Upselling and Cross-Selling: The Sells That Quietly Grow Order Value

TL;DR

Upselling means getting a shopper to trade up to a better or larger version of what they are already buying. Cross-selling means adding a complementary item to the order. Upselling is about value, cross-selling is about volume, and both do the same job: they raise the average order value without costing you a fresh acquisition.

That last point is the whole reason they matter. You have already paid to bring the customer in. Every extra pound they spend on the way out arrives at a much better margin than the first pound did, because there is no ad cost attached to it.

But the profit from an upsell is quieter and more fragile than most stores assume. The offer that raises the order value can also raise your returns, add a second shipment, and, in some cases, cost you the customer's next order. The take rate your upsell app reports is not the number that matters. What matters is the net, after returns, shipping, discounts, and any effect on the next purchase.

The single most useful thing to understand is timing. An offer made after the purchase is confirmed captures extra revenue without risking the original sale. An offer pushed too hard before checkout can cost you the whole order. The rest of this guide covers the difference between the two, where each works, the costs almost nobody prices in, and the point where selling more quietly starts losing money.

Upselling vs Cross-Selling, in Plain Terms

The two terms get mixed up constantly, so here is the clean version.

Upselling is trading up. The shopper is looking at a product, and you offer a better one: a larger size, a premium tier, a bundle of three instead of one, the model with more features. They were going to spend, and you help them spend a bit more on a better version. Upselling is about value.

Cross-selling is adding on. The shopper has chosen a product, and you offer something that goes with it: the case for the phone, the beans for the grinder, the socks for the shoes. You are not changing their choice, you are completing it. Cross-selling is about volume.

A simple way to remember it: upselling makes the one thing bigger, cross-selling adds a second thing. The coffee brand that offers a bigger bag is upselling. The same brand offering a milk frother alongside the bag is cross-selling.

The distinction looks academic, but it matters more than most guides admit, because the two behave differently once the money has changed hands. A cross-sell adds a separate item that the customer chose to want, so it tends to stick. An upsell nudges the customer past the tier they had settled on, which means a share of upsell buyers wake up the next morning having spent more than they meant to. That difference shows up later in returns, in refunds, and in whether they come back. We will come back to it, because almost nobody does.

The Number Everyone Quotes, and What It Really Says

If you read anything about recommendations, you will meet this claim: 35 percent of everything Amazon sells comes from product recommendations. It is quoted everywhere, usually to prove that upselling and cross-selling are enormous.

It is worth knowing where that number actually comes from, because it is shakier than its fame suggests. The figure traces back to a single sentence in a 2013 McKinsey article, which stated it with no source, no dataset, and no footnote. Amazon has never published or confirmed it. Every version you see today, more than a decade later, is ultimately repeating that one unsourced line.

This does not mean recommendations do not work. They plainly do. It means the specific 35 percent figure is folklore, not measurement, and you should not build a business case on it. The more useful and better-sourced version of the same idea comes from McKinsey's later research, which found that companies that excel at personalization generate 40 percent more revenue from those activities than average performers. That is a real, methodologically grounded finding, and it points the same direction: relevant recommendations earn more.

The lesson to carry forward is not the number. It is the habit. Treat the famous statistics with a little suspicion, and judge your own upselling by what it actually adds to your average order value, net of everything, not by what worked for Amazon. That habit of suspicion is exactly what the rest of this guide applies to your own upsell reporting, which is more misleading than the Amazon figure.

Where in the Journey Each One Works

An upsell or cross-sell can appear at several points, and each spot has a different character.

On the product page. This is where upselling fits most naturally. A shopper reading about the standard model can be shown the premium one, or a "buy two, save" option. They are still deciding, so a nudge toward a better version rarely feels pushy.

In the cart. A good place for a light cross-sell: the small complementary item that completes the order. This is the classic "add gift wrap" or "customers also added" moment. Keep it modest, because the shopper is on their way to checkout and you do not want to stall them.

At checkout. Handle with care. The customer is committing, and anything that adds a decision here can cost you the sale. A single, low-friction add-on can work. A wall of offers does not.

After the purchase, on the confirmation or thank-you page. This is the quietly powerful one, and it gets its own section next.

The general rule that emerges is simple: the closer you are to the moment of commitment, the lighter your touch should be, right up until the purchase is done, at which point you can be bolder again because there is nothing left to lose on that order.

