Are your product reviews visible to AI agents?

A practical guide for ecommerce brands on Shopify

Product reviews are one of the strongest trust signals an ecommerce brand can have. But for many Shopify brands, those reviews are invisible to AI agents, not because they are not there, but because of how they are technically delivered to the page. This post explains why that happens, what to look for, and how to make sure your reviews are working as hard as they should.

Why reviews matter in AI search

When a shopper asks an AI agent to recommend a vitamin supplement, a moisturizer, or a sleep aid, the response is built from evidence. AI agents are trying to form a confident, defensible recommendation, and the content they trust most is authentic, specific, and verifiable.

Product reviews sit at the top of that trust hierarchy. They are written by real buyers with no brand affiliation, they describe specific use cases and results, and they are the kind of content that signals to an AI agent that a product has earned its reputation rather than simply claimed it.

For beauty, wellness, and supplement brands especially, this matters. These are categories where purchase decisions are personal and where a review that says "I have combination skin and this cleared my congestion within two weeks" carries more weight than any product description could. AI agents can use that kind of specific, contextual evidence to match your product to the right query and the right person.

The challenge is that this only works if AI agents can actually read the reviews in the first place.

How AI agents weight different types of content

Understanding why reviews matter requires understanding the broader content landscape AI systems draw from. Based on observed patterns in AI recommendations, consistent with analysis from Yotpo and the broader SEO and GEO research community, there is a clear hierarchy in how different content types influence AI recommendations.

User-generated content sits at the top of that hierarchy. Verified buyer reviews on your own site, reviews on retailer and marketplace pages, Reddit discussions, YouTube reviews, expert comparisons, community Q&As, and affiliate content all sit in this tier. This content is weighted most heavily in observed AI recommendations because it reflects genuine, unfiltered product experience and is the hardest to manufacture at scale.

Influencer, affiliate, and media content sits in the middle. More credible than brand content, but still recognized as potentially incentivized.

Brand-owned content sits at the bottom. Your homepage copy, your campaign messaging, your product descriptions. Useful context, but consistently the least influential layer in shaping AI recommendations.

It is worth noting that this is an observed pattern rather than a published algorithmic rule. AI systems do not officially document a content weighting hierarchy. But the pattern is consistent enough across categories and platforms that structuring your content strategy around it is well-supported.

For beauty, wellness, and supplement brands, getting the top tier right is the foundation of AI visibility. Your reviews are your most credible asset, and making sure AI agents can access them is what allows that credibility to translate into recommendations.

The technical gap most Shopify brands are not aware of

Here is where things get specific. Even with hundreds of verified reviews, there is a good chance AI agents are retrieving only a fraction of that content, or none of it at all.

Most Shopify brands use third-party review apps including Yotpo, Judge.me, Loox, Stamped, and Klaviyo Reviews. These tools are well-built for displaying reviews to human shoppers. The challenge is that the way most of them deliver content to the page creates a gap for AI agents and automated crawlers.

In a standard integration, review content is loaded via JavaScript after the page has already been served. Star ratings are rendered visually as icon graphics rather than readable text. The aggregate rating, the average score and total review count, is injected dynamically rather than being present in the original HTML that an AI agent retrieves.

AI agents that retrieve raw HTML without executing JavaScript will find individual testimonial quotes written directly into the page, or marketing copy referencing a review count, but no structured data telling them what the actual rating is, how many reviews exist, or what those reviews say in aggregate.

The result is that a brand with 400 verified five-star reviews may be effectively invisible from a review standpoint to the AI agent a shopper is using to make their purchase decision.

Which AI agents are affected, and to what degree

This gap is most significant for AI agents that crawl pages directly without executing JavaScript. Research from Vercel tracking over half a billion crawl requests found no evidence of JavaScript execution across ChatGPT, Perplexity, Claude, Meta, and ByteDance's crawler. For these agents, content loaded via JavaScript is simply not retrieved.

The picture is more nuanced for AI systems that rely on pre-rendered search indexes rather than direct crawling. ChatGPT uses Bing's index for real-time retrieval, and Gemini uses Google's index. Both Bingbot and Googlebot have JavaScript rendering capabilities, which means some dynamically loaded review content may already be indexed and available to these AI systems via their search index pipelines.

In practice this means the gap is worst for direct AI crawling and somewhat less severe for AI systems relying on pre-rendered indexes. But it is not a reason to deprioritise the fix. Structured data in server-rendered HTML is the most reliable signal across all pipelines, both direct crawling and index-based retrieval, and it is worth implementing regardless of which AI system you are most focused on.

