For twenty years, a public relations team measured a campaign by the clips it produced: how many outlets ran the story, how large their audiences were, whether the coverage skewed positive. That scoreboard still exists, but a second one now sits beside it, and it increasingly carries more weight. When someone asks ChatGPT, Gemini, or Perplexity about a company, the assistant does not return a stack of articles to sort through. It composes an answer and names a handful of sources. Whether a brand’s earned coverage sits among those cited sources has become one of the highest-stakes questions in modern communications.
The encouraging part is that this shift rewards the craft PR has always practiced, now aimed at a new reader. The campaign that earns trusted, relevant coverage is the same campaign an answer engine learns to cite. Getting there takes a deliberate change in how a program is planned, pitched, and measured.
What LLM Discoverability Means for PR
LLM discoverability is the degree to which AI answer engines retrieve, trust, and cite a brand, or the coverage written about it, when they respond to a user’s question. It is earned through authority, clean structure, and freshness rather than bought through ad spend, which places it squarely in the domain of earned media rather than paid.
The mechanism behind it explains why PR sits at the center. When a model answers a live question, it runs a retrieval step, pulling a small set of documents from the web before it writes anything, then composes a response from what it found. That architecture, retrieval-augmented generation, means a brand that is not retrieved cannot be cited. The sources that get pulled are weighted heavily toward reputable, third-party material, which is exactly what a strong earned-media program creates. A brand’s own marketing pages rarely win that competition on their own.
The scarcity of slots raises the stakes further. Where a Google results page offers ten links for a user to weigh, an AI answer typically references only a few domains. According to the 2026 Status Labs white paper on AI and reputation, models tend to cite between two and seven sources per response. Being left out of that short set means effectively not existing for the question, with no second page to be discovered on later.
Why Earned Media Became the Strongest Lever
The clearest evidence for prioritizing earned media comes from independent research. A 2025 study by University of Toronto researchers ran large-scale, controlled experiments comparing how AI search engines choose sources against how Google does. They found that AI systems show “a systematic and overwhelming bias toward earned media,” meaning third-party, authoritative sources, over brand-owned and social content. Google distributes its citations more evenly across owned and earned material; AI answers do not.
That finding rewrites the value of a placement. Coverage in a credible, on-topic outlet does more than reach that outlet’s readers. It becomes a signal an answer engine uses to decide who the authority on a subject is. The controlled work on optimizing for these systems points the same way. Princeton-led research presented at KDD 2024 tested nine techniques across thousands of queries and found that citing authoritative sources and adding verifiable specifics lifted a source’s visibility in AI answers by as much as 40 percent, with the largest gains going to lower-ranked pages that added credible citations. Third-party validation, the core product of PR, is the most direct route into a cited answer.
The audience on the other side of that answer is now enormous. ChatGPT alone reached 900 million weekly active users by early 2026, up from 800 million a few months earlier, and that is one platform among several. A growing share of the people forming a first impression of a brand are reading a synthesized answer, not a press clip, which makes the question of what those answers say a board-level concern rather than a communications footnote.
The New Measure of a Campaign
Traditional PR measurement counts reach and sentiment across coverage. A program aligned to LLM discoverability adds a different question: is the earned coverage being cited by AI engines, and do those citations describe the brand accurately? The two scoreboards overlap, since good coverage still serves both, but the second one changes what counts as success.
The practical consequence is that a campaign stops expiring with the news cycle. A well-placed article that a model begins citing keeps working long after the initial traffic spike fades, because it has entered the pool of sources the engine trusts for that topic. Measurement has to follow the value. Tracking how often each engine cites a brand’s coverage, whether the citation is accurate, and how the mix of cited sources shifts over time becomes the real key performance indicator, alongside the familiar reach and sentiment metrics.
Building a Campaign That Gets Cited
Aligning a PR program with how models choose sources comes down to three moves that map onto muscles most teams already have.
The first is targeting the questions. Models retrieve around the language of a user’s prompt, so a campaign has to be built on the real questions people ask about a brand and its category rather than on internal messaging pillars. Auditing what the major engines already say, and which outlets they already cite, shows which publications shape the AI narrative for a category before a single pitch goes out.
The second is earning the right coverage. Authority and topical relevance beat raw circulation here, because models weigh the credibility and subject fit of the citing domain. A handful of trusted, on-topic articles in publications an engine already relies on will usually outperform a wide spray of low-authority pickups.
The third is structuring the story for extraction. The units a model lifts into an answer are clean, quotable facts, so every placement should carry one: a dated statistic, a crisp definition, or a named-expert quote that stands on its own when pulled out of context. A release written around a single strong, quotable fact gives journalists the extractable material that models later cite.
Run together, and reinforced with answer-first content on the brand’s own domain that echoes the same facts and terms, these moves compound. The engine finds the same story corroborated across earned and owned surfaces, which is the pattern it reads as settled.
What Undermines the Effort
The fastest way to waste a discoverability campaign is to try to buy the outcome. Answer engines weigh editorial, earned sources far above paid placements, so a program leaning on advertorial or sponsored content tends to underperform in citations even when it looks busy in a coverage report. The coverage has to be earned to carry weight.
Manufactured volume fails for the same reason. Padding the web with thin, near-identical mentions produces no corroboration a model can trust and can dilute the strong, citable coverage a brand actually wants surfaced. Depth and credibility move AI visibility. Quantity on its own does not.
How Status Labs Runs PR for AI Visibility
Status Labs rebuilt its earned-media approach around this shift several years ago, and it now plans campaigns with AI answer engines as a core audience from the outset. The firm’s analysis of the discipline frames the goal directly: earn the coverage AI engines want to cite, and make that coverage easy to extract.
In practice, the work runs as a sequence. It starts with an AI baseline audit, querying the major engines on the questions that matter for a brand and logging which outlets and articles each one cites. That map guides where to pitch. Messaging is then translated into the language of real user prompts, coverage is concentrated in authoritative and topically relevant outlets, and each placement is engineered around a single extractable asset a model can lift cleanly. Owned content reinforces the campaign with answer-first pages that echo its facts, and machine-readable structure, consistent entity naming and organization markup, helps crawlers parse the coverage as citation-ready. The final step is measurement against citations rather than clips alone, tracked query by query as the source mix moves.
Grounding that method is a body of first-party research the firm publishes openly, including its 2026 white paper on AI and reputation and the field notes it shares on its YouTube channel. The through-line across all of it is that durable AI visibility is built the way credibility has always been built, by earning the recognition of sources others already trust, and then making that recognition easy for a machine to read.
So, how do you align a PR campaign with LLM discoverability? Audit what the engines cite about you today, build messaging around the questions people actually ask, concentrate on authoritative and relevant outlets, engineer a clean citable fact into every story, and reinforce it with structured owned content. Do that consistently, and a campaign stops ending with the news cycle and starts becoming part of the answer people receive about the brand.
