The Data: Why Earned Media is the #1 Driver of AI Search Visibility
Introduction
The data surrounding Generative Engine Optimisation (GEO) suggests that earned media is the single most dominant driver of AI Search Visibility.
If you want tools like ChatGPT, Claude, and Google AI Overviews to recommend your company, you need to understand exactly what types of content these algorithms trust.
What the Research Says
Global academic and industry studies paint an incredibly clear picture of how LLMs select data:
-
The Dominance of Earned Content: A landmark study from the University of Toronto found that AI search results were 77.6% Earned, 22.1% Brand (Owned), and a mere 0.3% Social.
-
The Software Sector Reality: When looking specifically at the software category, AI search cited 74.2% Earned and 25.8% Brand—with social results dropping down to absolute zero.
-
The Trust Factor: Observations from Meltwater reinforce this data, stating that traditional press coverage, expert commentary, analyst reports, and reputable niche publications provide the strongest signals to LLMs, while social media mentions serve only as secondary indicators.
Redefining “Earned Media” for the AI Era
To leverage this data, PR and marketing professionals need a slight shift in mindset. For an LLM, “earned media” doesn’t strictly mean a feature in a mainstream newspaper. It encompasses any credible, high-authority, third-party endorsement.
According to data from Blinq (based on a large-scale study of 70,000 LLM responses), review sites and user-generated content account for nearly a quarter (23.3%) of AI sources. Depending on your industry, an LLM might pull heavily from:
-
High-domain authority outlets (e.g., Reuters, Financial Times, Forbes, Axios).
-
Dedicated vertical industry and niche publications.
-
Independent product review platforms, corporate websites, and financial blogs.
-
Community spaces and reference sites like Wikipedia.
The Golden Rules of AI Search Visibility
To ensure your brand gets cited, you cannot view communication as a series of isolated campaigns. AI models show a strong preference for recency—with studies from Muck Rack showing a high citation rate for stories published within the past 12 months, spiking heavily for news within the past 7 days.
Tessa Bron from Progress Communications (our member agency for Benelux) suggests because LLMs retain historical training data, past reputational issues can be resurfaced in a split second. Consistency and coherence over time are our only safeguards.
Download our complete PR Manager’s Guide to GEO (no email required).
