Standards Packs
Standards Packs are industry classification systems built directly into merchi.ai. When a Standards Pack is active on your account, the AI automatically identifies the most relevant classification for each product before generating content — and uses that classification to guide the output.
You do not need to configure or manage the classification yourself. It runs in the background every time you process a product.
What Standards Packs Do
When you process a product, merchi.ai analyses the product signal (from images, input data, or both) and queries the active standards to find the closest matching classification. The top candidates are injected into the AI prompt alongside your writing knowledge assets.
This means the AI knows not just what the product is, but how it is officially classified — including the standardised attribute set for that classification. This produces more consistent, technically accurate, and channel-ready content without any manual effort.
Available Standards
ETIM — Technical Product Classification
ETIM (Electrotechnical Information Model) is the international standard for classifying technical and industrial products. It is widely used in the electrical, HVAC, plumbing, and building services sectors.
Each ETIM Class has a defined set of Features — structured attributes like nominal diameter, connection type, rated voltage, or flow rate — with controlled value lists. When merchi.ai identifies the correct ETIM Class for a product, it extracts values for those features and includes them in the generated output.
Typical use cases: Electrical components, plumbing fittings, HVAC equipment, industrial supplies, building materials.
What you get in the output:
- ETIM Group and Class code and name
- Extracted values for each applicable ETIM Feature
- Technically accurate attribute descriptions grounded in the standard
Shopify Standard Product Taxonomy
The Shopify Standard Product Taxonomy is a universal product classification system maintained by Shopify and designed to be compatible with major commerce platforms and marketplaces.
The taxonomy covers over 14,000 product categories across all verticals, from Apparel to Electronics to Food & Beverage. Each category has a defined set of attributes (such as Material, Colour, Size Format, or Target Gender) with controlled value lists.
When merchi.ai identifies the closest Shopify taxonomy category for a product, it uses the category’s attribute set to structure the output more precisely.
Typical use cases: Retail, fashion, general merchandise, marketplace-ready product data.
What you get in the output:
- Matched Shopify taxonomy category path (e.g. Apparel & Accessories > Clothing > Tops)
- Attribute values aligned to that category’s standard attribute set
- Output structured for compatibility with Shopify and major marketplaces
Google Product Taxonomy
Google Product Taxonomy is the classification system used by Google Shopping and Google Merchant Center. It covers over 6,000 product categories and is the standard expected by Google for product feed submissions.
The taxonomy is a flat list — each category has a numeric Google ID and a full path string (e.g. “Apparel & Accessories > Clothing > Dresses”). Unlike ETIM, Shopify, and GS1 GPC, it is classification-only: there are no standard attribute definitions per category. Its primary value is ensuring the correct Google category ID is output for channel and feed use cases.
Typical use cases: Google Shopping feeds, Google Merchant Center product data, any channel requiring a Google Product Category field.
What you get in the output:
- Matched Google category ID (numeric)
- Full category path string ready for feed submission
- Output compatible with Google Merchant Center requirements
GS1 GPC — Global Product Classification
GS1 GPC (Global Product Classification) is the international standard maintained by GS1 — the organisation behind barcodes and the GTIN system. GPC classifies products into Segments, Families, Classes, and Bricks, with each Brick having a defined set of Attribute Types and Values.
GPC is used extensively in FMCG, grocery, healthcare, and supply chain contexts where interoperability between trading partners matters.
Typical use cases: Grocery, FMCG, healthcare products, B2B supply chain, GS1-compliant catalogues.
What you get in the output:
- Matched GPC Brick code and full hierarchy path
- Attribute values aligned to the Brick’s official attribute set
- Output structured for GS1-compliant data exchange
How Classification Works
For each product processed, merchi.ai:
- Builds a product signal — a compact text description derived from input data (if available) or a lightweight AI vision pass on the product image
- Embeds that signal as a vector and searches each active standards database for the closest semantic matches
- Returns the top candidate classifications (typically up to 10 per standard) with their full attribute definitions
- Injects these candidates into the AI prompt, where the model selects the best match and extracts attribute values accordingly
The entire classification pipeline runs automatically before the main generation step. If no strong match is found above the similarity threshold, the standard is skipped for that product and generation continues without it — there is no failure mode that blocks your output.
Viewing Your Active Standards Packs
Your active Standards Packs are shown at the top of the Writing Knowledge settings page. Each entry shows:
- The standard name and version
- The release date
- The number of classifications in the database
Standards Packs are account-level settings managed by the merchi.ai team. If you would like to add, change, or remove a Standards Pack on your account, contact support.
Standards Packs and Your Schema
Standards Packs work alongside your schema blocks. If your schema includes fields for taxonomy, classification codes, or structured attributes, the AI will use the identified standard to populate those fields accurately.
For example, a schema block defined as “Select the most appropriate ETIM Class and extract its Feature values” will be populated with the correct class and values for each product automatically — no manual lookup required.
See Schema Configuration for guidance on setting up blocks that take advantage of standards classification.
Frequently Asked Questions
Does the classification slow down processing? The classification step adds a small amount of time (typically under 2 seconds per product) because it runs a semantic search before the main generation. For most use cases this is negligible, and the improvement in output quality more than compensates.
What if the product doesn’t match any classification? If no candidate meets the similarity threshold, the standard is skipped for that product and generation continues normally using your writing knowledge assets. No error is raised and no output is blocked.
Can I see which classification was used for a product? The classification information is included in the generated output fields if your schema is configured to capture it. Check with your account team if you need classification codes surfaced in a specific output field.
Can I use multiple Standards Packs at the same time? Yes. If multiple Standards Packs are active on your account, all of them run in parallel for every product. Each standard’s candidates are injected separately into the prompt so the AI can draw on all of them simultaneously.
The classification looks wrong for some products — what should I do? Classification accuracy depends on the quality of the product signal. Products with rich input data (title, description, specifications) will classify more accurately than image-only products. If you are consistently seeing poor classification for a category, contact support with a few examples so we can investigate.