On Press Gazette's News Yacht, Publishers Prepare for a Two-Web Future
On the Equativ boat at Cannes Lions, media leaders traded festival futurism for the practical work of building for an agentic web that already shapes traffic, citations and trust.
On the Equativ boat at Cannes, executives from Hearst, TIME, The Guardian, Axios and The Washington Post set out how publishers are rebuilding for an agentic web that already shapes traffic, citations and trust.
Dominic Ponsford, Editor in Chief at Press Gazette, noted early in the session that there is no news beach at the Cannes Lions Festival. The observation is not new, but at Press Gazette’s News Yacht session, held on the first day of the festival and co-sponsored by Q5 and PA Media, the absence felt more pointed than usual. The discussion that followed skipped the festival’s usual futurism and went straight to the infrastructure publishers are already building for an AI-mediated web.
Katie Vanneck-Smith, Chief Executive of Hearst UK, was blunt about what publishers stand to lose. AI content needs to be labelled the way food is labelled: consumers should be able to see the composition of what they are reading. The publishers that survive will be the ones that have worked out what makes them distinctively human rather than the ones that have simply achieved the fastest AI output rate. The relationship a reader has with a trusted brand (’I’m a Times reader, and read Red Magazine, that’s part of who I am’) is not something a content machine replicates, and treating it as a given is precisely how publishers lose readers.
The agentic web is not coming. It is already here.
Mark Howard, Chief Operating Officer at TIME, started by walking through what two years of aggressive infrastructure investment looks like in practice. TIME has four LLM partnerships, runs Tollbit and Scalepost through its Fastly CDN, and joined Cloudflare’s Independence Day protocol last July, blocking unknown bots by default and shifting to an allow list. Unrecognised crawlers are routed to a paywall. Unsurprisingly, none currently pay.
The more consequential development is what TIME calls the markdown page: a stripped version of its content, metadata-rich, served specifically to bots. The company is now appending additional data to these pages, described variously as content enrichment or context forwarding, and building material explicitly designed for LLM consumption that no human will read.
Howard also described a product built around the gap between brand self-presentation and LLM reality. TIME queries the four major AI answer platforms and compares what they say about a brand against what that brand says about itself. The delta is typically 15 to 40%. TIME is now building branded content programmes to close that gap, routing content through its markdown pages where its domain authority with LLM bots is already high, and extending the approach to YouTube for long-tail discoverability.
“There are now two customers,” Howard said. The human web: pages designed to load fast, display well, monetise through advertising. And the agentic web, where bots are no longer simply bad traffic but are actively informing purchase decisions, and in some cases making them directly.
Publishers account for 84% of AI citations. Most have not factored that in.
Isabel Perry, Global EVP of Strategy at DEPT, pushed back against the standard pessimism. While Reddit is frequently cited as dominating AI responses, publishers themselves capture 84% and consistently appear within the top three channels by category. DEPT clients are reporting 60 to 70% zero-click rates from Google search; that part is real. But publisher authority inside LLM outputs is, by the data, growing.
Perry called the strategic response generative engine optimisation: building author credibility, brand consistency, and positive sentiment systematically across all channels, because language models look for patterns and authority signals rather than ranking. Jacquelyn Cameron, Chief Revenue Officer at Axios, gave the most commercially concrete version of that argument. Axios writes in Smart Brevity: what’s new and why it matters, placed up top. That structure naturally aligns with how LLMs scrape content. The top 30% of any article is what gets retrieved. Axios is building on that commercially with a brand library product that lets companies use Axios as a vessel to position authoritative information in front of the models that shape purchase decisions before a buyer sends a single signal back to a brand.
Trust as asset,
trust as problem
Imogen Fox, Global Chief Advertising Officer at The Guardian, anchored the trust discussion in data rather than aspiration. The Reuters Institute Digital News Report, cited in the room, confirms that trust in news brands remains low. The Guardian is partly insulated: 1.4 million paying supporters across 180 countries, with traffic growing including in the US. But Fox was clear this is a brand dividend that not every publisher can draw on, and that the platforms still refuse to share the granular AI referral data that would let publishers properly measure what they are losing.
The Guardian’s structural response is SPUR, an initiative establishing standards for responsible licensing, alongside a firm editorial commitment to human-centred journalism.
Compass, not GPS
Karl Wells, Chief Revenue Officer at The Washington Post, described a different internal challenge: how to give journalists useful data without turning data into instruction. The Post built an audience behaviour score, a single unified metric combining monetisation drivers and engagement signals, to create common language across newsroom and commercial teams. Journalists track it and compete on it. It simplifies without dictating.
The framing that emerged from his conversations with Executive Editor Matt Murray was compass rather than GPS. Data provides direction and context. It does not provide routing.
Subscriptions ask too much of new audiences
Wells was direct about the Gen Z problem. Asking a younger reader to subscribe before they have formed any relationship with a publication is structurally flawed: the commitment implied has no equivalent in how that generation accesses anything else. More flexible entry points are not a concession to lower standards. They are the only commercially rational path to building audiences that will eventually convert.
Nichification is part of the answer. “We’re no longer talking about how everyone consumes news,” Cameron said. “It’s about how everybody forms their reality.” Axios is built around that insight. Future operates across more than 175 specialist brands for the same reason: depth of relevance earns the relationship that precedes the transaction.
The creator economics question has no clean answer
Hannah Blake, Managing Director of New Media at Mail Metro Media, was candid about creator retention. Journalists take salaries. Creators carry audiences that may outlast any employment relationship, and flat payment structures don’t reflect that value. Mail Metro is experimenting with commission models and joint ventures, Goal Hanger cited as one reference point, where talent shares in revenue rather than simply receiving a fee.
“We haven’t got the full answer yet,” Blake said.
The session’s closing observation came from the same direction. When AI content is sufficiently pervasive that human provenance becomes a premium, physical events and in-person activations stop being margin items and start looking like strategic assets. People will seek human interaction in proportion to how synthetic their digital environment becomes. Whether that translates into sustainable commercial strategy before the window closes is the question nobody at the session could answer.