The workflow at a glance
- Product identity
- Relationship types
- Compatibility checks
- Relevant suggestions
What we need before implementation
- Product codes and variant structure
- Verified compatibility
- Relationship editor
At a glance
Organize variants, product families and related items into useful recommendations with compatibility and availability checks.
Different relationships serve different purposes
A size variation, compatible accessory and higher-priced alternative are not the same relationship. Map variants, families, upsells and cross-sells separately. The goal is to support product selection without confusing the catalogue. Store relationships using stable references so they survive title or category changes. Existing product identities remain intact; linked recommendations should not create additional duplicate products or interfere with how variations are priced and stocked.
Compatibility needs evidence
For technical products, a shared category is insufficient proof of compatibility. Use verified models, references and attributes. A related accessory may require a particular device version. Display the relationship and its conditions clearly. AI may propose candidates for review but should not independently approve technical compatibility. Evidence from manufacturers or reviewed product records should remain available to whoever maintains the relationship.
Display eligible and available items
Recommendations pass visibility, pricing and availability checks. Withdrawn products should not remain active suggestions. B2B customers see eligible items and prices for their accounts. Define how many options appear and where, preventing irrelevant cards from crowding the cart. Business priorities may affect order without overriding relevance. When a suggested item is unavailable, apply an agreed fallback instead of silently recommending a technically incompatible alternative.
Give the team control
Distinguish manually curated relationships from those imported through ERP or supplier data. The team can approve, exclude or correct suggestions. Catalogue changes trigger checks for outdated links. Bulk imports start with a small sample in which both ends of every relationship are verified. Avoid multiplying products simply to create artificial choice. Retain the source of each relationship so corrections are not overwritten by the next automated import.
A practical example and evaluation
For a photography retailer, lenses and batteries appear according to verified camera models rather than brand alone. Test a withdrawn camera, a model change and an unavailable accessory. Observe whether recommendations are used and whether compatibility questions decrease. Assess useful commercial choices and reduced mistakes without promising every suggestion will increase order value. Review relationships periodically, especially when suppliers replace product generations or change the accessories bundled with a device.
Acceptance checks before launch
- Withdrawn items leave active recommendations
- B2B restrictions apply
- Manual exclusions remain protected
FROM READING TO ACTION
A clear plan for your platform.
Tell us where your platform struggles and what needs to improve. Our team can connect the diagnosis with code, performance and infrastructure work.
Discuss your platform