The product data spec: an overview
A map of Google's product data specification, the required and recommended attribute groups (identity, price and availability, images, category, variants, shipping) and where to go deeper on each.
Google's product data spec is the rulebook for what every product in your feed must and should contain. It groups attributes into a handful of jobs: who the product is, what it costs and whether it's in stock, what it looks like, where it sits in Google's categories, how variants relate, and how it ships. This overview shows the whole map and links to the deep guide for each part.
What the spec is
The product data specification is Google's definition of the attributes a product feed can carry, which are required, which are recommended, and what valid values look like. Merchant Center validates every product against it. Get it right and products are accepted, matched and shown; get it wrong and they're disapproved or quietly starved of reach.
The spec is long, but it organises into a few clear jobs. Think of each group as answering one question about the product.
The attribute groups at a glance
| Group | The question it answers | Key attributes | Required? |
|---|---|---|---|
| Identity | Which exact product is this? | id, title, description, gtin, mpn, brand, identifier_exists |
Required (identifiers conditional) |
| Category | Where does it sit in Google's taxonomy? | google_product_category, product_type |
GPC effectively required |
| Price & availability | What does it cost, is it in stock? | price, sale_price, availability, link |
Required |
| Images | What does it look like? | image_link, additional_image_link |
Required |
| Variants | How do sizes/colours relate? | item_group_id, color, size, material, pattern |
Conditional |
| Demographics | Who is it for? | gender, age_group |
Conditional (apparel) |
| Shipping & tax | How does it get there, what's added? | shipping, tax |
Conditional |
| Merchandising | How do you want to segment it? | custom_label_0–custom_label_4 |
Optional |
Below, each group in one paragraph: with a pointer to its deep guide.
Identity: who the product is
The attributes that name the product uniquely. id is your internal key (stable, never reused). title and description are the text Google matches against shopper queries. And the product identifiers(gtin, mpn, brand, identifier_exists) are what tie your offer to Google's catalog. This group does the heaviest lifting: it drives both catalog matching and query relevance. Weak identity data caps everything downstream. Depth: GTIN, MPN and brand.
Category: where it sits
google_product_category places the product in Google's own taxonomy; product_type is your own free-text hierarchy for organising and reporting. A specific category ("Apparel & Accessories > Clothing > Dresses > Cocktail Dresses") unlocks far more than a top-level guess ("Apparel & Accessories"), because Google uses it to decide which searches and filters your product belongs in.
Price and availability: cost and stock
price, optional sale_price, availability, and the link to the buyable landing page. This group is the one Google continuously re-checks against your live page: a feed price or stock state that disagrees with what the crawler sees is the leading cause of disapprovals. Freshness of delivery matters as much as correctness here.
Images: what it looks like
image_link (the required primary image) and additional_image_link (extras). Images must be a clean shot of the actual product with no promotional text, watermarks or placeholders, hosted at a URL Google's crawler can reach. Image problems are a top-five disapproval cause.
Variants: how versions relate
When one product comes in several colours or sizes, item_group_id binds those variants into a family, while color, size, material and pattern distinguish them. Get this wrong (reusing one GTIN across variants, or leaving them unlinked) and you get either disapprovals or a messy, duplicated presence on the Shopping surfaces.
Demographics: who it's for
gender and age_group. Required for apparel and strongly recommended wherever they apply, because they power the filters shoppers use to narrow results. A product missing them simply doesn't appear in "women's" or "kids'" refined searches.
Shipping and tax: getting it there
shipping and tax at the product level, layered on top of your account-level settings. Most merchants configure shipping once at account level; per-product overrides are only for genuinely different rates. A product in a country with no shipping configuration is disapproved.
Merchandising: your own segmentation
custom_label_0 through custom_label_4 carry no meaning to Google: they're yours, for segmenting products in campaigns (margin band, season, bestseller, clearance). Optional, but the backbone of any serious campaign structure.
Deep dive "Required" is a floor, not a target
The most expensive misreading of the spec is treating "required" as the finish line. Meeting the required attributes gets a product accepted: it does not get it seen. The gap between accepted and competitive is almost entirely in the recommended and conditional attributes:
- Recommended attributes are where the long tail lives. A product with only the required fields shows for broad, generic queries and disappears from every refined, filtered or specific search. Add colour, size, gender, age group and material where the category expects them and the same product becomes eligible for hundreds of narrow, high-intent queries it was previously invisible to.
- Category-conditional requirements bite quietly. An attribute that's "recommended" globally can be effectively required for your category: a valid GTIN for media,
age_groupandgenderfor apparel. Miss these and you're not disapproved outright; you're just absent from the searches that convert. - The spec has priorities, and so should you. Not all attributes move the needle equally. Identity (identifiers, title, category) governs whether Google can match and understand the product at all: fix these first. Price/availability accuracy governs trust and disapprovals: fix these second. Recommended attributes govern reach: fix these third, biggest categories first. Merchandising labels are last, and only matter once the rest is sound.
- Completeness compounds. Each attribute you add doesn't just satisfy a rule; it gives Google another way to match, filter and rank your product. A fully-specced product isn't 20% better than a minimally-specced one: it's eligible for an order of magnitude more queries.
A single number that rolls this up (how complete and competitive your feed is against the priorities Google actually rewards) is what the Feed Health Score measures, so "are we meeting the spec well enough to compete?" has an answer you can watch move.
How to use this map
Work the groups in priority order, not feed order:
- Identity first: correct identifiers, strong titles, a specific category. This decides eligibility.
- Price and availability: accurate and fresh, matching your live page. This decides trust.
- Images and variants: compliant primary images, correctly linked variant families.
- Recommended and conditional attributes: colour, size, gender, age group, material where the category expects them. This decides reach.
- Merchandising labels: once the above is sound, structure for campaigns.
Each group above links to its deep guide. If you're still getting the account itself connected, start at Setting up Google Merchant Center; to see how these attributes turn into placements, read How Google Shopping listings work.
Frequently asked questions
What's the difference between a required and a recommended attribute?
Required attributes are the minimum for a product to be accepted and shown, omit one and the product is disapproved. Recommended attributes aren't gatekeepers for approval, but they strongly affect how often and how well your product surfaces. In practice, "recommended" means "required to compete", especially colour, size, gender and age group where the category expects them.
Do all products need the same attributes?
No. Some are universal (id, title, link, image, price, availability), but others are category-conditional, apparel needs colour, size, gender and age group; media needs a valid GTIN; multi-variant products need item_group_id. The spec is a floor plus category-specific rooms.
If I only fix one group, which matters most?
Identity, identifiers, title and category. They drive catalog matching and query relevance, which decide whether a product is even eligible to show. Price, images and the rest matter, but a product Google can't match or understand never gets far enough for them to count.