How it works
There are two families:
- Structured data sits in fields: size, material, dimensions, price, stock. It's used to filter and compare.
- Unstructured data is free text: descriptions, guides, FAQs, customer reviews. It's used to explain and advise.
It comes from several tools: the e-commerce platform (Shopify, PrestaShop…), a PIM if the brand has one, the ERP for stock, the reviews tool. An AI assistant needs both families. Fields let it find the right products, text lets it explain why they fit.
Good product data meets three criteria: it's complete, consistent across products (same units, same field names) and up to date.
Why it matters in e-commerce
An AI assistant answers from product data. If an accessory's compatibility isn't written anywhere, it can't confirm it, and the shopper hesitates or gets it wrong. Questions about size, fit, compatibility and dimensions are also the ones that drive returns when the answer is missing.
The same data feeds consumer AI assistants like ChatGPT. An incomplete product page hurts the brand on its own site and outside it.
Real-life example
On a furniture brand's site, a shopper asks: "Will this loft bed fit under an 8-foot ceiling?" If the bed's total height is in a field on the product page, the assistant answers right away. If it only appears on a diagram image, the assistant can't read it, and the shopper has to search alone or contact customer service.
Common confusion
People often mix up product data and product feeds. A feed is an export of the data in a given format, for a given channel (Google Shopping, a marketplace, an AI assistant). Fix an error in the feed without fixing it at the source, and it comes back with the next update.