Academic Research Got Cited Less Than Commercial Content. Read That Again.

We analyzed 5,472 citations generated by ChatGPT, Claude, Gemini, and Perplexity in response to health-related prompts built to mimic real research behavior. Academic and peer-reviewed sources got cited less often than commercial content.

Read that again.

Health media was the single largest source category, at nearly one in three citations. Commercial and affiliate-driven content came second at 23%. Academic and research sources, despite their authority, accounted for less than a quarter. And 40% of everything cited was published in 2025, with citations dropping off sharply for older material.

Now add the demand side. Forrester found 89% of B2B buyers use generative AI across every stage of the purchase process. Rock Health's 2025 consumer survey found nearly a third of respondents use AI chatbots for health information, up 16% from 2024. These are not experimenters at the margins. They are procurement committees and clinical operations leads triaging a crowded vendor landscape, and AI is the first filter they reach for.

Picture a VP of Clinical Operations evaluating care coordination platforms. Before she answers a cold email or attends a webinar, she asks a chatbot what the leading platforms are and what to look for. In seconds she has vendors, differentiators, and evaluation criteria. Your SDR did not set that shortlist. Your campaign did not shape it.

AI does not reward authority. It rewards accessibility. A well-structured article on a credible third-party platform outperforms a dense white paper living only on your domain. That is not a content quality judgment, it is a machine-comprehensibility one.

Recency is now a discipline, not a nice-to-have. If an asset has not been touched in twelve months, it is likely invisible to the systems your buyers are using today.

If your brand does not appear in AI answers yet: do these three things this week

1. Run the visibility audit. Open two or three chatbots. Type the category question your buyer would type. Note whether you appear, how you are described, who appears next to you, and what evaluation criteria the model surfaces. Twenty minutes, no budget.
2. Count your owned versus external ratio. Audit six months of output. If more than 70% published on your own domain, you have a distribution problem, not a content problem.
3. Pick two external channels and commit to one substantive piece a month. Trade publications, association sites, indexed podcast platforms, healthcare business media. That alone moves your citation surface over two quarters.

If you already appear but the description is wrong: do these three instead

1. Build a gap map from buyer questions, not a content calendar. Integration requirements. Implementation timeline. Expected outcomes in the first twelve months. If you cannot find a direct answer in your library, that is the roadmap.
2. Refresh your five to seven highest-value assets this quarter. New data, a current case reference, a revised stat. Republish with a visible date. Redistribute externally.
3. Track the competitive set the AI names, not the one in your CRM. They are often different, and the AI set is the one shaping the shortlist.

The reframe

GEO is not SEO with a new acronym. Search ranking decides which link a buyer clicks. AI citation decides whether you are described at all.

The pipeline number is a lagging indicator. Whether AI systems can find and cite your brand is a leading question.

If you do not appear in the research phase, you are not losing at the bottom of the funnel. You never entered it.

And if you don't want to do it alone, let us know.

‍