A shopper buys a laptop. At checkout the site offers a sleeve that fits it. If she adds it, that is a cross-sell. The sleeve costs the retailer almost nothing to offer, but it only works when it fits the laptop and the moment.
What is cross-selling?
Cross-selling is getting a customer to buy different, related items in addition to what they are already buying. Zapier says the items do not need to be near-identical, but they do need to be complementary (Zapier). A sleeve, a mouse or a warranty fits a laptop. A second laptop does not.
The term sits beside upselling, which moves the buyer to a bigger or better version of the same item. Cross-selling widens the basket, and upselling lifts the price of one line in it. Programs use both, and they measure them with different numbers. Our entry on upselling covers the second.
How does cross-selling work?
A cross-sell runs on three choices: which product, which customer and which moment. Amazon’s engineers pick the product with item-to-item collaborative filtering, which builds recommendations in real time (Linden, Smith and York, 2003). The paper says the store changes with each customer’s interests. It shows programming titles to a software engineer and baby toys to a new mother. A shelf built for one person sells more than a shelf built for everyone. Our entry on the recommendation engine covers the tools.
The moment matters as much as the product. Li and colleagues built a model for how to introduce the right product to the right customer at the right time, on the right channel. They tested it on a national bank’s data. Households differed in how they responded. The offers also helped households move along the financial path. The educational effect was 83%, the advertising effect 15% and the instant promotional effect 2% (Li, Sun and Montgomery, 2011). Timing is the lever.
A bundle is a fixed cross-sell. Two or three items are sold together at one price, and the shopper takes the set or leaves it. A recommendation is a looser cross-sell, since the shopper picks any item from a list. Each offer shown also takes up space and attention, so count the cost of the slot as well as the sale.
Banks show the long form. A household with a checking account often adds a card, a loan or a savings product later. The best next offer depends on where the household is on that path. That is the stage idea behind the bank model, and it applies to any program with a product ladder.
| Tactic | When it runs | Example | What a loyalty team measures |
|---|---|---|---|
| Product page recommendation | While browsing | “Customers also bought” | Click and add-to-basket rate |
| Checkout add-on | Before payment | A sleeve with a laptop | Attach rate and basket value |
| Post-purchase message | Days after the order | An accessory email to buyers | Second-purchase rate by member |
| Pickup or visit offer | When the member is in the store | One offer at pickup | Add-on spend per pickup |
| Cross-brand offer | Across a group of brands | A hotel guest offered a restaurant credit | Members who shop two brands |
Does cross-selling raise sales?
Evidence exists, though rarely as a single clean number. Gallino and Moreno studied a buy online, pick up in store rollout at one retailer. Store sales and traffic rose. Part of the gain came from customers who used pickup and bought extra products in the store, a cross-selling effect (Gallino and Moreno, 2014).
Retailers state the goal in their filings. Ulta Beauty says its marketing works to win and keep loyalty members, improve retention, raise shopping frequency and raise spend per member (Ulta Beauty 10-K). Kroger says its data science lets it use customer data to build personal experiences (Kroger 10-K). Neither filing gives a cross-sell lift, so treat any single figure you read elsewhere with care.
Measure it with an attach rate: the share of orders that include the add-on. Compare the rate for members who saw the offer with a holdout who did not. The gap is the lift. Then check basket value, margin and returns, because an add-on that comes back is not a sale.
Set up the test before the campaign. Pick the unit to randomize, either the order or the member. Run it for at least a full purchase cycle. Write down the attach rate, the margin and the return rate before you look at results. A test that starts with the metrics fixed is harder to bend after the fact.
What are the risks and rules?
Cross-selling goes wrong when the add-on is not wanted. The FTC’s 2022 staff report on dark patterns lists “sneak-into-basket,” which adds items to the cart without the shopper’s permission. It also lists tricking shoppers into buying unwanted items with a pre-checked box (FTC staff report, 2022). An offer should be a choice the shopper makes. None of this is legal advice.
Privacy law adds a second limit. Under the GDPR, a person can object to direct marketing at any time. The data shall then no longer be processed for that purpose (GDPR Article 21). A cross-sell program needs a working opt-out.
The same rule protects the program. A member who finds an unwanted item on an order learns to distrust offers. Keep add-ons clear and unchecked. Make them easy to remove. Count removals as a signal.
How do loyalty programs use cross-selling?
A member record is what makes a cross-sell relevant. The program knows what the member bought, when and where, so it can pick the next product. Our entries on offer engines and average order value cover the tool and the metric.
Multi-brand programs cross-sell between brands. SHARE Rewards runs on GRAVTY for Majid Al Futtaim. It spans malls, retail, restaurants, hotels and leisure, with 15M+ members. Loyalty Juggernaut reports a 35% rise in cross-brand engagement and 4.5M individualized offers a month (Loyalty Juggernaut).
A program can also pay for the add-on. A bonus on the related item turns the offer into a reward. Our entry on bonus points events covers how to run one, and points accelerators covers the rule behind it.
Four rules keep a cross-sell program honest. Offer only items that fit the order. Show one offer at a time. Skip members who opted out. Stop offering an item a member has declined several times.
For illustration, a program shows one add-on at checkout on 100,000 orders a month. A total of 8% of orders take it at $14. That is 8,000 add-ons and $112,000. Some of those buyers would have bought the item anyway. Hold out a random 5% of orders from the add-on, and compare their basket value with the rest to find the true lift. Report the holdout size beside the lift, and price the add-on’s margin against any discount.
How GRAVTY handles cross-selling
On lji.io, one GRAVTY agentic system selects the next-best offer, benefit or experience for each member. It acts within rules the enterprise sets (Loyalty Juggernaut). The same member shows up in every retailer, banner and channel. A cross-brand offer needs that base (GRAVTY for retail).
Frequently asked questions
What is the difference between cross-selling and upselling?
Cross-selling adds a related item. Upselling upgrades the item already in the basket. Zapier calls cross-selling getting customers to buy different, related items on top of their purchase (Zapier). Programs measure the first by basket size and the second by price per item.
What are examples of cross-selling?
A sleeve with a laptop, a case with a phone and a warranty with an appliance are common examples. Amazon builds such links from what customers who buy one item tend to buy next (Linden, Smith and York, 2003). Pickup adds a store example, since customers who collect an order sometimes buy more items (Gallino and Moreno, 2014).
When is the best time to cross-sell?
It depends on the customer, and a model can find the time. In a national bank’s data, households differed in how they responded to cross-selling offers, and the educational effect was 83% of the total (Li, Sun and Montgomery, 2011). Test timing against a holdout.
Is cross-selling the same as a recommendation engine?
No. A recommendation engine picks products. Cross-selling is the sales act that uses its picks. Amazon says its item-to-item method makes recommendations in real time (Linden, Smith and York, 2003). See our entry on the recommendation engine.
Can cross-selling be unfair to customers?
Yes, when it adds items without a clear choice. The FTC names two dark patterns: sneak-into-basket, and pre-checked boxes that trick shoppers into unwanted items (FTC staff report, 2022). Make every add-on optional and easy to remove.
