11 min read · Ecommerce Growth · Last updated July 2026
Quick answer: The right product recommendations in the right locations add 20–25% to average order value. Product page recommendations convert at 3–5%, post-purchase page recommendations at 8–12%. The key is matching the recommendation type (cross-sell, upsell, or co-purchase) to the moment in the customer journey.
Introduction
Amazon’s “Frequently Bought Together” and “Customers Who Bought This Also Bought” sections reportedly generate 35% of Amazon’s total revenue. That number — 35% of the world’s largest ecommerce platform from recommendation placement alone — should reframe how seriously you take this.
For most ecommerce stores, product recommendations are an afterthought: an app is installed, default settings are kept, and the widget sits at the bottom of the product page where 80% of visitors never scroll. Maximum app subscription cost, minimum revenue impact.
This guide covers exactly where to place recommendations, which type of recommendation to use at each placement, and how to set up the algorithm to surface meaningful suggestions.
What you’ll learn:
– The 3 types of product recommendations and when to use each
– The 5 placement locations and their conversion benchmarks
– How recommendation algorithms work (and how to override them)
– The Shopify apps worth using at each price point
Table of Contents
- The 3 Types of Product Recommendations
- Cross-Sell: Different Category, Complementary Need
- Upsell: Same Category, Better Version
- Frequently Bought Together: Data-Driven Co-Purchase
- The 5 Placement Locations
- Placement 1: Product Page
- Placement 2: Cart Page
- Placement 3: Post-Purchase Page
- Placement 4: Post-Purchase Email
- Placement 5: Homepage
- How Recommendation Algorithms Work
- Shopify Recommendation Apps
- Interactive Tools
- FAQ
The 3 Types of Product Recommendations
Cross-sell: Recommending a product from a different category that complements what the customer is viewing or buying.
– Camera → Camera bag + memory card + lens cleaning kit
– Coffee beans → Coffee grinder + scale + pour-over kettle
– Running shoes → Running socks + insoles + GPS watch
The goal: increase order value by adding complementary items the customer would have bought separately anyway.
Upsell: Recommending a premium version or upgrade within the same category.
– Basic protein powder (1kg) → Premium version (2kg, better value per serving)
– Standard plan → Premium plan
– Sunscreen SPF 30 → SPF 50+ with added moisturizer
The goal: shift the customer to a higher AOV version of the same purchase.
Frequently Bought Together (FBT): Algorithm-driven co-purchase recommendations based on actual order data.
– “92% of customers who bought [Product A] also bought [Product B]”
– Unlike cross-sell (manual curation), FBT is data-driven — it surfaces what customers actually buy together
The goal: surface non-obvious but high-conversion product combinations that you might not have manually curated.
Cross-Sell: Different Category, Complementary Need
Cross-selling is most powerful when the recommendation solves a problem the customer hasn’t thought about yet.
The vacuum-bag example: A customer buying a vacuum cleaner needs replacement bags. Most stores forget to mention it. The customer buys the vacuum, runs out of bags in 3 months, and orders them from Amazon. That’s a cross-sell opportunity you lost.
Designing effective cross-sells:
-
Map your product catalog by use case, not category. A kitchen knife (category: cutlery) cross-sells with a cutting board (category: kitchen accessories) and a knife sharpener (category: tools). Use-case mapping surfaces these connections that category mapping misses.
-
Price anchor your cross-sells. The recommended product should be 20–40% of the hero product’s price. Recommending a $150 accessory alongside a $30 product feels jarring. Recommending a $15 accessory alongside a $80 product feels natural.
-
“Complete the experience” framing works better than generic “You might also like.” Tell customers WHY the complementary product enhances what they’re already buying.
Upsell: Same Category, Better Version
Upselling works when the value proposition for the upgrade is immediately obvious and the price difference isn’t jarring.
