My supplement brand has strong on-page SEO, but we are invisible in AI Overviews. Is it possible that AI models are prioritizing third-party reviews and medical journals over direct merchant sites? How can I improve my ‘off-page’ AI presence so that LLMs recognize my product as a top recommendation in the supplement category?
Hi @mary123,
You’re right. AI models are very cautious with health products. They prefer “neutral” sources like medical journals or review blogs over store pages. This is because of YMYL (Your Money or Your Life) standards.
To get noticed by AI, you need to build trust outside of your own site:
- Third-party reviews: Get featured in “Top 10” lists on health blogs. AI loves these lists.
- Expert mentions: Reach out to dietitians or doctors to mention your brand online.
- Community talk: AI scans forums like Reddit. Authentic customer discussions there help a lot.
- Technical clarity: AI needs to understand your site structure perfectly.
A quick tip: to make sure AI can read your store’s data properly, use a tool like SearchPie: SEO, Speed & Schema. It handles the complex “Schema” and technical SEO that AI models look for. It’s an easy way to ensure your store is “AI-ready” while you focus on building those outside mentions.
Hope this helps solving your problems,
Try “Shopify catalogue” - Shopify is rolling out new features. I think limited to us market for now.
https://help.shopify.com/en/manual/shopify-catalog
Plus point:
Improve seo
Improve image sizes
Make the product data complete
Llms.txt, and llms-full.txt - there are still not much data about effectiveness of these standards
Ai overviews prefer social mentions. Mentions on forums like reddit/quora (even if less liked) are given priority. People are misusing that. If i mention some Shopify app here, it will get listed easily in the ai overviews even if nobody likes it.
Twitter started indexing it’s post and blogs.
Hi @mary123,
I’m Vineet from Identixweb, a Shopify development company.
Yes, that is possible, especially in the supplement category. AI Overviews may place more weight on independent reviews, qualified health experts, medical references, and established publishers because those sources appear less commercially biased than a brand recommending its own product.
To improve your off-page AI presence, I’d focus on two things:
- Genuine coverage from relevant health, fitness, or nutrition publishers
- Reviews from qualified experts, with sponsorships clearly disclosed
I’d also target specific searches where your product has a genuine advantage instead of trying to rank as the “best supplement” across a broad category. Which product and AI Overview queries are you currently targeting?
@mary123 ,
Yes thats very likely. AI Overviews and other LLMs dont rely only on your website they also look for signals from trusted third party sources. For supplements that often includes expert reviews reputable health websites scientific research, news mentions, and customer reviews.
To improve your off page AI presence focus on earning high quality backlinks, getting mentioned in reputable publications encouraging authentic customer reviews and building topical authority with expert backed content. The stronger your overall online reputation the more likely AI platforms are to recognize and reference your brand over time.
Everyone above is right that off-site trust signals matter, so I’ll add the part you can actually control on your own store. AI models tend to cite a source when they can pull a specific, checkable fact out of it, not a “best supplement” claim. For supplements that means the details most brands bury: exact ingredient amounts per serving, standardised extract percentages, third-party lab testing or a COA, allergen and sourcing info. If those live as real text and Product schema rather than baked into label images, you become extractable.
The reason journals and review sites win isn’t only that they read as neutral, it’s that they state facts an LLM can lift and attribute. If your product page is the clearest place on the web to get your actual composition, you start surfacing as the source for those factual queries, even while the “is it any good” queries stay with third parties. Off-page reputation and on-page verifiable data are two different jobs and you need both.
llms.txt is still unproven so I wouldn’t lean on it yet. Are your ingredient details currently live as page text, or mostly sitting inside your label images right now?
Thank you all, these are helpful. I will try fixing what you guys are suggesting first and get back if there are any further problems.
Yes, for supplements, independent sources usually carry more weight than merchant claims. I would not spend money on random Top 10 placements though. Those pages are often low trust.
- Pick 10 specific queries tied to your formula, not broad terms like best supplement. Record which sources AI Overviews cite, then pitch those publishers with product facts, testing data, and samples.
- Make COAs publicly accessible by batch, and publish ingredient amounts, extract standardization, allergens, sourcing, and testing methods as crawlable text.
- Ask qualified dietitians or pharmacists for honest reviews. Disclose payment or free products, and never script health claims.
- Build genuine reviews on established retail and review platforms, then monitor brand mentions and correct inaccurate ingredient or dosage information.
Also connect claims to actual studies on the ingredient and dose. Do not imply the study proves your finished product works unless that exact product was tested.
Everything above is sound, but it’s all one half of the picture. AI assistants source product answers from two different places, and the fixes are completely different.
Reader A: the open web. Your rendered pages plus schema markup, reached by crawling. This is where on-page facts, JSON-LD, and off-page trust signals do their work. Worth adding: AI crawlers read server-rendered HTML, not the fully hydrated DOM. If your ingredient panel or COA text loads client-side, it may simply not exist for them, worth checking what’s actually in the raw HTML before assuming the content is visible.