Why Post-Purchase Is the Safest Place to Offer

There is a reason experienced stores put their strongest offers after the checkout, on the thank-you page.

The logic is clean: the sale is already closed. The customer has paid. A post-purchase offer captures extra revenue at zero risk to the order you just secured, because that order cannot be un-placed. If the shopper ignores the offer, you still have the original sale. If they take it, you have added margin with no acquisition cost attached.

Compare that to a pre-checkout upsell, which is always a gamble. Push too hard before payment and you can introduce doubt, add friction, and lose not just the upsell but the entire cart. The post-purchase moment removes that downside entirely.

The best-executed one-click upsells, where the customer can accept the extra item with a single tap and without re-entering payment details, are where this really pays. Re-entering card details is exactly the kind of friction that kills a post-purchase offer, so the smoothest ones feel like a natural continuation of the order rather than a new transaction.

This is also why upselling belongs in the same family as everything else that happens after the order ships. The post-purchase window is not just for the thank-you note. It is a genuine revenue moment. But "safest" is not the same as "free," and the two sections after the example pull apart costs that the post-purchase upsell quietly carries.

A Real Example: Turning the Thank-You Page Into Revenue

The clearest way to see this work is a single store.

Hard Hat Watches, a Shopify store, built a short post-purchase upsell funnel: after a customer completed their order, they were offered a related item they could add with one click, no re-entered payment, no return trip to checkout. The offer sat entirely after the sale was already secured.

The results, as reported through the store's upsell app, were striking. That funnel converted at 30 percent across more than 10,000 views and generated 311,000 dollars in upsell revenue in its first ten months, all from the post-purchase moment alone. Five months later, that figure had roughly doubled to 620,000 dollars.

Two things are worth drawing out. First, none of that revenue put the original orders at risk, because every offer came after checkout. Second, a 30 percent take rate means nearly one in three buyers accepted a well-matched offer at the right moment, which tells you the appetite is real when the timing and relevance are right.

One honest caveat, and it is the doorway into the rest of this guide: this figure is the gross revenue the upsell app attributed to itself, reported by the store through that software, and not independently audited. That is not the same as the profit the store actually kept, and it is not even the same as the revenue the upsell truly created. Understanding the gap between those three numbers is where most of the real money in upselling is won or lost, and it is the part the internet almost never discusses.

The Attribution Trap: Why Your Upsell App Overstates Its Own Value

Here is the uncomfortable mechanic behind almost every upsell success story, including the one above.

An upsell app counts, as its own revenue, every order that included an item it showed. That sounds fair until you ask the obvious question: how many of those customers would have bought that item anyway? Some meaningful share of people who accept a cross-sell were already going to add it, or would have come back for it next week. When the app claims that revenue, it is taking credit for sales that would have happened without it. The technical name for this is a lack of incrementality, and it is the single biggest reason upsell numbers look better on the dashboard than in the bank.

Think about what "attributed revenue" actually contains. It contains three different things blended into one figure: revenue the offer genuinely created that would not otherwise exist, revenue that was simply pulled forward from a future order the customer would have placed anyway, and revenue from customers who would have bought that exact item in the same session regardless. Only the first of those three is real, incremental profit. The other two are the app congratulating itself for weather it did not make.

The way serious operators cut through this is a holdout test. You withhold the upsell from a random slice of traffic, say ten percent, and you compare the total revenue per session of the group that saw the offer against the group that did not. Not the take rate. Not the attributed revenue. The total revenue per visitor across the whole group, including the people who ignored the offer and the people who abandoned because of it. The difference between the two groups is the offer's true incremental contribution. It is almost always smaller than the attributed figure, and occasionally it is negative, which means the app was reporting six-figure "revenue" while quietly costing the store money.

Very few stores ever run this test, because the app does not encourage it and the attributed number feels good. But it is the only honest measure of whether an upsell is working. If you take one non-obvious idea from this guide, make it this: your upsell is worth the holdout difference, not the dashboard number. Everything that follows is downstream of that distinction.

The Margin Math Almost Nobody Actually Runs

Attribution is the first correction. The second is margin, and it is arithmetic almost nobody does before switching an upsell on.

Take a worked example. The numbers here are illustrative, chosen to make the mechanism visible rather than to represent any real store, but the structure is exactly the calculation you should run on your own figures.

Suppose your post-purchase upsell offers a 40 dollar add-on. The app reports a 20 percent take rate, and it looks like a triumph. Now walk the single accepted offer all the way through the business.

