By Steve Merrill, Founder of WRKNG Digital — July 28, 2026
What do Perplexity and Google AI Overviews actually see when they look at your Shopify store?
They see raw HTML text and structured data. Not your fonts, not your hero video, not the star rating your review app paints on after the page loads. They read the page a shopper never reads, and that page is usually thinner than you think.
I've pulled the raw source on a lot of Shopify stores this year. The gap between what the owner sees and what the machine sees is the whole game right now.
Here's what I mean. You open your product page in a browser. You see a clean gallery, a size selector, 340 reviews at 4.8 stars, a description that scrolls. Perplexity opens the same URL and gets a text file. If the good stuff isn't in that text file as words or as schema, it may as well not exist.
How does an AI engine read a Shopify page differently than a human?
A human reads the rendered page after your browser runs every script. An AI crawler often reads the raw HTML before most of that JavaScript executes. So two very different versions of your store exist at the same time, and you only ever look at one of them.
Google has said for years that its crawlers can render JavaScript, but rendering is a second, slower pass, and it isn't guaranteed for every page or every engine. Perplexity's crawler leans on the initial HTML response. Perplexity documents its crawlers and how they fetch pages, and the short version is this: give it text in the HTML, not text that appears later.
Three things live in that raw response and drive what gets quoted:
- Body text. The words physically present in your HTML. Descriptions, specs, FAQs, shipping details.
- Structured data. The JSON-LD schema that hands over price, availability, brand, and ratings in a format the engine trusts.
- Extractable signals. Headings, lists, and alt text that tell the engine how the page is organized and what each part means.
Everything a shopper loves and the machine can't read falls into a fourth bucket. The invisible one.
Why does Product schema decide whether you get cited?
Product schema hands the engine clean, machine-readable facts. Price, availability, brand, GTIN, and aggregate rating in a format it doesn't have to guess at. When those fields resolve without errors, the AI can state your price and stock with confidence and put your store in the answer.
Shopify themes ship with some structured data, but it's inconsistent by theme and often incomplete. I've audited stores where the Product schema had a name and an image and nothing else. No price. No availability. No rating. The engine reads that and treats the product as a mystery box.
Google's own product structured data guidance is direct about the fields that matter for shopping surfaces, and Schema.org's Product spec lists the full set. The ones that move the needle for AI answers are price, priceCurrency, availability, brand, and aggregateRating. Miss those and you're asking the engine to trust a store it can barely describe.
Run your product page through Google's Rich Results Test. If it throws warnings on price or availability, that's not a cosmetic problem. That's the AI reading a broken record of your catalog.
Why don't your reviews show up in AI answers?
Because most review apps paint the stars on with JavaScript after the page loads. The 4.8 and the 340 reviews your shopper sees never enter the raw HTML. The crawler reads a product page with zero review signal, so the AI has nothing to quote about how people rate you.
This one burns stores constantly. You paid for social proof. Your customers see it. The machine reads a blank.
The fix is making the review data render as real markup in the HTML, backed by AggregateRating schema with a matching ratingValue and reviewCount. Judge.me, Loox, and the rest can do server-side rendering or schema output, but it's usually off by default or theme-dependent. Check the raw source. If you can't find your star rating by reading the HTML, neither can Perplexity.
What content can these engines not extract at all?
Anything trapped in an image or a video. Your size chart as a graphic. Your fit notes baked into a lifestyle photo. Your ingredient list on a packaging shot. The engine sees an image tag and, at best, whatever alt text you wrote. Usually that's nothing useful.
Shopify stores love to design instead of write. A gorgeous product page can carry almost no extractable text. When Perplexity gets asked "is this jacket warm enough for winter," it needs a sentence on the page that says so. A photo of someone in the snow doesn't count.
Put the specs into words. Fit, materials, sizing, use cases, care. Plain sentences the extractor can lift and quote. That's the difference between being the answer and being skipped.
How do you close the gap between what AI sees and what your store shows?
Start by looking at the page the way a bot does, then fix what's missing. Four steps, in order.
- Fetch your page like a crawler. Pull the URL with curl or Google's URL Inspection tool and read the raw HTML. Whatever text you can't find there, the AI never saw.
- Validate your Product schema. Run the page through the Rich Results Test. Confirm price, availability, brand, and aggregateRating all resolve with no warnings.
- Move review data into the HTML. Get star ratings and counts rendering as real markup with matching AggregateRating schema, not a widget that loads after the fact.
- Write the specs into body copy. Take the facts out of your images and put them into sentences on the page.
Do those four and you've handed both engines a page they can actually read, trust, and quote.
The bottom line
Your store has two versions. The one your shoppers love and the one the machines read. Right now most owners only ever look at the pretty one.
Perplexity and Google AI Overviews are picking answers out of the plain one. Text and schema. That's it. If your best selling points live in images and JavaScript widgets, you're invisible in the exact place shoppers are starting to buy.
We map this gap for ecommerce brands every week, page by page, and fix the raw source so the AI reads what your customers already see. If you want us to look at your store, start here: WRKNG Digital agentic commerce audit.
Frequently asked questions
Do Perplexity and Google AI Overviews read my Shopify store the same way a shopper does?
No. A shopper sees rendered images, animations, and interactive widgets. The AI engines read raw HTML text and structured data. Anything that only appears after JavaScript runs, or lives inside an image, is often invisible to them.
Does Product schema actually change whether I get cited?
Yes. Product and AggregateRating schema give the engine machine-readable price, availability, brand, and rating data. When that data is present and clean, the AI can quote your store with confidence instead of guessing or skipping you.
Why do my reviews show up on the page but not get quoted by AI?
Most review apps inject stars through JavaScript after the page loads. The number a human sees never enters the raw HTML, so the crawler reads a product page with zero review signal even though your shoppers see 4.8 stars.
How do I check what an AI engine sees on my store?
Fetch the page with curl or Google's URL Inspection tool and read the raw HTML. Then run it through Google's Rich Results Test. What's missing from that view is what the AI is missing too.
