Skip to main content
A brand is the maker of a product — the label under which it’s sold. In MerchantOps a brand is reference data: it identifies the brand and its official site, and it is shared across the platform rather than owned by any one organization. For the wider vocabulary, see Core concepts.

Brands are shared reference data

Brand identity is the same no matter who is selling the product, so brands are always shared across the platform. Unlike products, variants, and pricing — which are scoped to your organization — a brand carries no per-organization ownership. There is one canonical identity for a brand that every organization references.
Because brands are shared identity, editing a brand is a change to reference data, not to your private catalog. Your product data (properties, variants, pricing, publish status) stays scoped to your organization.

How products reference a brand

Each product points at the brand it belongs to. The brand supplies identity — the name shown in the catalog and the official site that enrichment uses when it looks for authoritative product data.
brand identity
The brand’s name as it appears in the catalog.
enrichment source
The brand’s website. Enrichment discovers and scrapes the brand site to pull authoritative content. On identity and factual fields — name, brand, vendor ID, product page URL, categories, images — the scrape is the first-choice source.
An accurate brand and site produce better enrichment: the brand site is the first-choice source for identity and factual fields, and it anchors every other step of the run. If a product’s brand is missing or wrong, enrichment has a weaker anchor to work from.
The brand site is not the top source for every field. Descriptive and SEO content — description, features, and the meta fields — prefers freshly generated copy by default, with the brand-site scrape ranked next. Priority is configured per property; see How enrichment works.

Brands are registered automatically — and sometimes need a human

You rarely create a brand by hand. A brand that MerchantOps hasn’t seen before is registered for you when it shows up: on a CSV upload, when a product is created from a URL, or during enrichment. Registration includes finding the brand’s official site, and that is where a person occasionally has to step in. MerchantOps works the site out two independent ways — it asks a language model, and it runs a web search — and when the two disagree it refuses to guess. Instead it records both candidates and flags the brand for review. That refusal is deliberate: for an ambiguous short brand name, both sources can confidently return a different company, and picking one at random would point every future enrichment run at the wrong website.

The review queue

Flagged brands collect in Settings → Catalog Settings → Brand URLs. Each one shows the brand and the two candidate URLs, and you resolve it by picking one — or by typing a third URL, which is the answer whenever both candidates are wrong. Resolving a brand does three things at once: it sets the chosen URL as the brand’s official site, marks the brand verified, and clears the review flag. It requires permission to verify brands.
An unresolved brand has no site on record, so enrichment falls back to guessing a domain from the brand’s name — which cannot work for exactly the ambiguous names that got flagged in the first place. Products of a flagged brand keep enriching, but without the brand-site scrape that supplies their identity and factual fields. Clear the queue before a large enrichment run.
Because brands are shared reference data, resolving one is a change everybody sees. You are correcting a fact about the brand, not setting a preference for your organization.

Aliases and country URLs

A brand can also carry aliases — alternative spellings that should resolve to the same brand, so an incoming TNF and The North Face don’t become two brands — and more than one URL, each tagged with a country, for brands that run separate regional sites.
Aliases, additional country URLs, and marking a brand verified outside the review queue are available through the API rather than as screens in the dashboard today. The dashboard covers the review queue itself.

When brand-site scraping is paused

A brand’s site doesn’t always cooperate. Some sites sit behind bot protection that answers with a CAPTCHA or challenge page instead of the product page; others serve a page that turns out to describe a different product. MerchantOps tracks those outcomes per brand and, past a point, stops scraping that brand for a while rather than retrying into a wall. Every scrape counts against your plan’s search allowance whether or not it came back with anything usable, so a brand that cannot be read is a brand worth pausing. There are two tiers, and they’re deliberately far apart: The difference is what each one tells you. A wrong-product match, or a page that merely looks like a block page, is often specific to one product or one search, and the next product may scrape fine — so it takes a run of them, and the pause is short. A site actively answering with a bot-protection challenge is a property of the whole site: it isn’t about the product that was asked for, and nothing about trying again a minute later changes the answer. One is enough, and the pause is long. When the pause expires, MerchantOps doesn’t resume at full speed. It spends a single attempt to see whether the site is reachable again. If that one attempt is challenged too, the brand goes straight back to a full pause — so a site that stays closed costs about one attempt a month instead of a steady drip of them.
While a brand is paused, its products still enrich — from the lakehouse and generated content — but no fresh brand-site data is available to them. Identity and factual fields that normally come from the brand site can stay empty, and the completeness score for that brand’s products is measurably lower during the pause. That is a deliberate trade against paying for attempts that cannot succeed. Products enriched during a pause carry a scrape warning naming the brand and the date the pause lifts, and re-running them after that date picks the brand site back up.
The pause applies to the brand’s site scrape only. Nothing else about the brand changes: it stays valid reference data, products keep pointing at it, and publishing is unaffected.

Products & variants

How a product references its brand.

Core concepts

The lakehouse, enrichment, and the rest of the vocabulary.