Product Recommendation Email Examples for Cross-Sell and Replenishment

“You may also like” is a placeholder, not a strategy. Start with the relationship between the customer and the product, then choose the message and timing.

Recommendation patterns

Do not recommend products that are out of stock, restricted in the customer’s region, or incompatible with the original purchase.

Complementary product example

Subject: Complete your {{product_name}} setup

Customers who bought {{product_name}} often add:

{{complement_product_grid}}

We chose these because {{recommendation_reason}}.

[Shop compatible products]

Replenishment example

Subject: Ready for another {{product_name}}?

You purchased {{product_name}} on {{purchase_date}}. Based on your previous order, you may be ready to reorder.

[Reorder {{product_name}}]

Not ready? Snooze this reminder or update your preferences.

Recommendation quality

For every product block, define:

• source event;
• lookback window;
• relationship or ranking rule;
• availability check;
• price and currency freshness;
• fallback collection;
• exclusion list;
• measurement event.

Explain the relationship when possible. A reason makes a recommendation feel less arbitrary and gives the customer a way to correct it.

Suppression and frequency

Suppress a recommendation after the customer buys the recommended product, unsubscribes from the topic, or enters a higher-priority flow. Cap recommendation emails across campaigns; otherwise a purchase can trigger a post-purchase recommendation, replenishment recommendation, and general newsletter block at once.

QA checklist

• Recommendations are available and relevant.
• Product relationship is explainable.
• Price and inventory are current.
• Fallback content is clean.
• Purchases and preferences suppress or change the message.
• Tracking distinguishes recommendation clicks from ordinary campaign clicks.

See ecommerce personalization examples for broader context and replenishment examples for timing.

Pick the recommendation relationship first

Use a relationship the catalog and customer state can support:

If the source is a ranking model, keep the ranking reason and model version for review. A customer-facing “because you bought…” claim should be backed by a specific event, not a vague similarity score.

Protect the customer experience

Check availability, region, price, variant, age restrictions, and compatibility before rendering. Exclude products the customer just bought, returned, or complained about. Give the block an empty state: a useful guide, a category link, or no block at all.

Use one primary recommendation action. A grid with six unrelated CTAs makes measurement and choice harder. If several products are shown, order them by the same documented rule and label a general collection honestly.

Test the relationship

Create fixtures for a new customer, a repeat buyer, an unavailable recommendation, a returned purchase, a changed preference, and a customer already in a higher-priority flow. Compare against a fallback and track completed purchases, returns, support contacts, unsubscribes, and recommendation availability. “Recommended” is a promise of relevance; the test should measure whether the promise was kept.