AI strategy · Reuters · 2025

From search to conversation

I found Reuters was the most-cited news source on ChatGPT, and turned that into a distribution strategy and a paywall that works wherever readers are.

60%: citation share on ChatGPT
3×: conversion vs. Google search referrals

Context

Every publisher was watching search traffic fall and debating whether to block AI crawlers. Nobody had good data on what AI referrals were actually worth.

The problem

Search referrals were collapsing across the industry. The question was where readers were going instead.

What the evidence said

A scraping and analysis framework showed Reuters was the #1 cited news organisation on ChatGPT, and that ChatGPT-referred readers converted at 3× Google search.

Options on the table

Rejected
Block AI crawlers and wait for licensing deals
Protects content in the short term, but gives up the channel with the best-converting readers.
Rejected
Optimise for AI citations only
Grows visibility without capturing value; citations don't pay if the reader never meets a paywall.
Chosen
Distributed paywall: convert readers wherever the content appears
Keeps the upside of being cited and builds a subscription path into third-party surfaces.

What shipped

Scraping and analysis framework measuring citation share across AI assistants
Subscriptions on Rails, a distributed paywall for third-party surfaces
Conversion paths in the Reuters ChatGPT app and Reuters MCP servers

Outcome

The finding reset Reuters' AI distribution strategy. Subscriptions on Rails converts readers across partner publishers, the Reuters ChatGPT app and Reuters MCP servers.

My role, and the team's

Me: Built the measurement framework, found the result, wrote the strategy and designed Subscriptions on Rails.
The team: Partnerships ran publisher conversations; engineering built the paywall service; legal shaped the terms.

From the working file

Strategy one-pagerAI assistants are a channel

Observation: Search referrals falling; AI-assistant referrals small but converting at 3× search.
Bet: Readers who arrive via an assistant are high-intent. Meet them with a paywall, not a wall.
Not doing: Blanket crawler blocking; chasing citation share without a conversion path.
First milestone: Paywall live in one partner surface; conversion measured against search baseline.

What I'd do differently

The analysis sat as a slide for too long. I'd publish an internal dashboard on day one so the number could argue for itself.