Upsell mechanics that convert:
– Bundle pricing: “Get the Pro Kit (includes X, Y, Z) for $79 vs buying them separately for $110 — save $31”
– Size upgrade math: “The 1kg costs $29. The 2kg costs $49 — that’s $0.50 less per serving”
– Feature highlight: Show the specific feature the upgraded version includes that the standard version lacks
Where upsells work:
– Product page: below the main CTA, before the customer adds to cart (“Looking at [Basic]? [Premium] includes [Feature] for $X more”)
– Cart page: before checkout (“Upgrade your [Product] to [Premium] for $X more”)
– Checkout page: “Add [Upgrade] for $X” — Rebuy’s checkout widget does this well
What ruins an upsell:
– Upselling something that’s 2x the price without a clear reason
– Showing the upsell in the wrong moment (mid-checkout flow is disruptive)
– Generic “you might also like” framing with no value articulation
Frequently Bought Together: Data-Driven Co-Purchase
FBT is the most powerful recommendation type for stores with sufficient order data (500+ orders minimum, ideally 2,000+).
How it works: The algorithm analyzes your order history and identifies products that appear in the same order at a statistically significant rate. A product pair that shows up together in 15% of orders when the average co-purchase rate is 3% is a strong FBT candidate.
What makes a good FBT recommendation:
– Co-purchase rate should be at least 5x the baseline (if average co-purchase rate is 2%, a 10%+ co-purchase rate is meaningful)
– Both products should be typically available (no point surfacing an out-of-stock item)
– Price combination should make sense (total cart value should feel reasonable)
FBT with insufficient data: If you have under 500 orders, FBT will surface poor recommendations because there isn’t enough signal. In this case, manually curate “Complete the Look” or “Complete the Set” bundles instead. The outcome is similar, but human-curated rather than algorithm-driven.
The 5 Placement Locations
Placement 1: Product Page
Position: Below the fold, after product description and reviews. NOT above the Add to Cart button.
Why below the fold: Your primary CTA is “Add to Cart.” Any recommendation above it competes with that goal and reduces conversion rate. Recommendations below the fold serve customers who are considering the product and want to see the full picture.
Best recommendation type: FBT and cross-sell.
Conversion rate benchmark: Product page recommendations convert at 3–5% (meaning 3–5% of recommendation impressions result in a click and eventual purchase).
Format that works: “Complete the set” with a single-click “Add all to cart” button. The add-all-to-cart format increases FBT conversion by 40–60% vs individual add-to-cart for each item.
Placement 2: Cart Page
Position: Below the cart items, above the checkout button.
Why: The customer has confirmed intent to buy. A cross-sell here feels helpful rather than promotional — they’re already in “buying mode.” This is your last chance to influence order value before checkout.
Best recommendation type: Cross-sell (not upsell — they’ve already chosen their products). Impulse-buy-priced items work best here (under $25 — easy to add without extended consideration).
Conversion rate benchmark: Cart page recommendations convert at 5–8% — higher than product page because purchase intent is confirmed.
Format that works: “Add to order — ships free with your current cart” removes shipping cost as an objection to adding the item.
Placement 3: Post-Purchase Page
Position: On the order confirmation page, immediately after the purchase completes.
Why this is your highest-converting placement: The customer has just paid. Their guard is down. They’re in a positive emotional state (just completed a successful purchase). Payment info is already entered. A single-click “Add to my order” purchase requires no additional friction.
Best recommendation type: Cross-sell or FBT — something clearly complementary to what they just bought.
Conversion rate benchmark: Post-purchase page recommendations convert at 8–12% — the highest of any placement. Some brands see 15%+ on well-matched recommendations.
The critical detail: The post-purchase recommendation must be processed as a separate charge after checkout. Use Rebuy or a post-purchase upsell app (AfterSell, ReConvert) — Shopify’s native checkout doesn’t support this natively without an app.
Placement 4: Post-Purchase Email
Position: In the 2nd or 3rd email of your post-purchase sequence (not the order confirmation email — that’s transactional).
Why: The customer has received their order, ideally used it, and is now in the consideration window for what comes next. A post-purchase email 7–14 days after delivery with “Customers who bought [Product] also love [Product B]” is relevant and timely.
Best recommendation type: Cross-sell for consumables, category expansion for durables.
Conversion rate benchmark: Post-purchase email cross-sell clicks convert at 4–8% — higher than standard promotional emails because context is established.
Placement 5: Homepage
Position: “Recently Viewed” section for returning visitors. “Bestsellers in your category” for returning visitors with purchase history.
