Beta
Documentation index
Concepts

Master product record and versioning

Noobit's core structure: each product's record, with its version history.

6 min read

The master product record is Noobit's core data model. Every product that comes in — from a CSV, a sync, a webhook or the scanner — is converted to this structure before anything else. The pipeline and publishing always work from it, never from the supplier's original format.

Why it exists

In a real catalog, data comes from many places: a supplier's CSV, the WooCommerce sync, what someone on your team scanned with their phone, what Noobit found researching the web. If every source uses its own format, the code fills up with special cases and edits get mixed up.

The master product record brings everything into a single format: the same field names, the same conventions. Everything that comes after (content generation, image processing, publishing) works from a single source of truth.

What it contains

Identity
catalog name, EAN, SKU, brand and type (simple or variable)
Content
product title, short description, long description, SEO title/description, tags
Commercial
current price, was price (struck through, only when there is a sale), cost, stock, stock status, dimensions and weight
Taxonomy
categories, brand, attributes
Media
images (MAIN + gallery), with roles per variation where relevant
Variations
a list of child variations with their own attributes, SKUs and prices
Sources
what each origin contributed (a CSV, your store…), which you can turn on or off for the AI
Context for the AI
CSV columns the AI reads as context but that aren't fields of the product record
Custom fields
key-value pairs for whatever the base model doesn't cover

The AI rewrites the product title, never the catalog name. And it writes the content from its Product Knowledge: the facts from all your sources, each tagged with where it came from. We explain it in Concepts · Product Knowledge.

Version history

Every manual edit, every AI generation and every import creates a new version. The previous one isn't deleted: it stays, with its date and author, in the «Versiones» (Versions) tab of the product record. What comes in from the store (price, stock and other data) updates the current version without creating a new one.

Rolling back is instant
Going back to an earlier version just changes which version the product points to: the data already exists, so it's instant and nothing is lost. Your store doesn't change until you publish again.

Several sources, one record

When a product comes from two sources (a weekly CSV and your WooCommerce store, for example), Noobit keeps what each one contributes separately, and in the «Fuentes» (Sources) tab you decide which ones the AI reads. Hard data (price, stock, EAN, weight) always comes from its source. Written content (title, descriptions, SEO) only changes when you take action: neither the store nor a new CSV overwrites it, unless you tick «Sobrescribir contenido redactado» (Overwrite written content) when importing. If someone edits it in the store, you'll see it in «Conflictos» (Conflicts).

Prices: current price and was price

The product record stores the price your customer pays today and, only when there's a sale, the was price (the struck-through one). When you publish prices, the record takes precedence: if it has no was price, the sale disappears from the store too, including a scheduled sale that hasn't started yet.

That's why, before you publish prices for the first time, it's a good idea to sync prices from your store: that way Noobit learns your sale prices.

What isn't in the master product record

Some data deliberately lives outside the master product record:

  • Publishing: which version is published in which store
  • Tasks: the history of pipeline operations
  • Approvals: pending and resolved
  • Notifications, activity log and usage

This keeps the master product record clean (only data about the product itself) and lets you work on operational metadata without touching the content.

What this means in practice

When you export or prepare an output, what goes out is always the master product record. You don't have to think about whether a piece of data came from WooCommerce, a CSV or the scanner: it's already normalized. The complexity stays on the way in.