If you look across the sources that inform AI search results in consumer health and wellness, a handful of websites consistently underlie how consumer health answers get constructed. What’s interesting isn’t just which sites show up—it’s how differently each AI model leans on them, and what that says about how “truth” gets assembled.
Let’s start with the anchor: Healthline. Think of it as the baseline unit of influence. Other key sites influence can be framed relative to that.
Healthline: The Baseline (1.0)
Healthline is the closest thing to a universal translator across AI systems. Its influence is evenly distributed, showing up across nearly every major model rather than being concentrated in one.
What drives this dominance is breadth and format. Healthline doesn’t just do condition explainers or product roundups it does both, and does them in a way that’s structurally easy for language models to parse. Whether the query is “what is this symptom?” or “what should I take for it?”, Healthline is almost always relevant.
The result: it becomes a default reference layer. Not because any one model overweights it, but because all of them do enough to make it unavoidable.
Amazon: ~0.6 of Healthline
Amazon shows up differently, more a proxy for consumer behavior than a true health authority.
Its influence is concentrated in certain models that treat marketplace signals (sales rank, review volume, availability) as a form of credibility. In those systems, “what people buy” becomes shorthand for “what works.”
This creates an interesting dynamic: Amazon isn’t shaping the medical framing of answers, but it is shaping the recommendation layer. It’s where AI systems swap clinical reasoning for behavioral data. And while Amazon has clearly taken steps to block or limit AI crawlers those constraints seem to be applied to crawling the entire site as training data, AI crawlers seem to have no problem accessing individual Amazon PDPs.
WebMD: ~0.3 of Healthline
WebMD plays a more traditional role: it supports the clinical backbone of responses.
Its influence is strongest in models that lean heavily on structured medical authority. When queries are about symptoms, conditions, or standard treatments, WebMD becomes a stabilizing force—less about discovery, more about validation.
Compared to Healthline, it’s narrower in scope. It doesn’t compete as aggressively in product-oriented queries, which limits its overall footprint. But where it does show up, it carries a certain “this is what doctors say” gravitas that models seem reluctant to contradict.
Verywell Health: ~0.25 of Healthline
Verywell Health operates as the reliable understudy.
It rarely leads, but frequently appears alongside other sources, reinforcing similar conclusions. Its content sits in the middle ground—part educational, part recommendation-driven—which makes it broadly usable but less distinctive.
In practice, it functions as corroboration. When multiple sources are needed to make an answer feel complete, Verywell often fills that second slot.
Forbes Health: ~0.2 of Healthline (with a caveat)
Forbes Health is the outlier, not because of its format, but because of how unevenly it’s cited.
Its influence is heavily concentrated in a single AI family, which accounts for the vast majority of its citations. In other answer engines, it barely registers.
What drives its presence is its structure: ranked lists, product comparisons, and cleanly packaged recommendations. These are tailor-made for AI extraction, especially for queries that explicitly ask for “best” options.
The catch is that its overall influence can look larger than it really is if you don’t account for that model specific concentration.
The Bigger Pattern
What emerges from all of this is that AI systems don’t agree on what “authority” looks like—they just each have a coherent version of it.
Some prioritize editorial health content (Healthline, Verywell)
Some lean into clinical reinforcement (WebMD)
Others incorporate consumer behavior signals (Amazon)
And a few show strong preferences for structured recommendation content (Forbes-style lists)
Healthline wins not because it’s the most specialized, but because it’s the most compatible with all of these approaches at once.
And which AI engine most frequently references Forbes? Google Gemini, which cites Forbes more than 300 times more often than Anthropic or Perplexity.