Skip to main content

​
Overview

Catalogs are containers for products. They provide a structured workflow for managing product content that mirrors git-based version control.

​
Catalog lifecycle

​
States

  • Draft — editable, not visible to external consumers
  • Published — locked for direct edits, visible via the public products API

​
Operations

Marks the catalog as live. Published products become available through the public /api/catalog/products endpoint. A published catalog cannot be directly edited — fork it first.
Reverts the catalog to draft state and removes its products from the public feed.
Creates a copy of the catalog. Use forks to iterate on content (add products, edit descriptions, run AI enhancements) without affecting the live version.
Applies changes from a fork back into the parent catalog. Conflict resolution is handled at the product level.

​
Catalog stats

Each catalog tracks aggregate statistics:
  • Total product count
  • Products with optimization score above threshold
  • Feed-ready products (all required fields populated)
  • Last published timestamp

​
Marketplace Hub

Published catalogs can be listed on the Marketplace Hub — the B2B content exchange where other workspaces can discover, preview, clone, or subscribe to your catalog.
  • Publish to Hub — make your catalog discoverable in the public feed
  • Subscribe — receive live sync updates when the publisher releases a new version
  • Clone — take a static snapshot of a catalog at a point in time
  • Conflict resolution — when a subscribed catalog updates, field-level conflicts can be resolved automatically or manually
See Marketplace for the full concept overview, and the Marketplace Hub API for programmatic access.

​
Data pipeline integration

When products are imported, the Bronze → Silver → Gold pipeline runs automatically or on demand:
  • Bronze — raw ingest, idempotency keyed by source_identifier
  • Silver — normalization (field mapping, duplicate detection, image URL validation)
  • Gold — optimization scoring (0–100) across 7 rubric stages
Batch actions trigger pipeline stages for a selection or the entire catalog:
  • Normalize (Silver) — run field normalization in bulk
  • Analyze (Gold) — compute optimization scores in bulk
Pipeline behavior (auto-trigger, Silver mappings, Gold weights) is configurable via Pipeline Settings.

​
Best practices

Use the fork → edit → merge workflow for seasonal updates. This lets your team iterate on “Spring 2026” content while “Winter 2025” stays live.
Run Batch Gold after importing a new catalog to get an initial quality baseline before publishing.
Last modified on March 18, 2026