The offer that looked like it added 40 dollars added about three dollars and forty cents once the real costs are counted, and that is before the attribution haircut from the previous section. Apply even a modest incrementality discount, say a third of those accepts would have bought anyway, and the true figure is lower still.

None of this means the upsell is bad. Three dollars and forty cents of near-zero-acquisition-cost margin, multiplied across thousands of orders, is real money and worth having. The point is that the gap between the headline 40 dollars and the actual 3.40 is enormous, and every decision you make about discount depth, shipping, and which product to offer moves that final number far more than the take rate does. A store optimising the take rate is often optimising the wrong end of the equation. The lever that matters most in that table is usually shipping, which is why it gets its own section.

The Returns Asymmetry Nobody Prices In

Here is a distinction that upsell guides never draw: upsells and cross-sells get returned at different rates, for different reasons, and it changes which one you should push.

A cross-sell is an item the customer actively chose to add. They wanted the phone case; they still want it when it arrives. Its return rate is roughly the return rate of any deliberately chosen product. It behaves.

An upsell, especially a trade-up to a larger size or a premium tier, is different. A share of upsell buyers were nudged past the option they had actually settled on. In the moment, the bigger bundle or the premium version felt worth it. When the box arrives, some of them feel they overreached, and a trade-up is unusually easy to regret, because the customer has a clear mental anchor: the cheaper version they originally intended to buy. That regret converts into returns, refunds, and, worse, the quiet resolution not to shop with you again.

This asymmetry has a practical consequence that flips a common assumption. Conventional advice treats the high-value trade-up as the prize because it adds the most to the order. But once you net out the higher return rate, a stack of modest, genuinely useful cross-sells can out-earn a smaller number of aggressive trade-ups, even though the trade-ups win on gross order value. The cross-sells stick. The trade-ups partly bounce.

The way to manage it is to look at return rate by offer type and by offer, not just in aggregate. If a particular upsell shows a return rate meaningfully above your store average, it is not really adding what the dashboard says, and the difference is landing in your reverse-logistics costs, which for online orders already run several points higher than in-store returns. An upsell with a high return rate is a loan the customer pays back, not a sale.

The Hidden Fulfilment Cost of a Post-Purchase Upsell

This is the one that catches even careful operators, and it is almost never written about.

A post-purchase upsell happens after the original order is placed. Depending on how your warehouse and your app are wired, that accepted upsell can generate a separate order that ships as its own parcel. The original order is already picked, packed, or even out the door. The upsell item follows behind it in a second box.

A second box means a second set of costs: another pick, another pack, another shipping label, sometimes another set of packaging materials. On a low-priced add-on, that second shipment can cost more than the add-on's entire margin. You accepted the upsell, the customer is happy, the dashboard is green, and you lost money on the parcel.

The fix is not to abandon post-purchase upsells. It is to make sure the accepted upsell merges into the original shipment wherever it possibly can. The window matters: an upsell accepted in the first seconds on the thank-you page, before the warehouse has picked the order, can often be added to the same box at almost no extra cost. The same upsell accepted an hour later, after the parcel is sealed, is a separate shipment with separate economics. This is why the speed and placement of the offer are not just conversion questions, they are margin questions. The offer that fires instantly and merges is worth far more per accept than the identical offer that fires late and ships alone.

Very few stores audit this. It is worth pulling a sample of accepted upsells and simply asking: did this ship in the same box as the original order, or did it cost me a second parcel? The answer often rewrites the economics of the whole program.

Does an Upsell Cost You the Next Order?

The deepest and least-discussed question of all: what does an upsell do to the customer's next purchase?

Every upsell analysis stops at the order it appears in. But a customer relationship is a sequence of orders, not one transaction, and an offer that wins today can quietly cost you tomorrow in two ways.

The first is regret. A customer who was pushed into a bigger spend than they intended, and who later feels it, does not always return the item. Sometimes they just do not come back. That lost second order never shows up next to the upsell that caused it, because no dashboard connects the two. The upsell looks like a pure win; the missing repurchase looks like ordinary churn. They are the same event.

The second is anchoring, and it cuts the other way. A cross-sell that introduces the customer to a genuinely useful complementary product can raise the next order, because you have expanded what they buy from you. The customer who discovers your grinder because you cross-sold it with the beans now reorders both. That is an upsell paying a dividend on future orders, and it is just as invisible to the dashboard as the regret case.

