What are product attributes?

Someone buying a backpack does not decide by the product name but by its attributes: how many liters it holds, how much it weighs, whether it takes the rain, whether a 15-inch laptop fits. Here "attributes" are not the ones on an HTML tag or the "brand attributes" of a marketing plan, which are about how a company is perceived: they are the concrete data that describes an item and can be checked. They matter because your store's filters depend on them, along with what you send to Google and what an AI assistant can compare when someone asks for the best 30-liter backpack.

How it works

An attribute is a name and value pair: "material: merino wool", "capacity: 30 liters", "compatibility: iPhone 15". What makes it an attribute, and not a sentence, is that it lives in a field of its own. A commerce platform stores it in the product's fields, in the options of its variants, or in custom fields, like Shopify's metafields or BigCommerce's custom fields; a large catalog is usually managed in a PIM. From that field, the same value reaches three places:

• The store's filters. The faceted navigation on a category page can only filter by what is in a field. "Waterproof" written in the description does not show up as a filter.
• The page's markup. Schema.org defines properties in its Product type for the most common attributes, such as color, material, size, pattern and weight, and a generic one, additionalProperty, for any characteristic without a property of its own.⁠^{[1]Schema.org. Product} Google recommends including color, material, pattern and size in the markup of merchant listings, with examples such as "Cotton/Polyester" for the material or "XL" for the size.⁠^{[2]Google Search Central. Merchant listing (Product, Offer) structured data}
• The feed. The product data specification of Google Merchant Center has an attribute for each of those data points, with its own rules. Color is written as text, with no numbers, no codes like #fff000 and no values like "multicolor"; material takes up to two secondary materials separated by a slash, like "cotton/polyester/elastane"; and technical specifications go in product_detail, each with its section name, its attribute name and its value.⁠^{[3]Google Merchant Center Help. Product data specification}

In some categories, the attribute stops being optional. For clothing and accessories, Merchant Center requires color, size, gender and age group on free listings. Color is also required for all clothing targeted at Brazil, France, Germany, Japan, the United Kingdom or the United States, and for any product sold in several colors.⁠^{[3]Google Merchant Center Help. Product data specification}

Attributes that separate variants and attributes that describe

Not all attributes do the same job. Some separate one version of the same product from another: size, color, material. Each combination is one of the product variants, with its own price, its own stock and its own identifier. Google supports six of those axes in its variant markup: color, size, suggested age, gender, material and pattern.⁠^{[4]Google Search Central. Product variant structured data (ProductGroup, Product)} The other attributes describe all the variants the same way: the weight of the backpack does not change with the color.

The distinction has practical consequences. An attribute that separates variants is stored once per variant; one that describes is stored once per product. If you mix them up, you end up with the weight repeated across twelve variants, or with the color written on the parent product when each child has a different one.

In a store

Fictional example. A mountain gear store sells a 30-liter backpack in two colors. Its original page says, in a single paragraph: "30-liter backpack ideal for hiking, light and water-resistant, with space for a laptop". Split into attributes, it looks like this:

Notice two things. "Light" and "ideal for hiking" are not attributes: they are opinions, and they stay in the product description. And "water-resistant" has become a value that can be checked, which is what someone about to compare needs. For the Merchant Center feed, Clione builds a supplemental feed that fills in color, material, size, gender and age group only when the store provides them, without guessing them.

How to check it

1. Open your best-selling category page and note which filters it offers. Every filter that is missing is an attribute that is not in a field.
2. Open three product pages and underline each checkable fact in the description (dimensions, materials, capacities). What only exists in the text is not yet an attribute.
3. Check the markup of one page, the one Google reads to show rich results: run it through the Rich Results Test and see whether the product carries color, material, size or pattern in its markup.⁠^{[2]Google Search Central. Merchant listing (Product, Offer) structured data}
4. Review the Merchant Center diagnostics for clothing and accessories: a missing color, size, gender or age group shows up as a product issue.⁠^{[3]Google Merchant Center Help. Product data specification}
5. Check the units: "30 l", "30 liters" and "30L" on three similar products are three different values to a filter.

Related concept: product description

A product description is running text written for a person: it says who the item is for, how it is used and why to pick it. Attributes are the data that description leans on. A good description does not replace them, it cites them; and a fact that only exists in the description reaches neither the filters nor the feed. When attributes are missing, the work of filling them in is part of catalog enrichment.

To keep going, the catalog and entities reading path puts the other terms in order, from the product page to product taxonomy.