How to recommend the right products based on page and cart activity

Most product suggestions in a Shopify store never change. The same four items sit under every product, in the same order, regardless of whether the shopper is on their very first visit or has a cart well above your average order value. Page type and cart state are the two cheapest signals already at your disposal, and they can turn a slot that is merely filled into a suggestion that is genuinely useful.

These signals work best on a surface the shopper opens themselves: a bell and feed sitting beside the popup you already run rather than replacing it. Atomato is that feed. No email service provider is required, and no additional overlay goes on your theme.

Start with page type

Different page types call for different messages, and the page a shopper is on is already known.

On a product page, the useful recommendation is complementary, not similar — the thing this bag goes with, not three near-identical alternatives. Showing lookalikes mostly invokes doubt. Showing the refill, the case, the second size, or the care product is additive. One or two items keeps the message easy to read in the feed.

On a collection page, the shopper is still comparing. The best-reviewed or best-selling item in that collection is worth more than a cross-sell here, because the question is still which one, not what else.

The homepage is the least known context. It is the right place for restocks, new drops, or a current offer rather than a product-specific recommendation. If you have nothing to say right now, showing nothing is absolutely fine.

Don't overload the cart page. Only show an item that fills a real gap in the order, and keep the message short. A visitor on the cart page is almost done, and an unwelcome interruption hinders more than it helps.

Use cart value thresholds

Cart value is the second signal, and it changes what a good recommendation looks like. Below your free shipping threshold, the honest message is the gap itself plus one item that would close it. Pick a product priced near the difference, not your most expensive one. Shoppers can tell when a suggestion is sized for them and when it is sized for you.

The Zibra case study illustrates this on a paintbrush store where a second size of a brush or an artist set is suggested in the feed to shoppers who initially wanted just one brush.

Above the threshold, stop pushing the shipping line and switch to a genuine add-on: a small consumable, a bundle partner, or a size upgrade. Repeating a threshold message the shopper has already cleared makes the whole feed feel automated.

Well above your average order value, the best move is often no recommendation at all. Confirm what they have, mention return or warranty terms if that is a real concern in your category, and let them check out.

Exclude gift cards

Gift cards break most cart-value logic. They inflate the cart total without indicating anything about what the shopper actually wants, they usually do not need shipping, and add-on suggestions attached to them read as noise. Exclude gift cards from threshold calculations and from recommendation triggers. The same goes for anything else in your catalogue that behaves oddly — deposits, warranties, donation line items, and shipping protection products.

If you sell subscriptions, treat that as its own condition. A subscription in the cart is a targeting signal about commitment, not a channel. The SKIO integration for subscription stores lets you target based on subscription state; it is not somewhere contacts get synced.

Returning versus first visit

A first-time visitor has not earned a personalised recommendation yet, and pretending otherwise is obvious. Better to show what the store is known for: the bestseller, the current campaign, the size guide if fit is the usual objection.

A returning shopper is different. They have browsing history on your store, so picking up where they left off — the product they viewed, the collection they spent time in, the item that has come back in stock since — is both more useful and less intrusive than a generic offer. This is also where restock and price-drop messages belong, because the shopper has already shown which items they care about.

Let them add to cart from the message

A recommendation the shopper has to act on somewhere else is mostly a banner. If the message says a product goes well with what they are looking at, the add to cart should be in the message itself, in the feed, without a page change. That is the difference between a suggestion and an ad they dismiss.

The feed model matters here too. Because the shopper opens the bell, messages wait rather than being thrown in front of them, so you can run more of them without the fatigue an overlay creates. A dismissed banner is gone; an unread message stays until the shopper is ready.

A reasonable starting set

Four rules are enough to begin: complementary item on product pages, threshold-closing item below free shipping, bestseller on collection pages, and restock or continue-where-you-left-off for returning visitors. Exclude gift cards from all of them.

Give each one a couple of weeks, then compare add-to-cart from the feed against how often the message was opened at all — that ratio tells you whether the recommendation was relevant or just present.

You can build all of this alongside the popup you already run. If you want to see how it works in practice, Atomato is on the Shopify App Store, and there is more detail at atomato.co.