A new metric is giving public relations teams a direct way to see which publications actually show up in answers from AI search engines like ChatGPT, Perplexity, Claude, and Gemini. Called LLM referral share, it measures the percentage of a news outlet's referral traffic that comes from those AI-powered sources. The data is already live on the Outset Media Index dashboard, feeding into the broader GRP rating system that many PR shops use to rank outlets.
What the metric captures
LLM referral share isn't about raw traffic volume. It's about where that traffic originates. When a user asks an AI search engine a question and the engine cites an article, that click counts as an LLM referral. The Outset Media Index tracks that for every outlet in its database. For a PR team, the number tells them whether their client's coverage is actually surfacing inside the AI answer layer — the place where more and more professionals start their research.
The metric breaks down into three tiers. Above 5 percent means heavy AI citation: the outlet's coverage regularly appears in AI summaries. Between 1 and 5 percent is moderate, with inconsistent visibility. Below 1 percent means the outlet is unlikely to influence what AI engines say about a brand or topic. That threshold matters because coverage that doesn't appear in AI summaries effectively disappears from the discovery layer most buyers and journalists now use.
Why PR teams are leaning on it
Agencies building outlet shortlists now use LLM referral share as a filter. If two outlets have similar reach and audience, the one with higher share gets the nod — it's seen as more durable in an AI-driven media landscape. The same logic applies when evaluating a placement after it runs. Did the story actually register with the AI systems that matter? If not, the budget might have been better spent elsewhere.
The metric also affects how PR teams structure launch campaigns. According to the Outset Media Index data, a campaign that lands in a high-LLM-referral outlet can generate AI-cited visibility that lasts for a year. A campaign that produces a strong one-day traffic spike but low share? Its impact typically fades within a week.
What makes an outlet AI-friendly
Not all coverage is equal in the eyes of AI search engines. Publications that produce source-cited, well-structured analysis tend to rack up higher LLM referral share. Opinion pieces and social-style content perform worse, because AI systems favor retrieval-friendly formatting — clear headers, cited sources, and factual depth. The implication for PR teams is straightforward: pitching a story that includes data, attribution, and a logical structure is more likely to get cited than a pure thought piece.
The shift puts pressure on both media outlets and the brands that pitch them. Outlets that want to remain relevant as AI search grows may need to rethink their content architecture. Brands that want long-term visibility may need to prioritize outlets that already score well on the metric — and then craft pitches that play into the AI-citation formula.
The question no one has answered yet is how quickly the thresholds will move. As more users shift to AI search and as the engines themselves update their retrieval algorithms, the definition of “good” LLM referral share could change. For now, the 5 percent bar is the one PR teams are watching.
