Most Shopify merchants leave a meaningful amount of revenue uncollected every month by relying on manual cross-sells. You pick a few related products on each product page, set them as "You may also like," and leave it. The customer either sees something useful and adds it to their cart, or ignores it and carries on.

The problem with manual recommendations is that they do not learn. They do not know that customers who bought a specific protein powder are far more likely to also buy a specific flavour of protein bar. They do not know that customers who viewed a particular sofa and then added it to their cart are twice as likely to buy a matching footstool than customers who arrived from a different collection. Algorithmic recommendation engines do know these things, because they are derived from actual purchase and behaviour data.

The difference in AOV impact between manual cross-sells and a well-configured algorithmic recommendation system is typically 10 to 25%, depending on category and placement. For a store doing £100,000 per month, that is £10,000 to £25,000 in additional revenue without changing your traffic or your pricing.

Here is how to build it.

How Do Algorithmic Recommendations Differ from Manual Cross-Sells?

Manual cross-sells are static rules: "When customer views Product A, show Products B, C, and D." They are based on the merchant's judgement about what is complementary and do not change with customer behaviour or purchase patterns.

Algorithmic recommendations are dynamic. They derive from statistical patterns in purchase and browse data:

Logic type	How it works	Best for
Collaborative filtering	"Customers like you also bought X" — based on shared purchase patterns	Post-first-purchase recommendations
Purchase-based	"Customers who bought this also bought X" — strict co-purchase analysis	Product page "Frequently Bought Together"
View-based	"Customers who viewed this also viewed X" — browse session analysis	Collection page and browse context
Content-based	Products matched by shared attributes (same brand, category, price range)	New products without purchase history
Personalised	Combines user history with collaborative patterns for individual suggestions	Returning customer homepage and email

The most effective recommendation systems use a combination of these logic types, switching between them based on context and the availability of sufficient data. A new product with no purchase history cannot generate purchase-based recommendations; it needs content-based or view-based logic until it accumulates sufficient data.

How Do LimeSpot, Rebuy, Wiser, and Frequently Bought Together Compare?

The four most commonly used recommendation apps on Shopify each take a different approach to the problem.

App	Pricing	Recommendation logic	Best for
Rebuy	From $99/month (Starter, up to $4.4M GMV). Pro at $249/month	AI + rules engine, highly configurable	Stores wanting full control over recommendation logic and placement
LimeSpot	From $18/month (Essentials). Growth at $49/month	AI personalisation across all placements	Stores wanting broad AI personalisation at lower cost
Wiser	From $9/month (Starter). Scale at $49/month	AI-powered with multiple logic types	Smaller stores wanting algorithmic recommendations at entry price
Frequently Bought Together	Free tier + $9.99/month (paid)	Purchase co-occurrence analysis	Stores focused specifically on product page bundle recommendations

Rebuy is the most powerful and most configurable option. Its "Smart Cart" feature integrates recommendation logic directly into the cart drawer, surfacing personalised upsells at the highest-intent moment in the purchase journey. Rebuy's rules engine allows merchants to set conditions (show this recommendation only if the cart contains product from category X, or only to returning customers) that manual cross-sells cannot replicate. It also handles post-purchase upsells on Shopify's native post-purchase page. The tradeoff is price and setup complexity.

LimeSpot is the strongest value proposition for mid-market stores wanting broad AI personalisation across all page types without Rebuy's configuration overhead. Its homepage personalisation for returning visitors is a differentiator: returning customers see a personalised homepage based on their browsing and purchase history rather than a static editorial layout.

Wiser is the most accessible entry point for algorithmic recommendations. At $9/month for the starter tier, it provides purchase-based and view-based recommendation logic across product page, cart, and checkout. For stores in the £5,000 to £30,000/month revenue range, the ROI is easily justified.

Frequently Bought Together is a focused tool rather than a full recommendation platform. It analyses purchase co-occurrence to surface the products most commonly bought together and presents them as a bundle on the product page. For brands where product bundling is a key AOV strategy, it delivers strong results at minimal cost.

Where Should Recommendations Be Placed?

Placement determines both the conversion rate of recommendations and the customer experience implications. Recommendations placed in too many locations become noise; placed strategically, they are genuinely useful.

Product page: The highest-traffic placement for recommendations. Two distinct sections work well: a "Frequently Bought Together" bundle at the top of the page (near the add-to-cart button) and a "You May Also Like" section further down. The near-CTA placement drives higher conversion because the customer is in the highest-intent moment when viewing the product they have already decided to potentially buy.

Cart drawer: The second most valuable placement. Customers with something in their cart have demonstrated intent. A recommendation in the cart drawer ("Customers who bought this also picked up X") at this stage converts at higher rates than the same recommendation on a cold product page. Rebuy's Smart Cart specifically optimises for this placement.

Post-purchase page: The post-purchase upsell is a Shopify native feature available through apps like Rebuy and CartHook (from $99/month). After completing a purchase, customers are presented with a one-click upsell offer that charges to their existing payment method without requiring another checkout flow. Conversion rates for post-purchase upsells typically range from 3 to 15% depending on the offer relevance and price point. The key is presenting an offer that is genuinely complementary and priced to feel like a natural add-on rather than a significant additional purchase.

