Digiday Publishing Summit: The Subscription Shift Is a Two-to-Three Year Project with No Quick Wins
At the Digiday Publishing Summit in Vail, Colorado (March 23–25), media executives described the transition to direct reader revenue as a 'two to three year project' with 'no quick wins.' Jon Roberts, Chief Innovation Officer at People Inc., keynoted on the 'revenue lag' in AI licensing deals with OpenAI, Meta, and Microsoft, while noting that People Inc. blocks 'tens of millions' of unauthorized AI bot crawl attempts daily. Medium-sized publishers described competing for limited consumer wallet share as 'eating each other.' Subscription prices rose 5% year-over-year in 2025. Social media channels with 100,000+ subscribers convert approximately 1,000 to paid members. 82% of journalists now use AI tools (Muck Rack data).
Analysis
The Digiday Publishing Summit has become the most reliable annual barometer of where the digital publishing industry's anxiety is concentrated — and the March 2026 edition is the most anxious in years.
The "two to three year project" framing for the subscription transition is honest in a way that most public statements from publishers are not. It acknowledges that the shift from ad-supported, traffic-driven revenue to direct reader revenue is not a product launch or a strategy pivot — it is a cultural transformation that requires retraining newsrooms, rebuilding content strategies, and rebuilding reader relationships from scratch. The instincts that made digital publishers successful in the SEO era — optimising for search intent, maximising page views, writing for the algorithm — are not just unhelpful in the subscription era; they are actively counterproductive. A reader who arrives via a Google search and bounces after reading one article is a success metric in the ad model and a failure in the subscription model.
The AI licensing revenue discussion is the most commercially interesting thread from the summit. Jon Roberts' description of a "revenue lag" in AI deals with OpenAI, Meta, and Microsoft reflects a structural problem: the deals are being signed, but the revenue is not yet flowing at the scale that would replace lost search traffic. This is partly a timing issue — many deals were signed in 2024 and 2025 and are still in their early phases — and partly a valuation issue. Publishers who signed early deals may have underpriced their content relative to what AI companies were willing to pay, and renegotiation is difficult once a deal is in place.
The 82% journalist AI adoption figure from Muck Rack is the data point that deserves the most attention. If more than four in five journalists are already using AI tools in their workflow, the question is no longer whether AI will change how journalism is produced, but whether the resulting content will be differentiated enough from AI-generated content to justify a subscription price. The answer, for the publishers at the Digiday summit, is that differentiation must come from relationships, access, and community — things that AI cannot replicate — rather than from the text itself.