Knows why they came. Shows you what it did about it.

No. It reads only what this visit does — what was searched, how many pages and categories, which prices were looked at, what went into the cart and came back out. A first-time visitor is read the same way as anyone else.

No. It ranks only where you switch Intent suggest on — per collection, and per upsell. Everywhere else it only reports. On the Pro plan it never ranks at all; it shows you the mix of visits your store gets.

Where it is on, products tagged with the visit’s intent move ahead of the others. Intent is one weighted signal in your ranking formula, so every other signal keeps its weight — and the collection editor shows what share of the score intent carries. Products need intent tags for it to change anything.

No. It reads five practical reasons to shop, from behaviour, with fixed rules rather than a model guessing — the app describes what each reason looks like. AI tags your products, writes the summaries in the Intent tab, and, in testing, learns which products each reason’s shoppers buy.

They see your normal order. Intent AI commits to a reading only when the visit gives it clear evidence; a visit it has not read is ranked exactly as it would be without Intent AI.

Run the Intent AI on/off test. A share of your shoppers get the same store without Intent AI, and the test compares how many shoppers in each group bought anything, anywhere on the store. It runs for a planned number of weeks and is read once at the end, so an early lucky streak cannot decide it.

No. What Intent AI learns — which products each intent’s shoppers buy, and a trained model — is shown to you in the Intent tab as in testing, measured against what you already have on visits it did not learn from, and kept off your storefront. Today your storefront is ranked by your formula and your product tags.