Three levels of review readability

When it comes to how well AI agents can read your review data, there are three levels worth understanding:

• Structured data in JSON-LD format

• Server-rendered HTML

• JavaScript rendering

Structured data in JSON-LD format is the gold standard. This is a small block of machine-readable code embedded directly in the server-rendered HTML of each product page. It tells AI agents, search engines, and crawlers exactly what the product is, what its average rating is, and how many reviews it has. It is present the moment the page is retrieved, requires no JavaScript to execute, and leaves no ambiguity about what the numbers mean. Once set up correctly it updates dynamically and requires very little ongoing maintenance.

Server-rendered plain HTML text is the second level. If the rating value and review count are written directly into the page as readable text, most crawlers and AI agents can extract them through pattern recognition. Less precise than structured data, but significantly more reliable than JavaScript-only rendering.

JavaScript-only visual rendering is the most common setup on Shopify stores using standard review app integrations. Stars displayed as SVG icons, ratings loaded after page render, no structured data present in the initial HTML. For AI agents crawling pages directly, this content is not retrievable. Even for index-based retrieval, structured data remains a stronger and more reliable signal.

What good review content looks like to an AI agent

Beyond the technical layer, the content of the reviews matters too. AI systems consistently surface reviews that are specific, recent, and contextually rich over generic praise, a pattern well documented in Yotpo's research into AI product discovery.

The reviews that appear most frequently in AI recommendations share a few characteristics. They come from verified buyers. They are recent, which signals an active and current product. They are specific, describing real use cases, skin types, routines, results, and comparisons. They include visual content where possible. And they exist within a broader community layer that includes Q&As and FAQs alongside individual reviews.

A review that says "great product, love it" contributes very little to AI visibility. A review that says "I have been using this for six weeks alongside a hyaluronic serum and it has completely resolved the dryness I was getting, I have dry to normal skin and live in a cold climate" is the kind of specific, contextual evidence AI systems can actually use to match a product to a relevant query.

How to close the gap

Closing this gap involves two workstreams running in parallel.

On the technical side, the priority is confirming that aggregate rating structured data is present in the server-rendered HTML of every product page. That means checking whether your review app outputs JSON-LD schema and whether it does so server-side rather than via JavaScript. Some apps require a higher plan tier to enable rich snippets, others need configuration changes in the theme. In some cases it requires adding the schema block directly to the product template. The good news is that once this is set up correctly, it updates dynamically and requires very little ongoing maintenance.

On the content side, the priority is generating reviews that AI agents can actually use. That means making it straightforward for buyers to leave detailed, specific feedback. Prompting for use case, skin type, routine context, or results rather than just a star rating. Building out community Q&A sections that address the questions shoppers are actually asking AI agents. And ensuring that review content is consistent and present across retailer and marketplace pages, since AI agents look beyond your own site when forming recommendations.

Where Glara fits in

This is one of the most common gaps Glara identifies when auditing Shopify brands across beauty, wellness, and FMCG categories. Structured review data is missing or inaccessible at the product level, which means AI systems are forming recommendations without the evidence that would most support them.

Glara audits AI visibility at product level, showing you which products have structured review data in place and which do not, where your review content is strong enough to support AI recommendations and where it falls short, and how your review visibility compares to competitors in your vertical who are already showing up in AI answers.

Critically, Glara does more than surface the gaps. Through Optimizations, Glara can generate and push the missing structured data directly to your Shopify store via the Shopify API, including product schema, aggregateRating markup, and enriched product attributes, without your team needing to manually update templates or involve a developer. The fix is not just identified, it is applied.

For categories where trust signals drive purchase decisions, getting this right is one of the highest leverage improvements an ecommerce team can make to their AI visibility.

The bottom line

Product reviews have always driven conversion. In an AI search world, they also drive discovery. Getting them in front of AI agents requires two things: making sure the technical layer delivers review data in a format AI agents can read, and making sure the review content itself is specific and credible enough to be used as evidence.

Most Shopify brands using standard review app integrations have a gap here worth addressing. The fix is specific and achievable, and for beauty, wellness, and supplement brands where reviews are the primary trust signal, it is worth prioritizing.

Sources: Vercel AI crawler research, 2025. Yotpo AI product discovery research, 2025 and 2026. Shopify GEO playbook, 2026.

Want to see whether your product reviews are readable by AI agents? Book a demo and Glara will show you exactly where the gaps are across your catalogue and fix them automatically.