EndcapHow ChatGPT decides which products to recommend
Updated 2 August 2026
When someone asks an assistant what to buy, it does not browse your theme, read your hero image or notice your brand fonts. It reads structured product data. That single fact reorders most of what merchants have been told about ecommerce presentation.
Two gates, in order
First inclusion: is the product in Shopify Catalog at all. Then matching: does the product data align with what the shopper described. Most merchants optimise for the second while failing the first.
What gets read
Shopify names the fields explicitly: title, description, images, type, vendor, collections, tags, barcode, and variant option names. Availability and price are syndicated live.
Notice what is absent from that list: your theme, your page speed, your alt text, your schema markup. Those matter for Google. They are not what an assistant is reading here.
Why specificity wins
Shoppers ask assistants in natural language and with constraints: something waterproof, machine washable, under a hundred, good for a toddler. A description that never mentions those attributes cannot be matched to those requests, however well written it is.
How to find out where you stand
Ask the questions yourself. Put ten buying questions from your category to an assistant and count how many name your store. Then ask which brands it named instead. That number, tracked over time, is the only honest measure of progress.
Check your own catalog
Endcap runs every check in this guide and shows you which products fail, and why.
Add to Shopify, free