QSortby Guide
Upsell

Intent suggest

Let the intent read choose between your ranked candidates, so a gift-buyer and a shopper replacing something get different suggestions from the same cart — or the same feed. Rules first; a learned model only where it has proved it picks better.

Last updated · Nov 7, 2026

Every offer ranks by the signal you pick — velocity, margin, conversion rate. Intent suggest adds a second, lighter pass on top: among the products your ranking already chose, it prefers the ones that suit what the shopper came for.

Applied live on Growth. On Pro the same read runs in shadow — visits are classified and reported in Results → Intent, and your storefront is untouched. (Every new install runs on the Growth feature set for its first 10 days — see Your first 10 days — so you’ll see it working before you’re paying for it.)

The five intents

Intent AI reads each visit as one of five reasons to be in the store:

IntentWhat the visit looks likeWhat sells
Treating themselvesUnhurried, lingering on the dearer items — often late evening or the weekend, in from social on a phone. No deadline, nothing particular in mind.Premium bundles and the nicer size. Don’t discount.
Replacing somethingArrived knowing what they want and moving fast — an urgent-reading search, quick scrolling, the cheaper end. Restocking, not exploring.Make the obvious choice easy to find. Fewer options, not more.
Buying a giftShopping for someone else — gift wording in the search, or ranging across unrelated categories at prices all over the map. Usually a first-time visitor.Gift sets, wrapping and delivery dates.
BrowsingLooking around with nothing picked out — many pages across many categories, a long session, nothing in the cart.Range and new arrivals. Nothing here is ready to be closed.
Close to buyingCame straight here, has been before, acting decisively — few pages, something in the cart, nothing taken back out.Stock, shipping and returns. Anything else is a distraction.

The read comes from how the visit behaves — scroll speed, dwell, cart activity, referrer, what was searched — not from who the shopper is. No purchase history is required, nothing identifies anyone, and the app reports a margin (how clearly one intent led) rather than dressing coverage up as confidence.

Where you’ll find it

A checkbox in the editor of the touch points that can identify a shopper: cart drawer, popup, in-checkout and thank-you page. On the cart drawer and the popup it only appears in automatic mode — a hand-picked list is your order, not ours.

It also appears on a Personalized collection and on the For You feed. There it works differently: the intent read becomes the Intent match signal inside your personalization rule, weighted alongside the others.

It’s off by default on every offer and every collection, including new ones. Turning it on is a decision you make per surface.

What it needs to work

Intent keywords for your industry. In Settings → Product data & AI → Intent keywords each intent is described by words matched against your product titles, tags and types. The shipped list is a starting point that is wrong somewhere in every shop — “fresh” is a scent family in home fragrance and a shelf life in groceries — so read it against your own product names and remove any word that means something else in your catalogue. A wrong word does not sit idle; it tags the wrong products.

A tagged catalogue. Every product QSortby has embedded carries one or two intents, the primary first. The Product intent tags list on the same tab shows what carries each intent; filter it to Weak matches only to see the words that are failing. You can replace the AI’s tags on any product — from the five, one or two — and your choice stays until you press Generate embeddings again, or hand the row back with Use AI’s pick.

A shopper the layer has read. A visit is read as it goes, and the read firms up as the shopper does more. A visitor who lands, buys in forty seconds and leaves may never be read at all.

What it does and doesn’t do

It nudges, it doesn’t take over. Your chosen signal still decides the pool and still dominates the order; the intent pass moves matching products up within it. A product that isn’t selling won’t be promoted into the cart drawer because it happens to carry a tag.

It’s also fail-quiet by design. No reading, no tagged products, or any hiccup reading them, and you get your normal ranked order — never an empty widget and never an error in front of a shopper.

What “learns” means here, honestly

Out of the box the pass is rules, not a model: a product’s intents come from its own text matched against the keywords; a shopper’s intent from their visit. Match the two and the product moves up. That works from day one.

Behind it, QSortby records what shoppers with each intent went on to buy at your store, and recomputes a per-intent pick nightly. Those learned picks replace the tag match only where they have shown, on your recent orders, that they pick better than your own best sellers — and the Intent tab says, per intent, whether that is on and why. Until then the storefront ranks by the tag match, and the guide says so rather than claiming a learned model is running.

Proving it

Because each shopper sees a different order, this layer is invisible from the storefront. Two places make it visible:

  • Results → Intent — what shoppers came for in the window you pick, how each kind behaves, what is trending with each, and when each peaks.
  • The on/off test — an experiment that switches intent off for half your shoppers on For You and every offer that ranks by intent, for a length you set. At the end it says how much more (or less) often the intent group bought, with a range — and if there is no clear difference, how large a difference the test could have seen. Home shows the result beside the running read.
Menus are tucked away for bigger images — hover the edges to peek them, or Focus to pin. Click any image to zoom.

Popular

Set up QSortby — two steps Getting started Auto-sort a collection Auto-sort Create a personalization rule Personalization Upsell offers — the six touch points Upsell Troubleshooting Help
↑↓ to navigate ↵ to open esc to close