So the real scorecard for an upsell is not its contribution to the order it sits in. It is its contribution to the customer's whole lifetime: this order, minus returns, plus or minus what it does to every order after. That is a harder number to measure, which is exactly why almost nobody measures it, and why so much upsell advice quietly optimises for a metric that can be actively misleading. The stores that get this right treat upselling as a retention decision, not a checkout decision, and they judge an offer by whether the customers who accepted it are worth more over the following year than the customers who did not.

What Makes an Offer Convert, and What Kills It

With all of that on the table, the practical patterns get clearer. The offers that work share a few traits, and the ones that fail break the same rules. Here is the rough picture of where different placements land.

Beyond placement, three things separate an offer that converts and keeps its profit from one that merely looks good.

Relevance. The offer has to make sense next to what the customer just bought. A phone case after a phone converts and sticks. A random discounted item they never looked at converts worse and comes back more. This is where good recommendation logic, whether powered by AI or a simple rule, earns its place: the closer the match, the higher the take rate and the lower the regret.

Restraint on the discount. A common instinct is to make the upsell cheap to force acceptance. The data points the other way, and so does the margin math above: modestly priced add-ons, ones that are a small fraction of the original order, convert well and preserve margin, while deep discounts buy acceptance you did not need and hand away profit. A small, relevant, fairly priced offer beats a big, aggressive, discounted one almost every time.

Fit with fulfilment. The best offer to make is one that can join the original shipment. If accepting the upsell means a second parcel, the offer has to clear a much higher margin bar to be worth it. Choose upsell items your warehouse can merge, and fire the offer fast enough that it still can.

When Selling More Loses Money

Pulling the failure modes together, here is where upselling and cross-selling turn negative.

The most cited warning comes from Harvard Business Review, whose research on the downside of cross-selling opens with a blunt line: companies work hard to persuade existing customers to buy more, and often that is a money-losing proposition. The argument is that a segment of customers, when pushed to buy across categories, become unprofitable through returns, service demands, and erratic buying. Selling more to the wrong customer can cost more than it earns.

Layer on the specific mechanisms from this guide and the picture sharpens. An upsell loses money when the attributed revenue was not incremental, so you paid to move a sale that was going to happen anyway. It loses money when the second shipment costs more than the add-on's margin. It loses money when the return rate on a trade-up quietly runs above your average. And it loses money when the extra spend today suppresses the customer's next order through regret. Any one of these can turn a green dashboard red, and none of them appear on the dashboard.

There is also the everyday failure of aggression. An offer that pushes too hard, or a checkout stacked with pop-ups, creates friction and fatigue. It is entirely possible to run an upsell that lifts acceptance on paper while raising cart abandonment by more than it adds, leaving you worse off overall. The acceptance rate looks good in isolation and the net result is negative, which is precisely why the holdout test matters.

None of this argues against upselling. It argues for measuring the right thing. The number that matters is not the take rate on the offer. It is whether total revenue per session, after abandonment, returns, discounts, shipping, and the effect on the next order, went up. Keep the offers relevant, keep them after the sale where you can, keep them modest, merge them into the original shipment, and test them against a holdout, and the risk mostly takes care of itself.

Why This Is Really a Retention Lever

Step back and the whole topic is a retention story wearing a sales-tactic costume.

The reason an upsell can be so profitable is the reason retention in general is so profitable: you are earning more from a customer you already have, without paying to acquire them again. Harvard Business Review's often-quoted figures make the scale clear. Acquiring a new customer costs five to twenty-five times more than keeping an existing one, and increasing retention by just five percent can lift profits by 25 to 95 percent, drawing on Bain & Company research.

An upsell or cross-sell, done well, is one of the cleanest ways to act on that. Every extra item a current customer buys raises their lifetime value at close to zero marginal acquisition cost. But the section above on the next order is the catch: an upsell only compounds into retention if the customer is glad they took it. Push the right add-on and you widen the relationship and lift future orders. Push the wrong one and you get a one-time bump followed by silence. That is why the stores that win at this treat upselling not as a checkout gimmick but as part of how they keep and grow customers over time, judged over the lifetime rather than the transaction.

