Do upsells work at checkout?

There are two very different things people call a checkout upsell, and they carry opposite risks.

The first is an offer shown before payment, in the cart or on the checkout page. It can lift order value, and it can also add a decision at the exact moment someone was about to commit. Every extra choice at that step is a chance to hesitate, open a new tab and never come back.

The second is an offer shown after payment has been taken, on the way to the thank you page. The order is already yours. If the shopper says no, nothing is lost. If they say yes, the item is added to the same order, usually without re-entering card details, and it ships in the same parcel. On Shopify this runs as a post-purchase step between payment and the order confirmation.

So the honest answer is that post-purchase offers work reliably and pre-payment offers work when handled carefully. If you are starting, start after payment, measure it, and only then experiment earlier in the flow.

Relevance does more work than the discount. A shopper who has just bought a coffee grinder will consider a bag of beans, a cleaning brush or a set of scales. The same shopper offered a random bestseller from another category will decline without reading it.

Build the rules from your own basket data rather than intuition. Export a year of orders and find the pairs that already appear together more often than chance. Those pairs are your offer list, and they are usually not the products you would have guessed.

Price matters as much as relevance. The offer should feel small next to the order that was just placed. A modest accessory after a large purchase converts far better than an item of similar value, because the shopper is comparing it to what they just spent rather than to nothing.

And it has to be one offer. A single item with a clear reason to add it outperforms a grid. If your tool shows three options by default, turn two of them off and see what happens.

Write the reason in a line of plain text. What it does, why it goes with what they bought, and what it costs added to this order.

Most failures are mechanical rather than strategic, and they repeat across stores.

The offer interrupts payment. A popup over the payment form, or a step that must be dismissed before continuing, converts a small gain into a real loss. Never put anything between a shopper and the pay button.

The offer restarts the transaction. If accepting means re-entering card details or going back through checkout, most people abandon the upsell and some abandon the order. This is the main reason post-purchase offers outperform.

The offer breaks the shipping promise. Adding an item that ships separately, or that pushes the order into a different delivery window, produces support tickets and refunds that wipe out the extra margin.

The offer is a discount on something they already bought. Showing a cheaper price for the item in the cart teaches customers to wait, and it generates immediate requests for a price adjustment.

And the offer is too slow. A post-purchase step that takes several seconds to load gets skipped, because people are already reading the confirmation in their head.

Fix these five and the offer itself usually starts working.

An upsell that your warehouse hates will be turned off within a month, so decide the practicalities before you launch.

Make sure accepted items merge into the original order rather than creating a second one. Two orders means two picks, two labels and often two delivery charges you absorbed for one customer.

Check stock in real time. Offering an item that is out of stock produces a cancellation on an order that had already gone through, which is worse than never offering it.

Set a limit on how often a returning customer sees the same offer. Someone who declined the same accessory four times should be shown something else or nothing.

Decide what happens with returns. If the customer returns the main item, what happens to the accessory bought alongside it. Write the rule down before support has to invent one.

And keep the thank you page useful. The post-purchase step should not replace order confirmation details. People want the order number, the delivery date and the confirmation email, and burying those to sell one more item costs trust for very little.

Two numbers tell you whether it is working. Take rate is the share of shoppers shown the offer who accept it. Incremental gross profit per order is what it actually earned you after the discount, the extra shipping and the returns.

Take rate on its own is misleading, because a large discount will lift it while destroying the margin. Track profit, not acceptance.

For pre-payment offers you also need a control. Run the offer for half your traffic and not the other half, and compare completed orders as well as order value. If order value rose and completion fell, work out which is bigger before celebrating.

For post-purchase offers a control matters less, since the order is already captured, but you should still compare return rates on orders with and without an accepted upsell. Items added on impulse come back more often, and a high return rate can turn a good take rate into a loss.

Review the offer list every quarter. The pairs that sell together change with the season and with your catalogue, and a rule set built eighteen months ago is usually recommending last year's products.