Recommendation block strategies

Strategies define how products are selected for a block. Each anchor type supports different strategies.

​

Interaction strategies

Use behavioral data to find products that customers frequently interact with together. Interaction data is computed periodically based on historical events.

​

Collection interaction strategies

Use behavioral data to find products related to a collection based on how shoppers who browse that collection interact with products. Unlike product interaction strategies (which find product-to-product relationships), collection interaction strategies compute collection-to-product relationships.

​

Similar products strategies

Use vector similarity to find products that are visually and semantically similar to a seed product. The base Similar Products strategy seeds from the anchor product on the page; the behavior-seeded variants seed from products in the shopper’s own session — recent views, cart, or orders — so recommendations follow real intent instead of the current page alone.

​

Similar Products

Supported anchors: Product, Collection How it works: The system generates embeddings for each product based on product images, titles, descriptions, and attributes. It then runs the full ranking pipeline — combining vector similarity with behavioral signals — and stores the top results for each product in a precomputed cache. When a customer views a product, the block serves results directly from this cache for fast response times. If no precomputed data exists yet (for example, for a newly added product), the system falls back to real-time vector search automatically. For collection anchors, the system automatically selects a representative product from the collection to use as the similarity source. Precomputed results and freshness:

• Results are recomputed daily when the existing data is more than three days old or new products have been added since the last run
• When a product’s embeddings are updated (for example, after editing its title, description, or images), that product’s similar products are recomputed incrementally
• All filters, facets, pagination, and sort orders continue to work as expected on precomputed results
• No configuration is required — precomputed caching is enabled automatically for stores with active similar products blocks

Use cases:

• “Similar Products” on product pages
• “You May Also Like” recommendations
• Visual discovery and exploration
• Collection-based similarity recommendations

​

Behavior-seeded similarity

Behavior-seeded strategies blend the embeddings of products the shopper has already interacted with into a single seed vector, then run similarity search against your catalog. Products the shopper has already seen, added to cart, or purchased are automatically excluded from the results so they always see something new.

​

Manual strategy

Curate products using a collection or hand-picked product IDs, with optional custom sorting. This strategy gives you complete control over which products appear in the block. Supported anchors: Product, Collection, Cart, None The example below shows selecting the manual strategy when you want to hand-pick which products appear in a block. Configuration:

• Hand-picked products: Select specific products by ID in the dashboard. The display order matches the order you define. When hand-picked product IDs are configured, they take priority over collection-based selection.
• Per-anchor product picks: For product-anchored blocks, you can hand-pick different recommendations for specific anchor products. When a customer views a product that has anchor-specific picks configured, those picks are returned instead of the default list. If no anchor-specific picks exist for the current product, the block falls back to the default hand-picked products.
• Collection anchor blocks: The collection is determined dynamically from the API request’s anchor_id parameter (no collection selection needed in dashboard)
• None anchor blocks: Select a collection to pull products from in the dashboard
• Optionally apply a sort order for custom sorting
• Apply merchandising rules: For collection-sourced blocks, toggle on to inherit the collection’s merchandising rule pins and banners. Off by default.

​

Apply merchandising rules

By default, manual blocks that pull products from a collection skip the collection’s merchandising rules — you get the raw collection contents in the configured sort order, without the pins, banners, or other rule effects that would normally apply on the collection page. Enable Apply merchandising rules when you want a collection-sourced block to behave like the collection itself:

• Pinned products surface at the top of the block, matching their pinned positions on the collection page.
• Banners configured on the collection’s merchandising rule are returned alongside the products.
• Other rule effects (for example, sell-out deletion) apply to the block’s results.

Leave it off when the block should show the collection’s products in a fixed order — for example, a “New Arrivals” rail that always sorts by date regardless of what’s pinned on the collection page.

​

Per-anchor product picks

Per-anchor product picks let you tailor manual block recommendations based on which product a customer is viewing. Instead of showing the same hand-picked products for every anchor, you can define unique product lists for specific anchors. How it works:

1. You configure a default set of hand-picked products for the block
2. Optionally, you add per-anchor overrides for specific products
3. When the block is requested with an anchor_id:
   • If the anchor has specific picks configured, those products are returned
   • Otherwise, the default hand-picked products are returned

This is useful when certain products have natural pairings that differ from your general recommendations. For example, a “Complete the Look” block might show different accessories depending on which clothing item a customer is viewing. Use cases:

• “Best Sellers” block (collection of top products, sorted by sales)
• “New Arrivals” block (collection of recent products, sorted by date)
• “Editor’s Picks” (curated collection with custom order)
• “Staff Picks” (hand-picked products in a specific order)
• Seasonal promotions (collection of seasonal products)
• “Complete the Look” with product-specific accessory pairings
• Targeted cross-sells that vary by anchor product

​

Personalized strategies

Personalized strategies resolve per request against the current shopper’s session and identity. They don’t precompute a product-to-product matrix — instead they pull products directly from the shopper’s own browsing, cart, order history, or replenishment cadence. If the shopper has no relevant session activity yet, the block defers to its fallback chain.

​

Trending strategies

Show what’s popular across the store without needing a shopper or anchor product. Trending strategies are ideal for top-of-funnel surfaces like the home page or 404 pages where you have no other context to work with.

​

Contextual strategies

Use signals from the page itself — the search query, the traffic source, or the cart total — to select products. Contextual strategies are aware of what brought the shopper to the page rather than who the shopper is.

​

Strategy availability by anchor type

Not all strategies are available for all anchor types. Use this matrix to plan which strategy to pick for the surface you’re building.