- Why do marketplaces outrank our own site in AI answers?
- Marketplace listings are heavily structured, constantly updated and carry a review corpus attached to each product, which makes them cheap for a model to parse and easy to attribute. A brand site frequently has the better information and presents it as prose inside a JavaScript-rendered template. The gap is usually machine readability rather than merit.
- Should we track every product?
- Tracking every product is not possible within any published prompt allowance and would not be useful if it were. Pick the category questions that decide a purchase, the comparisons against brands you genuinely lose to, and the handful of hero products that carry the range. Everything else is noise you are paying to collect.
- Do reviews affect what an assistant says about us?
- Reviews give a model corroborating material written by someone other than the brand, which is exactly the kind of source it prefers when a question implies risk or spend. The effect is on what an answer can say with confidence rather than on any ranking, and a thin review corpus tends to produce hedged answers rather than absent ones.
- Does Shopify or WooCommerce change any of this?
- Both server-render product pages by default, so the common fatal problem is usually absent. What varies is whether the theme emits complete product markup and whether the collection and content pages render the same way as the product pages. That is worth checking directly rather than assuming the platform handles it.
- Is this worth doing at our size?
- The honest test is whether your buyers research before purchasing. Considered purchases with comparison steps are already moving into assistants; impulse categories much less so. Run a free check and put twenty real buyer questions to an assistant yourself before spending anything, and let the answers decide it.