Upsell and Cross-Sell — Do Shoppers Actually Use Your Recommendations?
Recommendation widgets are on nearly every store, and many are ignored or get in the way of checkout. Learn how to measure whether your upsells and cross-sells are seen, clicked, and bought, and where they help or hurt.
Short answer: a recommendation helps when it is relevant, placed where the shopper is still deciding, and easy to add. It hurts when it interrupts checkout or sends a ready buyer back to browsing. Measure each widget separately: how many shoppers see it, click it, and add to cart, and whether orders complete at the same rate with it present.
Table of Contents
- Upsell and Cross-Sell Defined
- What to Measure
- Placement and Its Risks
- What Recordings Show
- Making Recommendations Work
- Key Takeaways
Upsell and Cross-Sell Defined
- Upsell: a better or larger version of what the shopper is considering.
- Cross-sell: something that goes with it, such as a case for a phone.
- Bundle: several items together at a combined price.
All aim to raise order value. All can backfire by adding choices at the wrong time.
What to Measure
deepsync("track", "recommendation_viewed", { widget: "cart_addons" });
deepsync("track", "recommendation_clicked", { widget: "cart_addons" });
deepsync("track", "recommendation_added", { widget: "cart_addons" });Fire the view event only when the widget is actually on screen.
| Measure | Why |
|---|---|
| Seen rate | Many widgets sit below where shoppers scroll |
| Click rate of those who saw it | Relevance |
| Add rate | Whether it sells |
| Order completion with widget against without | Whether it distracts |
| Order value with against without | The benefit |
| Returns of recommended items | Whether people regret the addition |
The comparison that matters is revenue per visitor, not order value alone. A widget that raises order value by five percent and lowers completion by eight percent is a net loss.
Placement and Its Risks
| Placement | Shopper's state | Risk |
|---|---|---|
| Product page, "goes well with" | Evaluating | Low. Can distract from the main item |
| Product page, "compare with" | Evaluating | Useful for upsell |
| Added-to-cart confirmation | Just committed | Good moment, if not a blocking popup |
| Cart page | Reviewing | Good, kept below the totals and checkout button |
| Checkout | Paying | High. Anything here risks the order |
| Post-purchase, confirmation page | Done | Very low. No order to lose |
| Follow-up email | Later | None |
The closer to payment, the more a recommendation costs if it distracts. The confirmation page is underused: the sale is safe and the shopper is in a buying frame of mind.
What Recordings Show
- The widget is never reached. It sits far down the product page.
- It is seen and scrolled past. The items are irrelevant: a different size of the same thing, an unrelated category.
- A click sends the shopper away from the cart to a product page, and they do not come back.
- A popup after add to cart that must be dismissed before continuing. Watch for instant closes (popups).
- Adding requires choosing a variant, which the widget cannot do, so the shopper is redirected.
- The widget pushes the checkout button below the fold on mobile.
- Out-of-stock items recommended.
- Items already in the cart recommended again.
- Slow loading, shifting the page as it appears.
Making Recommendations Work
- Relevance first. Genuinely complementary items, at a sensible price relative to the main product.
- One-tap add without leaving the page.
- Keep the shopper where they are. Never navigate a cart visitor away.
- Few options. Two or three, not twelve.
- Explain the reason. "Protects your new phone" is better than "You may also like".
- Keep the checkout button visible above any recommendations.
- Nothing blocking between cart and payment.
- Use the confirmation page for a low-risk offer.
- Tie it to a threshold. "Add this for ₹199 and get free delivery" gives a reason (delivery thresholds).
- Exclude out-of-stock and already-carted items.
- Reserve space so the widget does not shift the layout as it loads.
Remove widgets that are seen and never clicked. They add page weight and visual noise.
Measure revenue per visitor, not order value
A higher basket from fewer completed orders is not a win.
Key Takeaways
- Measure each widget separately, starting with whether it is seen.
- The nearer to payment, the higher the risk of distraction.
- Relevant, one-tap, non-blocking recommendations work best.
- Judge by revenue per visitor.
Conclusion
A recommendation is a suggestion from a good shop assistant: relevant, well timed, and never between the customer and the till. Check that yours behave that way.
Frequently Asked Questions
Related articles
Stay in the loop
Get the latest insights on product analytics and user behavior delivered to your inbox.



