How AI can help news publishers

Insights

• News houses are scrambling to stay afloat amid economic uncertainty and declining subscription revenues.
• They must use AI to retain high-value customers and optimize growth. Good AI predicts what content and pricing model will ensure long-term customer retention.
• Firms must organize their data estate, prioritize high impact AI initiatives, and ensure executive support for AI transformation.
• A build-versus-buy exercise is crucial. For firms lacking data science expertise, an off-the-shelf solution such as Infosys Publisher AI Cloud can be a game changer.
• This strategy will help firms tackle news avoidance and convince people in real time that their news is worth paying attention to and worth paying for.

Amid inflation worries, changing user preferences, and a growing tide of negative news, media firms are in a tricky situation. Many see significant valuation drops, with seven of the top 10 media companies in the US underperforming the S&P 500 index by 41% from September 2021 to September 2023.

BuzzFeed News and Vice Media, once digital trailblazers, both closed their doors in 2023. Similarly, the Wall Street Journal’s axing of 40 economist staffers from its Washington bureau in February 2024 shows how even bigger players are scrambling to increase margins in the face of economic uncertainty and declining subscription revenues.

As one analyst and manager in Germany put it: “I think one of the truths about the media industry is that it is an industry that is under a certain obvious strain for cash, for new business models, figuring out what their future is. Basically, this ‘What’s going to save us?’ question is all out there.”

The problem with journalism

Video streaming and news delivery on mobile apps and affiliate websites are primary revenue streams for many media organizations.

But many people are on the internet less, down 13% since the pandemic, according to GWI study. Further, more customers are avoiding divisive or depressing news stories, according to the 2022 Digital News Report by the Reuters Institute. “Turning my back on news is the only way I feel I can cope sometimes,” said one UK reader in Reuters’ follow-up report in 2023, which surfaced new concerns such as disruption by online platforms such as Meta and Google, misinformation and increasingly low levels of trust in news coverage.

The 2022 survey also asked people in leading economies how interested they are in news in general. In the UK, the proportion saying “extremely/very” has fallen from 70% to 43% since 2015. Also, many lack funds to pay for content that leaves them overwhelmed and confused.

Pew Research Center reveals an exodus of traditional news subscribers to social media channels such as TikTok and YouTube, hitting their sales and marketing revenues.

Digital news houses must retain customers and find new revenue opportunities to stay in the game, and they need to do it fast. One way forward is using artificial intelligence (AI) to work out how to make journalists more productive and business processes more efficient. On the customer side, they must decipher what’s working and what’s not, and choose between short- and long-term content strategies to retain high-value customers and optimize growth.

The only way forward?

AI might not always be a prerequisite for content strategy. News publishers can get a head start through wider content diversity across channels, more reliable reporting, and more positive content — all things that increasingly perceptive and money-tight customers say they are looking for.

But if your competitors are doing it — Netflix and Amazon are competitors, even though their margins come from video streaming — it’s foolish to take a wait-and-see approach to AI.

To cite a media manager in the US who was interviewed as part of Columbia Journalism Review’s, TOW Report: Artificial Intelligence in the News: How AI Retools, Rationalizes, and Reshapes Journalism and the Public Arena: “How can I use these AI technologies to increase my audience, to increase my subscriber base, to increase the time that people are spending on the page and scrolling and viewing my gorgeous ads that are alongside it? That’s a motivation for us.”

AI algorithms can train on disparate data across devices and channels, and forecast future demand for different kinds and forms of content. Firms gain insights into what people are likely to consume in the future (channel, format, time of day) based on current interests and viewing patterns, which can nudge readers and viewers on their news journey across affiliate sites, unlock paid content, and inform editors on the direction their brands should take in the medium and long term.

For instance, a publisher uses AI to personalize podcast recommendations based on users’ listening patterns. AI in this case works as natural language processing (NLP) and machine learning (ML) tools to extract and analyze metadata from existing podcasts, which are then combined with user data. “This AI is able to do something at a scale that would be possible otherwise, at a level of detail that also wouldn’t be possible,” one manager told Felix M. Simon, the report’s author, interviewed as part of the TOW Report. Here, AI improved audience experience and boosted user retention at scale.

Numerous other AI applications in journalism extend beyond the scope of this paper. Figure 1 gives a quick summary, with those use cases suggested in this paper highlighted in bold.

Figure 1. AI use cases in the journalism industry

Source: Adapted from TOW Report: Artificial Intelligence in the News: How AI Retools, Rationalizes, and Reshapes Journalism and the Public Arena.

