As AI disrupts search, some publishers build one-on-one reader relationships via text

In May of last year, Google unveiled Search Generative Experience (SGE) – now called AI Overviews – an experimental AI-powered feature that summarizes responses to search results, as opposed to providing users with a list of relevant links. Many publishers began to worry that the feature would divert web traffic away from their sites, causing them to lose out on views and potential advertising and subscription dollars.

Then, in September, OpenAI upgraded ChatGPT for paying users, enabling the chatbot to browse the web – thereby adding more fuel to publishers’ concerns about diverted web traffic.

Some publishers have attempted to solve the problem by inking licensing deals with AI companies, giving the green light to their copyrighted content being used to train models in exchange for attribution and financial compensation. Others have elected to fight back: The New York Times, for example, has filed a lawsuit against OpenAI and Microsoft over alleged IP violations.

However, others are taking an alternative approach to audience engagement: direct relationship-building.

One leader in the space is Subtext, a marketing platform founded in 2018 that enables media companies (as well as politicians and influencers) to directly reach their core audience members via text. The idea is to establish a personal, enduring connection with the most devoted readers – one that can’t be undermined by the emergence of AI-powered tools or mediated through big tech companies.

Lots of brands that have been painstakingly trying to maintain their presence on social media have “come to the realization that they were renting the relationship with their audience, as opposed to owning a really meaningful direct line of communication,” says Subtext co-founder and CEO Mike Donoghue. “We can empower [our] clients to own that line of communication, as opposed to nebulously renting it from X or [another platform].”

It’s a reciprocal relationship, Donoghue says: readers have the chance to ask questions about subjects they care about, while the reporters have “an opportunity to ask questions of their audience that maybe they wouldn’t otherwise get to ask.”

Subtext has partnered with a fleet of major publishing companies, including Vox, Forbes, The Washington Post, The New York Post, The New Yorker and Condé Nast. As part of some of those partnerships, readers can text directly with individual journalists about their areas of coverage.

According to Iliya Rybchin, partner at management consulting firm Elixirr, the influx of publishers embracing Subtext reflects a broader desire in the industry to cut out the middlemen that have historically controlled publishers’ ability to engage with their audiences. “I don’t see this as a short-term trend but an evolution to where the role of intermediaries, like search engines and social media networks, becomes less critical to media companies,“ he says. “The move to solutions like Subtext is all about regaining control.“

Many of Subtext’s clients seem to have turned to the company in direct response to the advent of generative AI-enabled search, seeking new ways to connect with their audiences: the company’s media customer base shot up by around 40% following the debut of SGE, according to a company spokesperson.

One of these customers is The Conversation, a nonprofit news publisher, which first began working with Subtext in December.

“We are very focused on the need to reach readers directly in this era of Google [generating] its own answers instead of sending readers to news sites, and the texting service is one way we are trying to bypass Google – and other platforms that have constantly shifting algorithms,“ a spokesperson from The Conversation told The Drum. “We might have done it anyway, as a way to meet readers where they are, but declining search traffic has definitely raised the importance of texting for us.“

In Donoghue’s view, AI upheaval in the news industry is creating more pressure on journalists to become specialized subject matter experts. But it’s precisely this expertise, he says, that keeps readers engaged with human writers – and the publications they work with – in the era of generative AI. Meanwhile, he predicts that the production of less intensively journalistic articles may be handed over to machines. “I’m not so sure that there will be journalists writing [clickbaity articles] for very much longer,” Donoghue says.

In other words, Donoghue rejects the idea that we’re entering an era in which publishing companies must reject or wholeheartedly embrace AI. It’s more likely, he posits, that the technology will overtake certain aspects of the publishing business while the roles and responsibilities of human writers change accordingly – including, perhaps, by being more directly engaged with individual readers via text.

Of course, in many ways, AI is already being embraced by some publishing companies. As part of their recently announced partnerships with OpenAI, The Atlantic and Vox Media have both been granted permission to incorporate the developer’s technology into new products.

Despite the willingness of some leading media companies to embrace generative AI, it will likely be a while before the average reader feels at ease with the burgeoning paradigm of semi- or fully automated news publishing and dissemination.

A report published last month by the Reuters Institute for the Study of Journalism found that more than half of Americans surveyed (52%) and an almost equal number of Europeans surveyed (47%) are uncomfortable with news generated mostly by AI, especially when it concerns fraught subjects like politics and crime.

At a time when most publications are devoting resources to maintaining a following on social media – and a growing number are starting to experiment with generative AI – the challenge for these companies will be finding unique ways to stand out from the crowd.

One of the ways in which this can be achieved, Donoghue says, is to lean into the unique expertise of human writers. “[Media companies are] going to look back in two to five years and say, ‘This was the time where we took stock of what it is that we do well; we took stock of our assets, particularly our journalists and our domain knowledge. And we invested in our ability to own a lot of communication with our audience, as opposed to farming that out.’ It’s going to be painful, but the industry is going to be better off for it in the long run because these are enduring connections that can be owned.”

For more on the latest happenings in AI, web3 and other cutting-edge technologies, sign up for The Emerging Tech Briefing newsletter.