Homepage (returning visitors): Personalised homepage recommendations for returning customers replace generic editorial content with products relevant to their browse and purchase history. LimeSpot handles this well. The impact is highest for stores with large catalogues where customers may not otherwise discover products relevant to their interests.

Email: Klaviyo's product block can surface recommendations based on purchase history for post-purchase flows. Combined with zero-party preference data, these email recommendations can be among the most relevant touchpoints in the customer relationship.

How Do You Avoid Recommendation Fatigue?

Recommendation fatigue occurs when recommendations become so pervasive and poorly targeted that customers start to ignore them entirely. The common causes:

Too many recommendation placements on a single page
Recommendations that are clearly irrelevant (recommending pet food accessories to a customer who bought a single-product gift)
Recommendations that repeat the same set of products regardless of context
Post-purchase upsells for products the customer already owns or has just bought

The mitigation principles:

Relevance over volume: One highly relevant recommendation converts better than five generic ones. Configuring exclusion rules (do not recommend products the customer already owns; do not recommend the same product category the customer just bought in) improves signal quality.

Context-matching logic: The recommendation logic on a product page should match the product context. View-based logic ("also viewed") is appropriate in a discovery context; purchase-based logic ("also bought") is more appropriate in a committed-buyer context like the cart.

Frequency capping in email: Klaviyo flows should not surface recommendation content in every email. Reserve recommendation blocks for specific flows (post-purchase, replenishment, win-back) rather than including them in every broadcast.

How Do You Measure Recommendation Revenue in Shopify Analytics?

Most recommendation apps provide in-app attribution reporting showing the revenue attributed to recommendations. Rebuy and LimeSpot both have dashboards showing recommended-product add-to-cart events and orders attributed to recommendation interactions.

The standard attribution model used by most apps is last-touch: if a customer adds a recommended product to their cart and completes the purchase, the recommendation gets credit for that product's revenue. This is a reasonable approximation but it overstates impact slightly (some customers would have added the product anyway).

For a more conservative measurement, compare AOV for orders where a recommended product was added versus orders with no recommendation interaction. This comparison controls for the baseline effect and gives a clearer picture of incremental AOV impact.

In Shopify Analytics, you can track this manually:

Revenue per order for the period before installing a recommendation app
Revenue per order for the equivalent period after installation
The difference, controlling for seasonal effects, is the approximate AOV impact

A well-configured recommendation system on a Shopify store with a broad catalogue typically lifts AOV by 10 to 25%. For a store with a £60 average order value, a 15% lift means an additional £9 per order. At 500 orders per month, that is £4,500 in monthly incremental revenue from a £99/month app investment.

Key Actions to Take Now

Evaluate your current cross-sell setup. If you are relying purely on Shopify's native "Related products" feature with no algorithmic logic, consider Wiser (entry at $9/month) or LimeSpot ($18/month) as accessible upgrades.
Identify your highest-AOV product category and prioritise configuring "Frequently Bought Together" recommendations for those product pages first.
Enable cart drawer recommendations via your chosen app. This placement typically delivers the highest recommendation conversion rate.
Set up a post-purchase upsell page via Rebuy or CartHook for your most-purchased product, targeting the complementary item most commonly bought in the same order.
Configure exclusion rules in your recommendation app to prevent recommending products the customer already owns or categories they have just purchased.
Record your AOV before implementing algorithmic recommendations and measure again after 60 days to calculate the incremental impact.
Review your recommendation placements every 90 days to ensure they remain relevant as your catalogue evolves and new purchase patterns emerge.

Frequently Asked Questions

Do I need Shopify Plus to use AI product recommendations?

No. All the apps mentioned in this guide work on standard Shopify plans. Shopify Plus unlocks additional customisation options (such as checkout upsells natively within the checkout flow), but the core recommendation functionality on product pages, cart, and post-purchase upsell pages is available on all plans.

How long does it take for AI recommendations to become accurate?

Collaborative filtering and purchase-based recommendation logic requires a baseline of purchase co-occurrence data to work accurately. Most recommendation apps need at least 200 to 500 orders across relevant product pairings before their algorithms produce reliable suggestions. In the early period (first few weeks), content-based logic (recommendations based on product attributes rather than behaviour data) bridges the gap. Expect 4 to 8 weeks before algorithmic recommendations are performing at their full potential.

What is a typical post-purchase upsell conversion rate?

Post-purchase upsell conversion rates on Shopify typically range from 3 to 15% of orders, depending on the relevance of the offer, the price point, and the category. Offers at 30 to 50% of the original order value tend to convert more reliably than those at higher price points. The "one-click" nature of the post-purchase upsell (no re-entering payment details) removes a significant friction point and is the primary reason these offers convert at all in the immediate post-purchase moment.

Can I use AI recommendations in Klaviyo emails?

Yes. Klaviyo's product recommendation block can surface products based on purchase history and browsing behaviour using Shopify data. More sophisticated personalisation (surfacing products based on quiz responses or zero-party preference data) requires custom setup using Klaviyo's template variables and custom properties. LimeSpot and Rebuy both have Klaviyo integrations that extend recommendation logic into email content, enabling the same algorithmic recommendation engine that runs on your store to also power your email product blocks.

AI Product Recommendations on Shopify: How to Increase AOV With Smarter Suggestions