Merchant Center in depth II: the feed specification, attribute by attribute

The feed is not a technical export: it is the ad. Every attribute is a keyword, a selling point or a filter that decides whether you show up at all. This lesson walks through the entire specification with the question "what does each field do for profitability?" in mind, not just "what does Google require?".

The required ones (without them, disapproved)

Attributes required by vertical: google_product_category in certain categories (apparel, media, software…), age_group, gender, color, size in apparel and accessories, energy_efficiency_class on household appliances (Europe), unit_pricing_measure wherever unit pricing law applies, multipack, is_bundle…

Unique identifiers: the rule and its exception

Google requires a GTIN when the product has one (it knows because the catalogue knows). With a valid GTIN: better matching, appearance in comparison listings and in "popular products". Without a GTIN for a product that has one: "incorrect/missing unique identifier" → limited or disapproved. Own-brand products with no barcode: identifier_exists: no + brand + optional mpn. Buying GTINs (GS1) is worth it if you sell own-brand products in volume.

The ones that change performance

title: the most profitable attribute

Google matches the query mainly against the title (and then against the description, product type and category). Proven rules:

• The important part first (it gets cut at about 70 characters on mobile): Brand + Product + Key attribute (model/size/colour/dimension) + detail. Apparel: Brand Garment Gender Colour Size; electronics: Brand Model Type Feature; homeware: Product Material Dimension Brand.
• Include the terms people actually search for (look at the search terms in your Shopping campaign): "men's running trainers" is worth more than "Air Zoom Pegasus 41".
• No shouty capitals, no promotional copy ("sale", "free delivery"), no symbols.
• Variants: each variant's title carries its colour/size → relevance for the query "blue t-shirt size M".
• You can rewrite them without touching the shop using Merchant Center rules or a supplemental feed (below); and you can automate it with the Suite (Shopping Ninja generates optimised titles from the spreadsheet — lesson 6).

Classification

• google_product_category: Google's taxonomy (ID or full path). Correct = showing up on that category's queries; generic ("Apparel & Accessories") = poor relevance. Choose the most specific one.
• product_type: your own free taxonomy (Home > Kitchen > Frying pans). Google uses it for relevance and you use it to structure campaigns (product groups by product_type). In practice it is compulsory for any serious structure.

Price and promotions

• sale_price (+ sale_price_effective_date): the reduced price; the struck-through price and the discount are shown; it triggers the "price drop" badge.
• shipping (per country/region, service, price) and shipping_weight: overrides the account setting per product; it must match the checkout.
• unit_pricing_measure / unit_pricing_base_measure: unit price (required for certain products in the EU).
• cost_of_goods_sold: cost of acquisition → Merchant Center works out the gross margin in its reports (lesson 5).

Images and detail

• additional_image_link (up to 10): alternative views, in use, close detail → more clicks and a better listing.
• product_highlight (up to 10 short bullet points) and product_detail (section/attribute/value): they feed the detailed listing and relevance.
• color, size, size_type, size_system, material, pattern, age_group, gender: Shopping filters; required in apparel.

Variants done properly

A product with sizes and colours = one row per variant, all sharing the same item_group_id, each with its own id, title (carrying the attribute), link (to the variant), image_link (of its colour), price, availability, color, size. Google shows one and groups the rest. Classic mistake: a single row for every size → "size not available" on the website, disapproval on availability.

Custom labels: the structural lever

custom_label_0 … custom_label_4 (up to 100 values each): free values that only you see and that you use to partition campaigns and bid differently: margin (high/medium/low), performance (top/middle/tail), seasonality, newness, stock (high/low), price (bands), own brand/third-party. Fill them from the ERP or with rules. Without labels, Shopping structure is limited to category/brand/type.

Data sources (how the feed arrives)

Structured data (schema.org/Product with offers) on the product page: Google uses it to verify price and availability; getting it right reduces disapprovals.

Supplemental feeds and rules: changing the feed without touching the shop

• Supplemental source: a file/Sheet with id + attributes that get added or overwritten on top of the primary feed (rewritten titles, custom labels, categories, highlights). This is the specialist's tool: the shop exports what it can; you fix the rest.
• Data source rules: transformations inside Merchant Center — "set title to brand + + title", "if product_type contains X, custom_label_0 = Y", "replace within description", "extract the size from the title". Powerful for simple things; for complex logic, Sheets + a script.
• Precedence: rules > supplemental > primary (configurable).

Common feed errors and their fixes

💡 Ninja trick: work the feed like a campaign, not like a file: measure which products get impressions and which do not (the products report), look at the search terms that trigger each one and rewrite the titles using those words. In real accounts, rewriting the titles of the 50 highest-spending products with the "brand + product + searched attribute" structure raises impressions and CTR without touching the bid — and that is exactly what Shopping Ninja automates in its feed titles tab (lesson 6).

What to remember

• The feed is the ad: title, category, product type and images decide impressions and clicks.
• Required attributes exact and synced with the website (a stable id, price with VAT, availability, a real GTIN or identifier_exists: no).
• The ones that move performance: title, a specific google_product_category, product_type, sale_price, additional images, highlights/detail, variants with item_group_id and custom labels for structure.
• Sources: platform/file to start with, the API for minute-by-minute price and stock, structured data + automatic item updates as a safety net.
• Supplemental sources and rules = optimising the feed without touching the shop.