The questions people actually ask.

It reads your product pages the way an AI assistant reads them, scores them on four dimensions, tells you which checks failed and why, proposes changes as before and after diffs, and writes the ones you approve back to Shopify. After a push it rescores the affected products so the movement is measured rather than claimed.

It removes the reasons an assistant cannot use your pages: unreachable content, missing identifiers, unanswered buyer questions, thin locales. Whether any specific assistant recommends you is decided inside systems no vendor controls, so we do not promise placements, rankings or sales lifts. We show you the score before and after, and you can read the checks yourself.

SEO tools optimise for human search results. The checks here are about machine readability and answerability: server rendered substance, product and offer markup, identifier coverage, whether a page answers a purchase question, and parity between locales. There is overlap, but a page can rank well and still be unusable to an agent.

No. Scoring, rules, schema, identifiers, markup and parity carry across are all deterministic and free of credits. Drafted prose is the only part that uses a language model, and it is optional. Some merchants only ever run rules.

No, and this is deliberate. There is no automatic publishing mode and no scheduled write. Every write follows an approval you made in the interface, and the record shows who approved it and when.

The structural checks are deterministic: markup either parses or it does not, an identifier is either present or absent. The answerability dimension is a simulation and therefore a judgement, made from what is visible on the page. We publish every check in the methodology so you can disagree with a specific one rather than the number as a whole.

Your catalog content as read through the Shopify API: products, variants, metafields, collections, translations and image alt text. The scores derived from it, the drafted content, and the record of changes you approved. Plus your account details and billing records. We never request or store orders, customers or payment data.

In the European Union. Our database and storage run on EU regions, and our own company is Luxembourg based. The subprocessor list names every provider and the region it processes in.

No. Your content is sent to language model providers only to draft the specific change you asked for, under agreements that prohibit training on it, and is not used to train any model of ours. The subprocessor page names the providers.

Only your account. A public scorecard exists only for the free domain check, and only through the signed link generated for that scan, which you can choose not to share. Connected catalog data is never public.

Yes, both, from Settings, Security. Export delivers everything we hold in a machine readable form. Deletion is permanent and cancels any active subscription. You can also write to privacy@getfeedmind.com.

No. Feedmind writes product fields, metafields, translations and image alt text through the Shopify API. It does not modify theme files and does not inject scripts into your storefront.

Access is revoked immediately by Shopify. We act on the mandatory data webhooks and delete your catalog data, scores and drafted content within thirty days, and we cancel any active subscription. Changes you already published stay in your store, because they are your content.

You can run the free domain check without any connection and without an account. Everything beyond that, catalog wide scoring, rules and fixes, needs the store connection because it works on your real product records.

Not yet. The free domain check works on any storefront, but the connected product is Shopify only today. If you are on another platform, tell us at hello@getfeedmind.com and we will tell you honestly whether we are close.