Product attributes: the guide to a complete catalog

What an attribute is and why it matters

An attribute is a product characteristic stored as data: the material of a table, the size of a T-shirt, the power of a drill. The difference from the description is that an attribute has a type and, often, a list of allowed values. That makes it useful for things a paragraph cannot do:

• Filtering in the store: “stainless steel only”.
• Comparing products side by side.
• Filling the columns each channel requires.
• Feeding the description and the structured data with consistent facts.

Store search depends on this: faceted search, the kind that lets you combine filters, only works when data is structured.

Attribute types

Not all data is the same, and treating it the same way causes many errors. The most common types are:

1. List: one value from a closed catalog (color: red).
2. Multi-select list: several values from the same catalog (uses: wood, metal).
3. Yes/No: a characteristic the product has or not (battery included).
4. Number: a quantity without a unit (pieces per pack).
5. Measurement: a number with its unit (250 ml, 750 W).
6. Free text: what does not fit in a list (care instructions).
7. Date.

Choosing the right type keeps “Red”, “red” and “RED” from ending up as three different colors.

Families: what applies to each product

A drill and a shirt don't need the same data. The family groups the attributes that apply to a type of product and says which are required. Without families, every product looks incomplete or is asked for too much.

Families also organize the work: the content team knows what is left to capture, and completeness is measured against what actually applies.

How to know what each product is missing

A catalog is never complete “in general”: it is complete for a channel. That is why it pays to measure two things:

• Completeness against the family: the attributes that belong to the product.
• Completeness against the channel: the columns each destination requires.

A list of what is missing, product by product, is more useful than a catalog-wide percentage. Google Merchant Center, for example, requires different data depending on the product type and the country, and the same goes for each marketplace.

Common mistakes

• Attributes inside the description. “Made of 2 mm stainless steel” inside a paragraph cannot be filtered.
• Unnormalized values. “M”, “Medium” and “Med” as three different sizes.
• Measurements without units. “250” does not say whether it is milliliters or grams.
• Attributes nobody uses. Every extra field is capture work; if nobody filters it and no channel asks for it, it is not needed.

Where to start

You don't need to define every attribute in the catalog at once. An order that works:

1. Pick one type of product that sells a lot or that a channel often rejects.
2. List what buyers ask about that product and what the channel where you sell the most requires. That union is the family.
3. Decide each attribute's type and, for lists, their allowed values. Merge the spelling variants you already have into a single value.
4. Fill the attributes from what is already in your spreadsheet, the spec sheets or the descriptions.
5. Measure what is missing product by product and fix the best sellers first.

Once the first family is ready, the rest come faster: many attributes, such as brand, color or material, repeat across families.

How Bruxel handles it

In Bruxel attributes are typed —list, multi-select, yes/no, text, number, date and measurement— and grouped into families that say what is required. Completeness is measured product by product and against each channel's columns. You can fill attributes with AI in batches from the product data, and nothing is published until someone reviews it.

If you are just starting, the post What is a PIM? explains the context, and the PIM page shows what it looks like in the product.