The problem with AI

Many AI use cases have already been adopted in the form of generative AI. Media companies are quietly integrating AI into their products to deliver more wholesome news experiences, according to Reuters. Nearly 28% say this is now a regular part of their activities.

However, it’s not easy for many to onboard new AI technologies into products and solutions. CXOs we speak to worry about irresponsible AI usage (trust, ethics, intellectual property issues) and an inability to scale the technology fast enough to make a significant impact. There is also the concern that many smaller media houses just don’t have enough clean data to train AI models and they are short on executive and data science skills to create AI products from scratch. Building fine-tuned generative or predictive AI models can cost millions, which struggling news providers can’t afford. Further, some firms prioritize a big bang initiative, don’t see results, and dial back to simple automation solutions, saving money where they can.

AI-first requires data-first

However, many media firms are sitting on hoards of data, which, if prepared properly, can make both build-from-scratch and new off-the-shelf AI solutions work for them.

With executive sponsorship and prioritization of high-impact use cases, news organizations can start their AI journey by preparing their data, both structured and unstructured, across various formats (machine-generated, social media, historical, etc.). Data preparation ensures that data assets are available, discoverable, accessible, and of high quality for use in ML and generative AI models.

Once all data is available at one place for AI, publishers can start connecting disparate trend data, including likes, social media sentiment, page views, bounce rates, and trending topics across channels by demographics.

This knowledge graph offers organizations a comprehensive view of their business ecosystem. It enables them to craft tailored, memorable experiences for users in optimal formats and timing.

It also provides curated product bundles to subscribers, increasing value for money and significantly reducing customer churn.

Schibsted in Norway uses data analytics to refine its subscription strategy; it now offers six national and local newspapers, 44 magazines, and exclusive podcasts in an all-access package, just a bit more than a single publication subscription cost.

Infosys Publisher AI Cloud

Media houses must rehearse a build-versus-buy exercise. With ample data scientists, data engineers, and readily available data, building such AI solutions is viable and feasible. However, off-the-shelf solutions are also available, which, with a bit of fine-tuning, can learn to inspire, engage, enroll, contextualize, and connect different customer groups on their news journey.

Infosys Publisher AI Cloud provides data management, ad targeting, and insights on customer preference across channels, along with reach and frequency measurement, engagement programs, and a connected ecosystem view that increases the potential for data monetization (Figure 2).

Figure 2. Workflow of the Infosys Publisher AI Cloud platform

Source: Infosys

Both open and closed generative AI models, accessible through Infosys Topaz, enable content curation and creation.

The solution targets three main outcomes.

• Subscriber intelligence: First-party data creates a dynamic paywall that unlocks certain articles to certain users based on willingness to pay, giving the publisher a way of offering pay-per-article or sending out a subscription request. In the TOW report, interviewees reported dynamic paywall conversion rates 2% to 10% higher than random policies. As Rohit Supekar, a data scientist at the New York Times, has described: “The Times achieved its goal of 10 million subscriptions and set a new target of 15 million subscribers by the end of 2027. This success has been possible in part due to continuous improvements in the paywall strategy over the years.”
• Knowing customer interests: Publishers have many internal and affiliate sites. The more the news organization knows about users and their preferences, the easier it is to navigate them from site to site.
• Theme-based ad targeting: First-party contextual ad targeting allows advertisers to target based on topics and themes derived from the recommendation engine.

Not long ago, a financial news publication rapidly lost subscribers due to manual processes and weak customer insights, with no indication about what was being viewed and which customers might be thinking of leaving. The firm used the Infosys solution for a technique known as “personalized save,” a predictive model based on engagement data and a complete picture of each customers’ user profile. The firm personalized its cross-sell and up-sell marketing offers, increasing customer retention by 20%.

Any business strategy is now an AI strategy

Let’s not overlook that AI serves as both an efficiency and productivity enhancer; transcription, summarization, and other content intelligence and process optimization tools rank high among AI use cases in journalism. These tools enable producers to do more work faster.

Indeed, AI, and generative AI in particular, has a big part to play in the future of quality news curation and sales and marketing analytics. If publishers can prepare their data estate and leverage off-the-shelf solutions that provide deep customer insights, the pay-off will be a more confident and consumer-focused content strategy, working across all channels to retain customers and grow business: the upshot is a healthier organization that grows and adapts to changes in the market — able to pivot in case times get even worse.

Media firms will first have to tackle periodic and specific news avoidance, while tailoring content to different demographics and reach them on the right channel. They must enhance news accessibility for hard-to-reach groups, work out what inspiring and positive news stories to commission, and convince people in real time that their news is worth paying attention to and worth paying for. AI is a powerful tool to help publishers achieve this.