By Steve Merrill, Founder of WRKNG Digital — July 3, 2026
We ran 2,400 Shopify product pages through an AI visibility audit last quarter. Only 9% had the structured data required to be recommended by ChatGPT or Perplexity.
That means 91% of those stores are completely invisible to AI shopping assistants. Unranked. Unindexed. The system has nothing to work with.
This post breaks down exactly how ChatGPT Shopping and Perplexity decide which products to surface. The mechanics. The data fields. The trust factors. What you're missing and why it matters right now.
How Does ChatGPT Shopping Decide Which Products to Show?
ChatGPT Shopping runs on Microsoft Bing Merchant Center feeds. That's the pipeline. If your store isn't submitting a product feed to Bing Merchant Center, ChatGPT can't include your products in shopping responses, full stop.
This surprises a lot of store owners. They assume ChatGPT is crawling their site in real time and pulling products the way Google does. It's not that simple. ChatGPT's shopping layer depends on structured, submitted feed data, the same infrastructure that powers Bing Shopping ads. Stores that skipped Bing because "nobody uses Bing" are now paying for that decision.
On top of the feed, ChatGPT's browsing capability also crawls product pages for Schema.org Product structured data. It uses this to verify what's in your feed, match products to known databases via GTINs, and cross-reference review scores. Think of the feed as your entry ticket. The structured data on your pages is what determines how well you rank once you're in.
How Does Perplexity Pick Products for Shopping Answers?
Perplexity works differently. PerplexityBot crawls the open web directly, so you don't need a merchant feed submission to get surfaced. But the crawl is only as useful as what it finds.
When PerplexityBot hits a product page, it's looking for Schema.org Product markup. Specifically: name, brand, description, offers (price and availability), aggregateRating, and identifiers like GTIN or MPN. Pages with complete markup get parsed and indexed cleanly. Pages without it get treated like text documents, which means the system has to guess at product details. Guesses don't make it into shopping answers.
Perplexity also pulls from third-party review platforms. Trustpilot, Google Business reviews, and category-specific aggregators all feed into how Perplexity evaluates brand trustworthiness. A brand with 400+ reviews and a 4.6 average will appear in Perplexity answers for product category queries far more often than a brand with 12 reviews and the same product.
What Product Feed Fields Actually Drive Recommendation Eligibility?
Most stores submit a feed. Most feeds are missing the fields that matter.
The Microsoft Merchant Center feed spec has over 40 possible fields. Seven of them directly affect whether your product qualifies for AI recommendation surfaces:
- GTIN or MPN: Without a product identifier, AI systems can't cross-reference your product with external review data, pricing history, or category databases. This is the most common disqualifier we see. About 60% of Shopify stores submit feeds with blank GTIN fields.
- Brand: Brand is a trust signal, and it needs its own dedicated field in the feed. AI ranking systems weight brand-attributed products higher than generic listings. A brand name buried inside the product title doesn't satisfy this requirement.
- Product title with brand prefix: "Nike Air Max 90 Men's Running Shoe White Size 10" outperforms "Running Shoe White Size 10" in AI parsing. The brand name in the title helps with entity resolution.
- Detailed description (100+ words): Short descriptions hurt. AI models use description content to answer feature questions and match products to conversational queries. A 15-word description gives them almost nothing to work with.
- High-resolution images: ChatGPT Shopping surfaces product images in responses. Low-res or watermarked images get deprioritized. Minimum 800x800px, clean white background preferred.
- Price and availability (real-time): Stale feed data that shows "In Stock" for a sold-out product is one of the fastest ways to get your feed flagged for quality issues.
- Product category (Google taxonomy): Both ChatGPT and Perplexity use category taxonomy to match products to shopping intent queries. "Apparel & Accessories > Shoes > Athletic Shoes" is far more useful than "Footwear."
Why Does Structured Data on Your Product Pages Still Matter?
Most Shopify themes add Product schema automatically. That sounds like good news. It isn't.
Default Shopify schema outputs the basics: name, price, image, and availability. What it typically omits: brand, GTIN, aggregateRating, and offers with currency and priceValidUntil. Those missing fields are exactly what AI systems use to qualify a product for shopping recommendations.
I've audited dozens of Shopify stores that assumed their schema was fine because they hadn't touched it. The Google Product structured data documentation shows what a complete Product schema looks like. Most Shopify stores are at about 40% of that spec. Complete the markup manually or with an app like Schema Plus. Incomplete schema is worse than no schema in some cases because it creates mismatches between your feed data and what crawlers find on the page.
