← Insights

Structured data is how a model learns what you are

Schema markup is treated as an SEO chore. It is closer to a declaration of identity, and most B2B sites contradict themselves in it.

Sarthak Jain, Co-founder · · 5 min read

Most teams add schema markup because a audit tool flagged it. They paste an Organization block, tick the box, and move on.

That block is doing more work than they think. It is the machine-readable version of "who is this company, and what is it for" - and when a model resolves your brand into an entity, that is one of the few unambiguous inputs it has.

The fragmentation problem

The common failure is not missing schema. It is contradictory schema.

A company ends up with an Organization name on the homepage that does not match the legal entity in the footer, which does not match the name on the careers site, which does not match the one on an old acquisition's domain that still resolves. Each page is individually correct. Together they describe three or four organisations.

A model reading that has to guess which one you are, or split you across several weakly-supported entities. Neither outcome cites you confidently.

What is actually worth marking up

For a B2B company, the useful set is short:

  • Organization, once, consistently, with the same legal name and sameAs links everywhere it appears.
  • WebSite with the canonical domain, so alternates and staging hosts do not compete.
  • BlogPosting or Article on writing, with a real author and dates. Authorship is one of the clearer authority signals available.
  • BreadcrumbList where the hierarchy is real, because it tells a crawler how the site actually nests.
  • FAQPage only where there are genuine questions and answers on the page. Marking up sales copy as an FAQ is the kind of thing that gets a site's rich results suppressed.

Everything else is usually optional.

Entity mapping is the part people skip

Markup declares what you are. Entity mapping is making sure the rest of the internet agrees.

That means the same name and description on your site, your LinkedIn, your Crunchbase entry, your GitHub organisation and anywhere else the company appears. It means retiring the old spellings rather than leaving them to rot. It is unglamorous, and it moves things that schema alone does not.

How to know it worked

Not by re-running the validator - that only proves the syntax parses.

Ask a model who you are, in a new session, and read what comes back. Does it name the right company, in the right category, doing the right thing? Record the answer with the date. That is the measurement that matters, and it is the same one we run in a baseline audit.

If this describes your situation, start with a baseline.

We document where you stand before recommending anything - so whatever we claim at 90 days has a before to be measured against.