Reader B: the Shopify Catalog. The structured feed Shopify syndicates directly to assistants. It does not read your storefront theme, so no amount of schema apps or page copy touches it. It reads your product records. Six inputs decide whether you surface:
-
Standard Product Taxonomy category - assign the most specific node to every product (Health & Beauty › Health Care › Vitamins & Supplements › …). This is the single highest-leverage field, because the category is what unlocks the attribute set. An uncategorised product exposes no structured attributes at all.
-
Populated attributes - the category gives you empty containers; you have to fill them. Form (capsule / powder / gummy), count, dosage form, dietary claims, flavour, allergen info. This is where lumine’s point lands structurally: a query only matches if the attribute actually carries a value.
-
Catalog Mapping (Admin → Agentic → review how product data is sent to the Catalog) - controls which fields source the title/description/category, and how variants group. Get this wrong and your 60ct / 120ct / flavour variants split into near-duplicate listings instead of clustering as one product.
-
Product identity - GTIN/EAN/UPC on every variant, and the real brand name in the Brand field, not the store name. Missing GTIN is one of the most common reasons products silently fail to index in AI feeds.
-
Knowledge Base app - install and populate Shopify’s Knowledge Base with your policies, shipping, returns, and usage/dosage guidance in your own words. This is what an assistant answers from when a shopper asks a follow-up mid-conversation. Without it, the agent either guesses or leaves you out of the answer entirely. For supplements this is disproportionately valuable, since buyers ask questions before they buy. Keep it factual and avoid therapeutic claims.
-
Listing completeness - multiple clean images, reviews, full variant coverage, accurate live price and stock. Shopify scores this, and sparse records get skipped. The first five make you eligible; this one gets you chosen.
Sequence that works: clean data first (category + attributes across the catalog, bulk-edited), then structure (Catalog Mapping + variant grouping), then identity (GTIN/brand), then content, then Knowledge Base and completeness gaps.
Start with an audit, not a fix. Pull the share of SKUs missing a category, missing attribute values, and missing GTIN. That number sizes the whole effort and gives you a baseline to improve against - and it’s usually a lot worse than merchants expect.
One caveat: Catalog availability is still region-gated, so check your market. But none of the underlying work is wasted if it isn’t live for you yet - clean taxonomy, attributes and identifiers feed your Reader A schema too. And agreed on llms.txt, no evidence yet that it’s doing anything.
I wouldn’t assume the direct merchant site is the problem. For supplements especially, I’d expect third-party sources to carry a lot of weight because there’s more context around the product than just the product page itself.
I’d look at what shows up around the brand when you search the actual questions you want to be recommended for. Reviews, retailer listings, relevant publications, comparisons, and mentions from sites that already have some authority in the category are probably more useful than just adding more keywords to your own pages.
I’d still keep the technical side clean though. I’ve been using SiteGuru for the regular SEO audit rather than treating AI visibility as a completely separate project. If important pages aren’t indexed properly or the site structure is messy, I’d fix that before spending much time trying to build an “AI presence.”
The tricky part is measurement. I wouldn’t judge it by whether ChatGPT gives the product as a recommendation once. I’d track a set of specific prompts over time and compare what sources those answers are pulling from. That should give you a much better idea of where the actual gaps are.
Strong Google SEO doesn’t carry over to AI Overviews, they reward different things. Google ranks page authority. AI quotes whatever answers the exact question cleanly.
Fastest fix: take your top pages and lead each section with the direct answer, then the detail. And go narrow, “is X safe with Y,” “how much Z per day”, the specific questions a big retailer only answers generically. That specificity is what gets you cited over bigger brands.
Hi,
Yes, I think you’re looking at the right problem. For supplements, having strong on-page SEO alone may not be enough. AI systems need enough trustworthy, consistent information to understand what your brand and products are, especially in a YMYL category.
I’d focus on a few areas:
- Build third-party authority: Get genuine mentions, reviews, expert references, and relevant publications talking about your brand. Don’t try to manufacture mentions just for AI visibility.
- Make your expertise clear: Provide detailed, factual product information and make it easy to identify the people/experts behind your content.
- Create content around real questions: Publish useful, specific content that answers the questions people actually ask about your category instead of just targeting commercial keywords.
- Improve AEO/technical signals: Make sure your content and structured data are easy for search engines and AI systems to understand and extract.
- Keep your information consistent: Your brand, products, ingredients, claims, and other important details should be clear and consistent across your website and external sources.
For the technical/AEO side, I’ve been working on AutoRank: SEO Blog & CRO. It includes AEO data for AI search engines along with technical SEO/site audits and AI-generated SEO content, so it can help identify the on-site gaps while you work on building your off-site authority. There’s also a free plan if you want to test it.
Hi, you say your page has strong on-page. Does it rank on the first page for the keywords your after?
It’s best to grab your keywords and check what is being pulled in the AIO. its it supplement competitors or thirdparty listicles etc
If its places you can get featured try that, most pages adding a product you’ll need an affiliate program or pay for an article update.
For off page, what we do is track sticky citations, so with ai search the results are non-deterministic - meaning they change. So it’s important to find citations coming up constantly in your niche and reach out to them for a product placement so best results.
I recently did a case study and found:
Authority helps but does not decide sticky citations. A DR 7 site out-cited Forbes by 76%, and five of the 25 permanent sources score under DR 60.