Cross-Selling

Cross-Selling powers “goes well with this” recommendations: given a product a shopper is viewing, it finds products from a different category that complement it — a shopper looking at shoes might see sock or bag recommendations. This is the complementary counterpart to Similar Products, which finds substitutes instead.

How it works

flowchart TD
    A[Shopper views a product] --> B[Match campaigns\nby source category]
    B --> C[Search target categories\nfor complements]
    C --> D{Ordering}
    D -->|Similarity| E[Merge by closeness]
    D -->|Rank| F[Round-robin\nacross categories]
    E --> G[Recommendations shown]
    F --> G

What it does

🎯 Source and target categories

You define which categories a product must belong to in order to trigger cross-sell recommendations (the source, e.g. shoes), and which categories the recommendations themselves are drawn from (the target, e.g. socks). Leaving either empty widens it to the whole catalog.
⚖️ Category and price awareness

Like Similar Products, results can be biased toward a similar price range and category proximity.
🎚️ Sensitivity controls

Require results to match the source product on chosen attributes.

Configuring in Panel

In Panel, open Site → Campaigns, and open or create a Cross-Selling campaign.

General settings — Name, Active, Default Campaign toggles.

Methods

Semantic Search — turn on and weight the AI models used for matching (e.g. Berry Punch Net, Volcanic Water Net, Kiba Net, Darjeeling Second Flush Net, depending on which are activated for the site).
Color — turn on color-aware matching if a color-capable Net is activated.

Influencers

Price impact — Price sensitivity, Reasonable price difference, and Scaling factor, same as Similar Products.
Category impact — bias results toward category proximity, judged by product type or Google product category.
Only Buyable Products — restrict results to in-stock products.
Age Groups / Gender — keep recommendations within the same age group or gender as the source product.

Interface controls

Preferred Image, maximum results — same as other services.
Filters — configure which attributes are filterable on the results panel.

Products — restrict the campaign’s source and target categories via Assigned Products.

With item grouping on, result variants are grouped.

Boosting is not currently available for Cross-Selling.

MCP Tool

Agents can call this service directly as the cross_selling MCP tool, which wraps the Cross-Selling REST operation — its request and response shapes match that operation’s schema.