Google's paltry AI payouts, and Reuters' paywall that lifted ad revenue

Jez Walters' What's New in Publishing round-up lands two signals that point in opposite directions. On one side, Google's AI Contribution programme, which pays when publisher content "significantly" contributes to AI answers, is paying out sums that look trivial: many payouts under US$1,000, US$60,000 for one tester and US$1 million annually for another, which for several small and mid-size publishers is "less than 0.1% of their advertising revenue". On the other, Reuters' machine-learning paywall is doing something counterintuitive, boosting subscriptions and ad revenue at the same time by deciding per-user whether to show a paywall based on the ad revenue that user can generate: "Overnight, the dollars started to flow." The same edition carries the harder underlying number: across Chartbeat's network, Google Search traffic to publishers plunged 40% between July 2025 and July 2026, with Discover down 34% and new readers down 14%, even as loyal reader counts rose 11%. The B2B information economy is collapsing alongside: Gartner's stock is down 71% from its late-2024 high and Forrester made a full-year net loss of US$120 million, with only 13% of B2B buyers using an analyst report in their most recent purchase decision, down 63% since 2022. Sundial CEO Kirk McDonald, formerly of GroupM North America, sums up the pivot: "Human creativity is becoming more valuable, not less."

The crux

the AI platforms' opening compensation offer is a rounding error against the traffic publishers are losing, which is why the durable value is shifting to owned audience and differentiated content.

Sell-side read

the Reuters paywall is the actionable model here, a pricing engine that treats each reader's ad value as a variable rather than treating the paywall as a blunt on/off switch.

Watch

whether Google's AI Contribution payouts scale beyond the pilot, or whether the under-0.1%-of-revenue framing hardens into the permanent economics.