Why: A returning visitor who previously bought skincare shouldn’t see kitchen products first on the homepage. Category-affinity sorting lifts CVR for return visitors by 5–10%.
Best recommendation type: Category-affinity based, recently viewed products.
How Recommendation Algorithms Work
Most Shopify recommendation apps use one of three approaches:
Rule-based (manual merchandising): You manually define which products to recommend together. Full control, but requires ongoing maintenance as catalog changes.
Co-purchase analysis: Algorithm analyzes order history for co-purchased items. Data-driven and improves over time. Requires 500+ orders for meaningful results.
Machine learning hybrid: Combines co-purchase data with category attributes, browse history, and individual customer data to generate personalized recommendations. Requires more data and typically costs more.
Overriding the algorithm: Most apps allow manual overrides. Use these for:
– High-margin products you want to push (even if not the most-recommended organically)
– New products without co-purchase history
– Bundles you’ve manually assembled with strong value proposition
Shopify Recommendation Apps
| App | Best For | Key Strength | Monthly Cost |
|---|---|---|---|
| Rebuy | Post-purchase, checkout upsell, AI recs | Best post-purchase page, smart cart | $99–$749 |
| LimeSpot | Product page, collection, email recs | Best value, good for growing stores | $18–$99 |
| Also Bought | FBT widget, simple implementation | Clean FBT display, easy setup | $9.99 flat |
| Frequently Bought Together | Single-focus FBT | Best pure FBT widget | $9.99 flat |
| ReConvert | Post-purchase page upsell | Drag-and-drop post-purchase builder | $49–$99 |
The combination that works for most stores ($50k–$500k/month):
– Rebuy for post-purchase page and cart recommendations
– LimeSpot for product page and collection page recommendations
– Klaviyo for post-purchase email recommendations
Interactive Tools
Widget 1: Recommendation Placement Strategy Matrix
Recommendation Placement Planner
Select your setup to see which placements to prioritize
Widget 2: Cross-Sell Revenue Calculator
Cross-Sell Revenue Calculator
Key takeaway: The post-purchase page is your highest-converting recommendation placement at 8–12% conversion rate — and most stores leave it completely blank. Set up a single cross-sell there before optimizing anywhere else.
FAQ
Should I show cross-sells on the product page or only in the cart?
Both — but with different intent. Product page cross-sells should be lower consideration items that complement the hero product. Cart cross-sells should be impulse buys at a lower price point. Don’t show the same recommendation in both places.
What’s the right number of recommendations to show at each placement?
Product page: 4–6 products (gives choice without overwhelming). Cart page: 2–4 products (they’re about to check out, don’t distract them). Post-purchase page: 1–3 products (focused, high-intent placement — more choices dilute conversion). Email: 3–4 products.
How do I know if my recommendation algorithm is working?
Monitor the “recommendation click-through rate” (CTR) and “recommendation revenue” metrics in your app dashboard. If CTR is under 1%, your recommendations are not relevant. If they’re over 5%, the algorithm is working well. Most apps show this data in their analytics dashboard.
Can I manually override algorithm recommendations for specific products?
Yes — all major recommendation apps support manual overrides. Use these for new products (no order data yet), high-margin products you want to push, or seasonal combinations that the algorithm might not surface. Review manual overrides quarterly to ensure they’re still relevant.
Does adding too many recommendation widgets slow my site down?
Yes — each recommendation widget loads additional JavaScript and makes API calls to fetch recommendations. Limit yourself to 2–3 active recommendation placements per page. Monitor your PageSpeed score after installing any recommendation app — a 10+ point drop suggests the app is loading too aggressively.
Conclusion
Product recommendations are not a set-and-forget feature. The placement, the recommendation type, and the algorithm all determine whether you see a 3% conversion rate or 12% on your recommendation widgets.
Start with the post-purchase page — it’s the highest-converting placement and easiest to implement with no risk to your main conversion flow. Then add cart page cross-sells, then product page FBT. Build from where the intent is highest and work backwards.
Ready to Grow Your Ecommerce Store?
Our ecommerce growth team handles SEO, email, Shopping, and CRO as one connected system. → Talk to Our Ecommerce Team
Written by the Ignited Nepal ecommerce team. ignitednepal.com