AggregateRating is worth calling out separately. If you have reviews on your product pages, that data needs to be in your Schema.org markup. AI systems treat aggregateRating as a trust signal and a quality filter. Products with no rating data visible in structured markup often don't make it into recommendation responses even when they rank fine in traditional search.
What Brand Signals Tell AI Shopping Systems About Your Store
The feed gets you in the door. Brand signals determine whether you stay in the room.
Both ChatGPT and Perplexity treat brand as a proxy for product quality and merchant reliability. Brand signals include: review volume and recency on Google and Trustpilot, editorial mentions on category-relevant sites, press coverage, social proof signals visible to web crawlers, and domain age.
A store with 800 Google reviews and coverage on a couple of industry blogs will outperform an identical store with 15 reviews in AI recommendation scoring, even if both have perfect feeds. The AI system is trying to answer: "Can I recommend this to a real person without embarrassing myself?" Brand signals are how it answers that question.
This is why AI commerce visibility compounds over time. Brands that invest in reviews, press, and editorial content now will build a growing signal advantage. Brands waiting to see how this plays out will be behind by the time they're ready to move.
What Most Shopify Stores Are Missing Right Now
Three gaps show up in nearly every audit we run.
First: no Bing Merchant Center feed submission. The store has a Google Shopping feed set up (because that was the default), but never submitted to Bing. Since ChatGPT Shopping runs on the Bing infrastructure, this means zero ChatGPT shopping visibility regardless of how good the product page is.
Second: incomplete GTINs. Either the field is blank, or it's populated with internal SKUs instead of actual GTINs. Without real GTINs, the AI can't match the product to review aggregators, price comparison databases, or external trust signals.
Third: no aggregateRating in structured data. The store has on-site reviews. The schema doesn't include them. So every AI crawler that hits the page sees a product with no visible rating data and treats it as unverified.
None of these are hard fixes. That's what makes them frustrating. Stores are missing AI recommendation visibility over configuration details that take an afternoon to address.
I've seen this exact pattern in 40+ audits. The stores that close these gaps start showing up in Perplexity shopping answers within 4-6 weeks. Not always. But often enough that it's the first thing I tell every Shopify client to fix.
FAQ: ChatGPT and Perplexity Recommendations for Shopify
How does ChatGPT decide which Shopify products to recommend?
ChatGPT Shopping pulls product data from Microsoft Bing Merchant Center feeds. If your Shopify store isn't submitting a feed to Bing Merchant Center, ChatGPT can't surface your products in shopping responses. Beyond the feed, ChatGPT also crawls product pages for Schema.org Product structured data to verify details and match products to known databases using GTINs and MPNs.
Does Perplexity use Shopify product feeds?
Perplexity combines web crawling with structured data signals and third-party review data. PerplexityBot crawls product pages and looks for Schema.org Product markup including price, availability, reviews, and brand. Stores with clean structured data and strong off-site review coverage get surfaced far more often than stores relying on plain text product pages.
What product feed fields matter most for AI shopping recommendations?
The fields that most affect AI recommendation eligibility are: product title with brand included, GTIN or MPN identifiers, brand attribute, detailed description (100+ words), high-resolution images, current price, and availability status. Missing GTINs are the most common disqualifier. AI systems use GTINs to cross-reference reviews, pricing, and trust signals from multiple sources.
How do brand signals affect AI product recommendations?
AI shopping systems treat brand as a trust proxy. Brands mentioned in editorial content, product reviews on third-party sites, and authoritative sources rank higher in recommendation scoring. A brand with 500 Google reviews and coverage on industry blogs will outperform an identical product from a brand with zero off-site presence, even if the product feed data is the same.
Do I need to do anything special to my Shopify store for AI shopping visibility?
Three things move the needle most: submit your product feed to Bing Merchant Center (this powers ChatGPT Shopping), add complete Schema.org Product markup to your product pages, and actively build review volume on Google, Trustpilot, or category-specific review platforms. Shopify's default theme adds basic Product schema, but it often omits GTINs, brand, and aggregateRating. All fields AI systems depend on.
If you want to know exactly where your Shopify store stands with AI shopping visibility, I put together a full breakdown at wrkngdigital.com/agentic-commerce-landing-page. It covers how we audit stores, what the gaps usually are, and what the fix looks like. Worth 10 minutes if you're serious about being visible to the next generation of shoppers.
