--- title: "Frequently bought together: how to turn co-purchases into recommendations" date: "2026-10-24" author: "Rafael Gijon" canonical: "https://crossup.ai/en/blog/frequently-bought-together-product-pairs" locale: "en" --- # Frequently bought together: how to turn co-purchases into recommendations ## In short "Frequently bought together" means products that show up again and again in the same orders in your store. It is a valuable signal for recommendations and bundles, but on its own it does not prove they complement each other. SalesPilot, CrossUp's AI, crosses it with your catalog, stock, the moment of the purchase and your rules before showing anything on Tiendanube. If the coffee maker and the paper filters keep selling together week after week, your customers are telling you what goes together. CrossUp turns that pattern into sales: SalesPilot, CrossUp's AI, spots which products repeat in the orders of your Tiendanube store and uses them to suggest the right complement on the product page, at add to cart or at checkout. This guide explains how to read that signal, when to trust it and when to turn it into a bundle. ## What does "frequently bought together" mean? "Frequently bought together" (or co-purchase) means that two products show up again and again in the same orders. It isn't about your store's best-sellers; it's about what sells *with* one specific product. A few illustrative examples of what tends to show up in different verticals: - **Pets:** dry food with treats; cat litter with a scoop. - **Beauty:** serum with eye cream; sunscreen with after-sun. - **Home:** coffee maker with filters; a sheet set with pillowcases. - **Technology:** a phone with a case and a screen protector. - **Apparel:** a bikini with a sarong; a jacket with a beanie. The signal is built from orders, not people: it looks at which products share an order, without tracking anyone. And it is only as good as your own store: the pairs that work in one decor store are not the ones that work in another, even if they sell similar products. That's why it pays to read your own orders instead of a generic list of "products that go together". ## How is a co-purchase relationship found? The first step is counting: in how many orders do both products appear together? But counting isn't enough, for two reasons. **First: one coincidence is not a pattern.** If two products appeared together in a single order, it may be chance. The signal starts to matter when it repeats. That's exactly what SalesPilot requires: to use co-purchase as evidence, the pair has to repeat in your orders, and the relationship is recalculated periodically with new orders. **Second: popular products show up with everything.** Picture a beauty store with 1,000 orders in a quarter (the numbers are illustrative). Makeup remover is in 400 of those orders: 4 out of every 10. If you look at the 100 orders that include a serum and see that 45 also have makeup remover, it looks like a strong relationship. But makeup remover is already in 4 out of 10 orders storewide; showing up in 4.5 out of 10 serum orders says almost nothing. Eye cream, on the other hand, appears in only 50 orders overall (5 out of 100) and in 30 of the 100 serum orders. With serum, it shows up six times more often than usual. That is the relationship worth using. The useful question isn't "how many times do they appear together?" but "how much more often than usual do they appear together?". It's the same logic to apply when you review your orders by hand. ## Does being bought together mean they are complementary? Not necessarily. Co-purchase is evidence of affinity, not a guarantee that one product completes the other. Some pairs are bought together for reasons that don't help you recommend: - **A promotion.** If you put two products on sale the same week, they will show up together even if they have nothing in common. - **A gift.** Around holidays, one shopper builds orders for several people and mixes unrelated products. - **Size or color doubts.** In apparel and footwear, people often buy the same model in two sizes and return one. They are bought together, but they are variants, not complements. - **Popularity.** As in the makeup remover example, a best-seller shows up with almost everything. That's why SalesPilot doesn't recommend on co-purchase alone. It crosses that signal with what it knows about your catalog, your stock, the moment of the purchase and your rules. That way of connecting signals is what we call [your store's product map](/en/blog/product-map-recommendations): there, co-purchase is one more relationship, with its own strength and context, next to complements, alternatives and variants. ## How does SalesPilot use this signal? In practice, co-purchase feeds [SalesPilot](/en/salespilot) recommendations in three steps: 1. **It builds candidates.** On the product page and at add to cart, it starts from what is bought together with that product. At checkout, it pools what is bought with everything already in the cart, without repeating what the shopper picked. 2. **It filters.** It drops anything unpublished or out of stock, anything the shopper already has, anything priced out of context and anything you excluded for that moment. 3. **It ranks.