Types of Recommendations
Not all product recommendations work the same way. Understanding the different recommendation types helps you choose the right approach for each location on your store.
Related Products
Products that are similar to or commonly bought alongside the current product. Based on purchase history or product metadata.
Recently Viewed
Products the current user has viewed in this session. Useful for bringing customers back to products they showed interest in.
Bestsellers
Your top-performing products by sales volume. A safe default recommendation that leverages social proof.
Personalised Recommendations
AI-powered suggestions based on the individual customer's browsing and purchase history. Requires a dedicated app and sufficient data to work well.
Native Shopify Recommendations
Shopify provides a product recommendations API that most modern themes support. Enable it in your theme settings on the product page template. The algorithm considers purchase history, product tags, and product type to generate suggestions automatically.
You can also manually set recommended products for specific items via metafields or by installing a simple app. Manual curation is worth doing for your top 20-30 products where you know exactly what pairs well.
Personalisation Apps
For larger catalogues and higher traffic stores, AI personalisation apps outperform static or algorithmic recommendations. LimeSpot and Rebuy are the two most used options in the UK market. Both use machine learning to adapt recommendations per visitor based on their behaviour.
These apps typically require 3-4 weeks of data to start personalising effectively. During that period, they fall back to bestseller-based recommendations.
Placement Strategy
Product page recommendations should appear below the product description or after the add-to-cart button. Label them clearly: "Frequently bought together", "Complete the look", or "You may also like" all perform well depending on your product category.
Cart recommendations appear most effectively as the last item before the checkout button. Post-purchase recommendations, shown on the order confirmation page, have no checkout friction and can drive impulse additional orders.
Recommendation Copy
The section headline matters. Test different labels for your recommendation sections. "Customers also bought" leverages social proof. "Complete your order" creates a sense of incompleteness. "People who bought this also bought" is collaborative social proof.
Measuring Impact
Track click-through rate on recommendation sections and the conversion rate and average order value of sessions that interact with them versus those that do not. Most recommendation apps provide this data natively. A well-configured recommendation engine should increase average order value by 5-15%.
