- The Shopify AI recommendations algorithm ranks products using sales history, product relationships, and shopper behavior, not random picks.
- New stores and new products get weaker suggestions because the algorithm has little data to learn from.
- You can tune the results with better product data, manual rules, and complementary product links.
- AI merchandising pays off most when it's paired with clean product pages and personalization.
The Shopify AI recommendations algorithm decides which products show up in "You may also like" and related blocks. It's not guessing. It ranks items based on data from your store and shopper patterns.
Get it right and those blocks push more items into each cart. Get it wrong and they show junk nobody wants.
Here's how the algorithm works, and how to tune your merchandising for more revenue.
What the Shopify AI recommendations algorithm does
Shopify's recommendation engine picks products to show on your pages. It powers the "Recommended products" and related-item sections built into most themes.
It pulls from Shopify's AI, which also sits behind the Shopify Magic AI tool for things like product copy. The goal is simple: show items a shopper is likely to buy next.
There are two main modes behind it:
- Automated - Shopify's AI chooses products for you based on data.
- Rule-based - you set filters, like same collection or price range.
- Manual - you hand-pick exactly which products appear.
Most stores run on automated by default. That's fine to start, but it leaves money on the table if you never tune it.
How the algorithm ranks products
The algorithm weighs several signals, then ranks the best matches for each page. Think of it as scoring every product against the one a shopper is viewing.
Here are the main inputs it uses:
| Signal | What it measures | Why it matters |
|---|---|---|
| Purchase history | Products bought together | Strongest signal for "also bought" blocks |
| Browsing behavior | Products viewed in the same session | Catches related interest before a sale |
| Product relationships | Shared type, tags, collection | Helps new products get matched |
| Conversion rate | How often an item sells when shown | Pushes proven sellers higher |
| Recency and stock | Live inventory and new arrivals | Stops it recommending sold-out items |
The engine favors products with a track record. An item that converts well when shown gets shown more. That's a flywheel, which is great once it starts but slow for new stores.
Don't expect strong recommendations on day one. The algorithm needs sales and traffic data to learn. A brand-new store leans mostly on product relationships, not behavior.
Why new products get weak recommendations
This is the "cold start" problem. The algorithm has no purchase or view data for a fresh product, so it can't score it well.
Until data builds up, Shopify falls back to simpler signals like collection and product type. That's why new arrivals often sit buried in related blocks.
You can speed this up. Clean product data gives the algorithm something to work with right away.
- Set accurate product type and vendor fields.
- Add specific, consistent tags.
- Group related items in the same collection.
- Link new products from popular pages so they get views.
Want AI merchandising that sells more?
We tune Shopify's recommendations, search, and personalization so the right products show at the right moment.
How to tune your merchandising for more revenue
You don't have to accept the default output. Here's a practical order to tune an AI ecommerce store for better results.
- Fix your product data first
Set product type, tags, and collections cleanly. Bad data means bad matches, no matter how smart the algorithm is.
- Pick the right recommendation mode
Use automated for high-traffic pages and manual rules for bundles or hero products you want to push.
- Add complementary product links
Tell Shopify which items pair well. This feeds "complete the look" and "frequently bought together" style blocks.
- Place blocks where they convert
Product pages and cart pages drive the most add-ons. Test position above and below the fold.
- Review the results monthly
Check which recommended products actually get clicks and sales. Swap out weak picks in manual sections.
Use complementary recommendations on product pages and "related" recommendations on collection pages. Shoppers deep in a product want add-ons. Shoppers browsing want alternatives.
For a deeper look at placement and testing, read our guide on what actually drives revenue with AI product recommendations.
Automated vs manual: which to use where
Both have a place. The trick is matching the mode to the job.
Automated AI
- Scales across every product
- Learns from real behavior
- Great for large catalogs
- Weak on brand-new items
Manual rules
- Full control over picks
- Perfect for bundles and launches
- Promotes high-margin items
- Needs upkeep as stock changes
Our take: let automation handle the long tail, and use manual rules for your top 20 products and promotions. That mix gets you reach plus control.
Where AI recommendations fit in the bigger picture
Recommendations are one piece. They work best alongside search, personalization, and segmentation.
Shopify also offers deeper tools through Shopify OpenAI features and the Magic suite. A ChatGPT Shopify integration or AI chatbot can suggest products in chat, using some of the same ranking logic.
Here's how the parts connect:
Want to connect these? Our AI search and merchandising service tunes the whole flow, and AI personalization tailors picks to each shopper.
You can also feed recommendation data into customer segmentation and email automation so your messages match what people browse.
Common merchandising mistakes to fix
Most stores lose add-on sales to a few fixable issues. Check these before anything else.
- Messy or missing product tags and types.
- Recommending out-of-stock items.
- Showing only same-category products, never complements.
- Burying related blocks at the bottom of long pages.
- Never reviewing what the blocks actually sell.
Fixing product data alone often lifts the quality of every automated suggestion. If your store is new, see our guide to what AI works first for ecommerce.
Frequently Asked Questions
What are product recommendations with AI?
AI product recommendations use data like past purchases, browsing, and product relationships to suggest items a shopper is likely to buy. On Shopify, this powers the "recommended" and "related products" blocks on your pages. The AI ranks products automatically instead of showing a fixed list.
How does the Shopify AI recommendations algorithm choose products?
It scores products using purchase history, session browsing, product type and tags, conversion rate, and stock levels. Items that sell well when shown get ranked higher over time. New products lean more on shared collections and tags until behavior data builds up.
Can I control which products Shopify recommends?
Yes. You can switch from automated to rule-based or manual recommendations in your theme or app settings. Manual mode lets you hand-pick items, which is ideal for bundles, launches, and high-margin products you want to promote.
Does the Shopify Magic AI tool affect recommendations?
Shopify Magic focuses on content like product descriptions and email copy, not the ranking of recommended products directly. But better product copy and clean data help the algorithm understand your items. Both run on Shopify's broader AI, including OpenAI-based features.
Why are my new products not showing in recommendations?
New products hit a "cold start" because the algorithm has no purchase or view data for them. Add accurate tags, product types, and collections, and link them from popular pages to gather views faster. Over time the AI learns and starts surfacing them more.
Want the recommendation engine working harder for you? Our team tunes Shopify's AI merchandising, search, and personalization so shoppers see the right products. See how our AI solutions for ecommerce can lift your average order value, or get a free quote when you're ready.
