How channels decide what they want

Every channel has its own idea of what a "good" product looks like, which is why your data has to be reshaped for each one. But channels aren't all unique snowflakes. They fall into families, and once you can spot the family, the differences get a lot less intimidating. Here's how to think about what a channel is really asking for.

A lot of channels borrow the same playbook. Google Merchant Center and Meta (Facebook and Instagram) are close cousins: both expect a product feed built around the same core fields, things like id, title, description, link, image_link, price, availability, brand, and gtin. The field names and rules line up so closely that if you've built a feed for one, you're most of the way to the other. Many advertising channels lean on this Google-style schema, so getting it right once pays off across a whole family of destinations.

Other channels don't invent their own schema at all. Instead they adopt a shared industry standard, which is especially common in B2B, retail, and industrial sectors. Standards like ETIM (technical and electrical products), eCl@ss (cross-industry), GS1/GPC, and UNSPSC are large, agreed-upon vocabularies that classify products and define which attributes each kind of product should carry. When a marketplace or retailer says "send us eCl@ss-classified data," you're conforming to that public standard rather than a one-off spec, and the same data can satisfy any partner who speaks it.

Strip away the surface differences and there's a common backbone underneath. Almost every channel wants the same handful of things: a stable identifier, a name or title, a price, an availability status, an image, and some notion of category. The core facts, the attributes that describe your product, barely change from one channel to the next. What changes is how those facts are named, formatted, and grouped.

The variation shows up in the details:

Of all the differences, categorization is usually the toughest. Your own product categories almost never line up with a channel's taxonomy or a standard's class system, so you have to translate between them: your "Trail shoes" might need to become Google's Apparel & Accessories > Shoes, or an ETIM class code. Get it wrong and products land in the wrong place or get rejected. Productsup handles this with the Classification Mapping list for mapping to a classification system, and Partner Taxonomy Mapping for mapping to a partner's categories.

Here's the idea that trips people up. Even within a single schema, your products don't all need the same attributes. A t-shirt needs size and color. A laptop needs screen size and memory. A bottle of wine needs volume and vintage. So "required" isn't one fixed checklist for your whole catalog. It depends on what each product actually is.

Industry standards make this explicit: a product's class defines its own set of attributes, so two products in the same feed can legitimately carry completely different fields. Channels express the same idea through conditionally mandatory attributes, a field that stays optional until a condition makes it required. For example, a marketplace might require a cell_composition attribute only for products that contain batteries. See Conditionally mandatory attributes for how that works.

The takeaway: a feed isn't one tidy rectangle where every row fills the same cells. Different products carry different attributes by design, often driven by their category or class, and that's exactly what the channel expects.

You don't track all of this by hand. An export template already encodes a channel's spec, including its attributes, accepted values, and the analyzer tests that validate them, conditionally mandatory fields included. Add the template, map your data to it, and the readiness score tells you whether you've met what that particular channel decided it wants.