Cross-selling and upselling: how to sell more to the same customer without turning into a pushy salesperson

Acquiring a new customer for an online store costs, depending on the sector, somewhere between five and seven times more than getting a customer who's already buying to add something more to their order. And yet plenty of stores pour their entire budget into attracting new traffic and put almost nothing into making better use of the traffic already sitting in the cart. That's where cross-selling and upselling come in: two age-old retail techniques, adapted to a product page and a digital cart.

Cross-selling: "this pairs with what you're already buying"

Cross-selling means offering a complementary product to whatever the person is already buying or has already bought. The classic example everyone recognises: you buy a phone and get offered a case, a screen protector, and headphones. None of those products replace the phone; they all complement it. The key to good cross-selling is that the suggested product has a logical, obvious relationship to what's already being bought, not that it's simply "another product in the store we'd also like to sell."

Upselling: "for a bit more, you get the better version"

Upselling means offering a superior version of the exact product someone is looking at, usually right before they decide which one to buy. Typical examples: the model with more storage, the same product in a higher-grade material, an extended warranty, or the annual subscription plan instead of the monthly one at a small discount. Unlike cross-selling, this doesn't add a new product to the cart: it swaps out the initial choice for a higher-value one.

Where to place them without being invasive

On the product page

This is the natural home for upselling: right below the price or purchase options, a block like "also available as" showing the higher-tier version, clearly displaying the price difference and the extra benefit (more capacity, better material, warranty included). For cross-selling, a block further down like "pairs well with" or "customers also bought" works better, because it doesn't interfere with the main purchase decision.

In the cart

The cart is prime territory for low-cost cross-selling: cheap, relevant add-ons (batteries, a charger, a maintenance product) right before checkout. Aggressive upselling doesn't work well here: the person has already decided what they want to buy, and trying to change their mind at the last step tends to generate resistance. The cart is for adding on, not for making someone second-guess a decision they've already made.

In the post-purchase email

A few days or weeks after the purchase (depending on the product), this is a great moment for cross-selling, almost never for upselling. If someone bought a coffee machine, an email saying "customers who bought this machine also bought these capsules or this descaler" makes sense and adds real value; trying to sell them a pricier coffee machine when they already have one at home doesn't.

The golden rule: real relevance, not random products

The mistake that sinks any cross-selling or upselling strategy is showing generic suggestions with no logical connection to what's being bought. A "recommended products" block that shows the same items no matter what's in the cart signals that the store doesn't care what you actually need, it just wants to sell something extra. The best-performing suggestions are almost always based on real joint-purchase data (what people typically buy alongside this product) or on a complementary relationship that's obvious to anyone, not on arbitrary rules set by hand without checking whether they make sense.

The line before it feels pushy

There's a clear line between helping and hounding. Some practical rules for staying on the right side of it:

• No more than three or four suggestions at once; a long list of "you might also like" overwhelms and gets ignored entirely.
• Never interrupt checkout with a last-minute pop-up offering something extra: it's the moment of highest psychological friction, and any obstacle there increases cart abandonment.
• The suggested upsell's price should stay within a reasonable range of the original product (as a common industry benchmark, not much more than 25-30% above it), because too big a price jump breaks the conversation and creates distrust rather than interest.
• The cross-sell suggested in the cart should be low-cost in absolute terms: adding $4 or $6 for an add-on gets decided in two seconds; adding $40 requires stopping to think, and that pause is friction that can end in abandoning the whole order, not just the extra.

How to know if it's working: average order value

The metric to watch is average order value (AOV): how much each customer spends per purchase on average. If your cross-selling and upselling are working, that number should climb steadily over the months, without an increase in traffic or in product prices. It's one of the cleanest ways to measure the real impact of these techniques, separated from the effect of more traffic or price increases.

Bundles: a variant that blends both techniques

There's a third approach that mixes cross-selling and upselling into a single move: the bundle, where a set of complementary products gets offered at a price slightly lower than buying them separately. For example, a camera together with a memory card and a case, with a 10% discount off the combined individual prices. This works especially well because it solves, all at once, the question of which add-ons someone needs (you've already chosen and justified them) and it also includes a clear price incentive, something plain cross-selling doesn't always offer. The risk is the same as with any low-relevance suggestion: a bundle of products with no logical connection to each other reads as an attempt to clear out the warehouse, not as genuine help.

When to automate and when to decide by hand

For large catalogues, automating suggestions with real joint-purchase data is nearly mandatory, since doing it by hand would be impossible to maintain. But for small catalogues, or for products with an especially high margin the business wants to push, manually deciding what gets suggested alongside what (instead of leaving it entirely to a generic algorithm) usually delivers better results, precisely because whoever knows the business understands complementary nuances that an automated system, based purely on purchase statistics, doesn't always catch.

Frequently asked questions