** It puts the pairs with the strongest relationship in your orders first and, on the product page, it can add similar alternatives when that signal is available. An illustrative example from a coffee store: someone adds a drip coffee maker to the cart. In the store's orders, that coffee maker often appears with paper filters, ground coffee and a thermal carafe. The filters are in stock and cost a fraction of the coffee maker; the carafe is sold out. SalesPilot offers the filters. The relationship with the carafe exists, but today it can't be sold. This kind of signal is the basis for showing [related products](/en/related-products) that make sense at each moment. If you want to see how those moments play out on Tiendanube, we cover it in the [practical guide to related products](/en/blog/related-products-tiendanube). ## Which signals should you use, and which not? | Signal | What it shows | What it doesn't prove | Possible use | Merchant control | |---|---|---|---|---| | Two products that repeat together in many orders | Real affinity in your store | That they are used together: it could be a promo or a gift | Candidate for the product page, add to cart and checkout | Exclude products or categories at each moment | | Two products together in one or two orders | That someone bought them together | A pattern | None yet: wait for more orders | Nothing to do | | A product that shows up in almost every order | That it's very popular | Affinity with any specific product | General suggestion, with care | Exclude it if it crowds out the rest | | The same model in two sizes or colors | Doubts about size or color | That one complements the other | None as a complement: it's a variant | Improve the size guide on the product page | | A pair you pin | Your commercial judgment | That it will convert | Manual suggestion on the product page or at add to cart | Full: triggers, recommended items and dates | | A bundle you publish | An offer you put together | That the discount is right | Priority offer on the product page and at add to cart | Products, discount, placement and dates | ## When should you create a bundle? When a pair repeats, makes sense to use together and leaves you a good margin, it can stop being a loose recommendation and become an offer: a bundle. The difference is that you decide the bundle. In CrossUp you pick 2 to 6 products, choose a fixed, progressive or no discount, where it appears (the product page, add to cart or your store's home page) and for how long. While it's active, the bundle takes priority: if the product a shopper is viewing belongs to a published bundle, CrossUp shows that bundle first on the product page and at add to cart. If you don't want to start from scratch, you can ask SalesPilot in one sentence, for example "build a coffee kit with the coffee maker, filters and a mug". The AI drafts the proposal from your catalog, and you review it, edit it and decide whether to publish it. We walk through it in [bundles and kits with CrossUp](/en/bundles) and in [how to build bundles people actually buy](/en/blog/combos-que-funcionan). One useful distinction: if what repeats in your orders is the *same* product (three packs of the same coffee, two units of the same cream), that isn't a bundle, it's a quantity pack. A [volume discount](/en/blog/volume-discounts-2-for-1-3-for-2-packs) such as a 3-for-2 fits better than a bundle of different products. ## How do you turn a co-purchase into a recommendation, step by step? 1. **Spot the pattern.** If you turn on SalesPilot, it does this for you with your orders. If you want to check by hand, look at what repeats in the orders of your best-selling products. 2. **Check the commercial sense.** Ask whether one product completes the other, or whether they appear together because of a promo, a gift or a size doubt. 3. **Check stock and price.** A complement that is out of stock or pricier than the main product won't work at add to cart. 4. **Pick the moment.** Direct complements at add to cart; refills on the thank-you page. 5. **Turn it into a rule or a bundle.** Pin the pair as a manual suggestion if you want to lock it in, or build a bundle if you want to give it its own offer. 6. **Measure.** Track how many times it was shown, how many times it was accepted and how much revenue was attributed. ## How do you measure the result? In four layers: how many times the recommendation was shown, how many times someone accepted it, how many of those acceptances ended in a purchase and how much value was attributed. CrossUp attributes order by order: it only counts the value of products added from the offer and actually bought. The product the shopper had already chosen and rejected recommendations stay out. With that measurement you can see which pairs really convert in your store and which ones to exclude or replace with a bundle. For examples by vertical, see how it applies to [food and beverage](/en/solutions-food-beverage) and [home and decor](/en/solutions-home-decor). Use your purchase patterns to recommend with SalesPilot: [install CrossUp on your Tiendanube store](/en/tiendanube) with a 15-day free trial and let your orders tell you what goes together. ## Frequently asked questions ### What are frequently bought together products? Products that repeatedly appear together in your store's orders. They can reveal an affinity that is useful for recommendations or bundles. ### Is every co-purchase useful for recommendations? No. A promotion, a gift or a size doubt can also put products in the same order, so SalesPilot crosses co-purchase with catalog, stock, moment and your rules. ### Does CrossUp track my customers to calculate co-purchases? No. It uses aggregate patterns from your store's orders: which products share the same order. It does not track people across sites or sell data. ### Can I pin a pair manually? Yes. On the product page and at add to cart you can create manual suggestions that work alongside SalesPilot. ### Can a co-purchase become a bundle? Yes. You can build the bundle by hand or ask SalesPilot for it; either way, you review products, discount and placement before publishing. --- [Home](https://crossup.ai/en) ยท HTML: https://crossup.ai/en/blog/frequently-bought-together-product-pairs