Common Mistakes That Cost More Than They Earn

1. Trusting the attributed revenue. The app counts sales that would have happened anyway. Run a holdout test and judge the upsell by the incremental difference, not the dashboard total.
2. Skipping the margin walk. A 40 dollar upsell is not 40 dollars of profit. Subtract cost of goods, processing, shipping, discount, and returns before you celebrate.
3. Ignoring the second parcel. A post-purchase upsell that ships separately can cost more than it earns. Merge it into the original shipment or do not offer it.
4. Pushing trade-ups over add-ons blindly. Trade-ups win on order value and lose on returns. Watch return rate by offer type, not just in aggregate.
5. Upselling before the sale is secure. Pushing hard at checkout risks the whole order. Move your strongest offers to after the purchase.
6. Offering things that do not match. An irrelevant add-on converts worse, returns more, and dents trust. Relevance is the whole game.
7. Leaning on deep discounts. A modest, relevant offer beats a heavily discounted one, and the discount often just erodes margin you did not need to give up.
8. Measuring the take rate instead of the lifetime. An upsell that lifts acceptance but raises returns or suppresses the next order is a loss wearing a win's clothing.

A Simple Order to Build In

If you are adding upselling and cross-selling to a store, build it in this order. Most of the gain is in the first two steps, and most of the safety is in the last two.

First, the post-purchase one-click upsell. It earns the most and risks the least, because it sits after the sale is closed. Make it a single, relevant, one-tap offer with no payment re-entry, and wire it so accepted items merge into the original shipment.

Second, a light in-cart cross-sell. One small, complementary item that completes the order. Keep it modest so it never stalls the path to checkout.

Third, product-page upsells. Offer the better version or the multi-buy where the shopper is still deciding. Low risk, steady contribution.

Fourth, measure the net, not the take rate. Run a holdout, watch revenue per session, returns by offer type, and shipping cost per accepted upsell. Cut anything that lifts acceptance while raising abandonment or returns.

Fifth, only then get clever. Personalised recommendations, bundles, and follow-up offers add on top once the basics are earning and you can prove the earnings are incremental.

Most stores do this backwards, chasing clever recommendation engines before they have even turned on a post-purchase offer or run a single holdout. The boring first step is usually the biggest one, and the boring last step is what stops the whole program quietly losing money.

FAQ

What is the difference between upselling and cross-selling?
Upselling gets a customer to trade up to a better or larger version of what they are buying. Cross-selling adds a complementary item to the order. Upselling is about value, cross-selling is about volume, and both raise the average order value.

Is upselling or cross-selling better for ecommerce?
Neither is universally better; they do different jobs and carry different risks. Cross-sells tend to stick, because the customer chose to add them. Trade-up upsells add more to the order but get returned more often, so once you net out returns a stack of modest cross-sells can out-earn a few aggressive trade-ups.

When is the best time to upsell?
Usually right after the purchase is confirmed, on the order confirmation or thank-you page, and fast enough that the item can still be merged into the original shipment. At that point the sale is already secure, so an extra offer adds revenue without risking the original order.

Does the Amazon 35 percent recommendations statistic hold up?
Not really. It traces to a single unsourced sentence in a 2013 McKinsey article and has never been confirmed by Amazon. Recommendations clearly work, but that specific figure is folklore, not measured fact.

Why do people say upsell apps overstate their value?
Because the app counts every order containing an item it showed as its own revenue, including sales the customer would have made anyway. The honest measure is a holdout test: withhold the offer from a random slice of traffic and compare total revenue per visitor. The true, incremental value is almost always smaller than the attributed figure.

Can upselling actually lose money?
Yes, in several ways that never show on the dashboard: the revenue was not incremental, a separate shipment cost more than the add-on's margin, a trade-up got returned, or the extra spend suppressed the customer's next order. Judge an upsell by net contribution over the customer's lifetime, not by its acceptance rate.

How do I stop upsells from annoying customers?
Keep them relevant to what the customer is buying, keep them modest rather than heavily discounted, limit yourself to one clear offer at a time, and place the strongest ones after checkout so they never add friction to the sale.

Where to Go From Here

The thread running through all of this is that the easy number lies. The take rate looks good, the attributed revenue looks better, and the real profit is whatever survives returns, shipping, discounts, incrementality, and the effect on the next order. Measure that, and upselling quietly grows order value instead of just appearing to. For the highest-value orders, some stores take the same idea to the phone, where a quick call to confirm or complete an order is also a natural, low-pressure moment to offer the obvious add-on. Kovax handles that side for Shopify stores, on order confirmation and support calls.