AI visibility for supplement and wellness brands.
This is the category Abovance was built in. We run a supplement brand of our own, and every method on this page was tested there first. It is also the category where the AI assistants are least likely to cite a brand's own site, so the work is specific.
Who do AI assistants cite when your buyers ask?
We asked ChatGPT search and Perplexity 50 common health and supplement questions on 5 October 2026 and logged every source: 537 citations from 241 sites. Health publishers and medical, academic and government sites took most of them. Brand and company sites took 11%.
Source: Abovance, Who ChatGPT and Perplexity cite on 50 health and supplement questions, October 2026, full dataset published. Claude was added to the engine set in later studies.
| Site | Citations | What it is |
|---|---|---|
| healthline.com | 46 | Health publisher |
| pmc.ncbi.nlm.nih.gov | 23 | Academic, open-access journals |
| health.clevelandclinic.org | 19 | Health system publisher |
| health.harvard.edu | 16 | Academic health publisher |
| webmd.com | 12 | Health publisher |
| forbes.com | 12 | General media, product roundups |
| goodrx.com | 10 | Health publisher |
| nccih.nih.gov | 9 | Government |
| everydayhealth.com | 9 | Health publisher |
| goodhousekeeping.com | 9 | General media, product roundups |
| Engine | Answers | Citations | Leans on |
|---|---|---|---|
| ChatGPT search | 50 | 109 | Fewer sources per answer; brand article libraries and study summaries when it cites a brand at all |
| Perplexity | 50 | 428 | Many sources per answer; publishers, magazines and review sites |
The product page is not what gets cited. When ChatGPT did link a brand's own site, the page was an article, buying guide, study summary or press release, never a product page. One leading practitioner brand was named in six answers and its site was cited in none of them: publishers were telling its story. Most of a supplement brand's AI visibility is decided on other people's sites, and the two engines disagreed on sources for 20 of the 50 questions, so checking one engine shows part of the picture.
What gets a supplement brand cited
The four conditions a page has to meet before a model cites it, applied to a supplement brand.
Crawlers can reach and parse the page
robots.txt that admits GPTBot, ClaudeBot and PerplexityBot, no crawl delays, a visible H1, an llms.txt at the root, and a sitemap in both search consoles. Shopify stores usually need llms.txt and the robots.txt rules added by hand.
One entity everywhere
Organization and Product schema with sameAs links, matching records on Wikidata, Crunchbase and Bing, Amazon and retailer listings that match the site, so every engine resolves the brand to one identity.
Evidence a model can quote
An article library on your own domain: ingredient pages with the clinical dose, the study it comes from (PubMed or PMC) and what the evidence does not show. That is the page type ChatGPT linked in our study. Automated QA checks every citation before it ships.
Named where the models already look
Being named by the publishers and review sites the engines read: journalist sourcing (Qwoted, Featured), Reddit threads answered under your own name, YouTube transcripts, and supplier or retailer pages that name you.
The rules every page has to clear first
Supplement marketing is governed by DSHEA and the FTC, and a page that converts by overstating a benefit is a liability. We write to these rules so that compliance sign-off is routine.
- Structure and function claims only. A supplement page may describe how an ingredient supports normal function; it may not claim to diagnose, treat, cure or prevent a disease. Every claim carries the FDA disclaimer where DSHEA requires it.
- Competent and reliable scientific evidence. The FTC's Health Products Compliance Guidance expects claims to be backed by evidence that matches the claim: the right ingredient, dose and population. We cite the study the dose came from, not a review of a different form.
- Endorsements and reviews disclosed. Testimonials and influencer content follow the FTC Endorsement Guides: material connections disclosed, typical results stated, no claims a product page could not make.
- Label and site agree. The Supplement Facts panel, the product page and the article library state the same doses and forms. Our founder publishes an open dose-disclosure audit of this category.
This is how we run the claims register, not legal advice. Your counsel or compliance officer signs off on the rules that apply to you; we build so that sign-off is easy to give.
Proof
CalmGut, now Shuven, is the founder's own supplement brand: a 17-page content rebuild with verified PubMed citations, llms.txt, Organization schema and a Wikidata entity. Its first sale came from a visitor referred by ChatGPT, and 11.1% of sessions referred from chatgpt.com converted, the only channel that did. Numbers are small and reported as measured in Search Console and GA4.
Check a question your buyers ask
Pick one or write your own, then run it in the free checker to see which sites each engine cites today and whether yours is one of them.
Frequently asked questions
Do AI assistants cite supplement brands' own websites?
Rarely. In our October 2026 study of 537 citations across 50 health and supplement questions, brand and company sites took 11% of citations, and the pages that were cited were articles, buying guides and study summaries, never product pages. Read the study.
Can a supplement brand do GEO without breaking DSHEA and FTC rules?
Yes. GEO is about being a source the engine trusts, which in this category means accurate ingredient pages with real clinical citations and no disease claims. The compliant page and the citable page are the same page. More on GEO for regulated health.
Which engines matter for a supplement brand?
All of them, because they disagree. ChatGPT search and Perplexity shared no source on 20 of our 50 questions. We track ChatGPT search, Perplexity and Claude, and report Google AI Overviews from Search Console.
What does it cost?
Engagements begin with a Foundation build, from $15,000. The GEO Retainer follows from $5,000 per month. The Deep Audit and Roadmap is a standalone $5,000 diagnostic, credited toward a Foundation started within 60 days.
Start with the audit.
The AI visibility audit runs the questions your buyers actually ask, maps who AI cites instead of you, and ranks what to fix first. Complimentary, by application. Engagements begin with a Foundation build, from $15,000; the GEO Retainer follows from $5,000 per month.