By Steve Merrill, Founder of WRKNG Digital | September 3, 2026
Your spec table looks clean to a shopper. To an AI assistant it's often a blur of unlabeled numbers. That gap is why the model quotes a competitor's specs and skips yours.
I've checked this on more than 40 Shopify stores this year. The specs are usually right on the page and unreadable to a machine. Same facts, wrong container.
Here's how to structure your product specs so an AI reads them correctly the first time.
Why does AI misread specs that look fine to me?
Because a spec only means something when it's labeled. A shopper sees "200" next to a picture of a jacket and knows it's grams. A model sees "200" with no field name and guesses, or drops it.
Most Shopify themes render specs as styled text or an image, not as labeled data. The label lives in the design, not in the source. Strip the design and the meaning goes with it.
Step one: name every spec as a field, not a row of text
Put each spec in a labeled attribute. Material: merino wool. Weight: 200gsm. Size range: XS to XXL. Not a paragraph, not a picture. A field with a name and a value.
Do this in your product data first, then mirror it on the page. When the label and the value travel together, a model can lift the pair without guessing.
Step two: put the specs in Product schema
Move the same labeled specs into your Product schema. Use the standard properties for what fits, and additionalProperty for the rest. Then run the page through the Schema.org validator to confirm the values are real, not blank.
Most themes ship the schema skeleton and leave the values empty. That's the trap. An empty field reads as no field. Fill them with live product data.
Step three: match the specs to your feed
Your specs also live in your product feed. Line them up with the Google Merchant product data specification so the feed, the schema, and the page all say the same thing.
When those three agree, a model trusts the spec. When they disagree, it treats the whole product as unreliable. Consistency is the point, not just presence.
Step four: write the top three specs into plain text
Take the three specs that decide the sale and state them in a plain sentence high on the page. "This base layer is 200gsm merino, rated to 20 degrees, sized XS to XXL."
That sentence is quotable. An assistant answering a spec question can lift it whole. This is the read we give stores in a WRKNG audit. Which specs a model can see, which it can't, and where your data disagrees with itself.
Further reading
Frequently Asked Questions
Why does AI get my Shopify product specs wrong?
Because specs only carry meaning when they're labeled, and most themes render specs as styled text or images where the label lives in the design instead of the data. A model reads the raw value with no field name and either guesses or drops it.
Where should product specs live for AI to read them?
Put each spec in a labeled field in your product data, mirror it in Product schema with real values, and match it to your feed. When the label and value travel together across all three, a model can lift them without guessing.
Do I need additionalProperty in Product schema?
Yes, for specs that don't map to a standard schema property. additionalProperty lets you attach named spec pairs like material or weight so an AI reads them as structured data rather than loose text.
How many specs should I put in plain text?
State the three specs that actually decide the sale in one plain sentence high on the page. That sentence is quotable, so an assistant answering a spec question can lift it whole instead of skipping your product.
Want to know if AI assistants can actually find and recommend your store? Get a free AI-visibility read on your Shopify store at WRKNG Digital. We show you exactly what ChatGPT, Perplexity, and Google AI see when they look at your products.
