Conversion & Analytics
Cross-Sell
Cross-sell is the practice of recommending related items alongside the shopper's current selection. A shopper buying a camera is offered a memory card and a case. Unlike upsell, cross-sell does not replace the original item; it complements it. AI cross-sell uses purchase patterns and product embeddings to choose which complement.
How it works
Cross-sell starts with a base item in cart or on the product page. The system looks for products that historically co-occur with that base in completed orders. The simplest version is "frequently bought together," which uses raw co-occurrence frequency. Modern systems layer on filters for stock, margin, and freshness, plus a re-ranker that boosts items the current shopper is likely to want.
The placements are predictable: a "frequently bought together" block on the product page, a "complete your order" row in cart, a post-purchase one-click offer, and an in-chat suggestion when the assistant detects intent. Each placement has a different acceptance rate. Product page cross-sells reach the largest audience but at low conversion; post-purchase one-click reaches a smaller audience at much higher conversion.
Good cross-sell rules also avoid the obvious traps. Do not recommend an item the shopper already has in cart. Do not recommend across incompatible variants (a phone case for a different phone). Do not stack five recommendations when one would do. Each rule sounds obvious until a launch produces oddly mismatched suggestions and someone has to explain why.
For example, a yoga mat in cart triggers strap, cleaning spray, and bag as a three-item complement. A second example: a coffee machine on the product page shows a descaler, a measuring scoop, and a starter bean pack with the reasons clearly labelled.
Why it matters for Shopify stores
Cross-sell is one of the most reliable AOV levers on a Shopify store. Unlike discounts, which compress margin, well-targeted cross-sells add line items without requiring a price cut. The accessories on a base purchase are often higher-margin than the base, which makes cross-sell economics compelling.
The pitfall is over-reliance. A storefront that drowns every product page in cross-sells trains shoppers to ignore the recommendations, which lowers their value over time. The right discipline is one strong cross-sell row per surface, with clear reasons for each suggestion. AI cross-sell helps choose the right row; it does not justify three rows.
Examples
- A coffee maker in cart triggers a "you might also need" row with a descaler, a starter bean pack, and a milk frother.
- A jeans purchase shows a belt and a matching shirt as a "complete the look" block on the cart page.
- A skincare cleanser shows the matching toner and moisturiser as a small bundle with a single-tap add-all option.
Related terms
AI Upsell
AI upsell is the use of a model to recommend a higher-priced or higher-margin variant of what the shopper is already considering. Unlike fixed upsell rules, an AI upsell picks the suggestion per shopper and per cart, which means it shows up only when there is a defensible reason to upgrade.
AI Product Recommendations
AI product recommendations are item suggestions chosen by a model based on a shopper's context: their query, their browsing, their past orders, and other shoppers' behaviour. They appear on home pages, product pages, carts, and inside chat, and they typically combine collaborative filtering with semantic search.
Shopping Cart AI
Shopping cart AI is software that reads a shopper's cart in real time and acts on it. Common actions include suggesting a complementary item, applying the right shipping or discount logic, recovering an abandonment, and forecasting the likelihood of conversion. It runs on top of the storefront, not in place of it.
Average Order Value (AOV)
Average order value (AOV) is the average amount a shopper spends per order, calculated as total revenue divided by number of orders over a period. It is one of the three primary levers in ecommerce alongside conversion rate and traffic. Common ways to lift AOV are upsells, cross-sells, bundles, and free-shipping thresholds.
Personalized Product Recommendations
Personalized product recommendations are suggestions tailored to one shopper based on their behaviour, profile, and context. Unlike generic best-seller lists, they change per visitor. They typically draw on browsing history, past orders, location, device, and any chat or survey signals the merchant has captured.
See it in action
Watch how Shop Me uses AI shopping assistance and conversation insights on a live Shopify-style store.
See Live DemoFAQ
How is cross-sell different from upsell?
Upsell replaces the original item with a higher-tier version (a 500ml bottle instead of 250ml). Cross-sell adds a complementary item to the original (the bottle plus a pump dispenser). Both raise AOV; they target different decisions and usually appear in different placements. A typical cart shows one of each, not both.
How many cross-sells should I show per product?
Three is a common default. Two is cleaner on small product pages and on mobile. More than four often becomes background noise. Test on your traffic; most stores find that going from three to five recommendations hurts click-through on the strongest one without raising overall add-to-cart rate.
Do AI cross-sells work for stores with small catalogs?
They can, but the gain over a hand-picked rule is small. A store with 20 SKUs and a few dozen orders a day can produce a strong "frequently bought together" map manually. Once your catalog grows past a few hundred SKUs and orders span many product combinations, AI cross-sell starts pulling ahead because the manual map cannot keep up.