What counts as an entity anchor?

A record about your company that lives somewhere other than your own website and can be cross-checked: a company page on a professional network, a directory or database record, a knowledge-base entry, a listing that links back to you. The point is not the link but the confirmation — something outside your control saying you exist and are what you claim.

Should we create a Wikidata or Wikipedia entry?

Not early, and not as a first move. Both have notability standards, and a rejected or deleted entry is worse than none because it is a public record of failing the bar. Our own rule is to wait until at least two independent editorial publications cover the company, then attempt it once, with the affiliation disclosed.

How many profiles are enough?

Consistency matters more than count. Three profiles that agree on name, description and founding year do more than ten that contradict each other, because contradiction is what makes a model hedge. Start with the two or three that are genuinely relevant to your category and get them identical.

Entity SEO and the knowledge graph

Entity SEO is the work of making a machine able to confirm that your company is a specific, real organisation rather than a string of characters. It is the single most common failure in our dataset: close to 100% of the 100 audited sites had no entity anchor at all.

why it decides whether you get named

When a model composes an answer it has to choose whom to name. Given two candidates, one it can confirm and one it cannot, it names the one it can — or it hedges about the one it cannot, which in regulated categories reads as a warning to your buyer.

This is also why brand recall and category recall come apart so sharply. In our dataset 93% of brands were described accurately when asked about directly, while 4.5% were named when a buyer asked who to pick. Recognition is not confirmation.

what an anchor actually is

A record outside your domain that describes the company and links back to it.
Consistent identifiers — the same legal-ish name, the same founding year, the same description, wherever a machine can read them.
A person attached to the company, with their own verifiable presence, linked in both directions.
Markup that connects it all: one organisation node, referenced everywhere, with sameAs pointing at the records that exist.

the order that stops the waste

Fix readability first. Entity work on a site a crawler cannot read is invisible — you are describing a company whose pages return nothing.
Make the site agree with itself. One name, one description, one founding year across the markup, the llms.txt and the page copy. Internal contradiction is the cheapest thing to fix and the most common.
Create two or three real records in places relevant to your category, filled with identical copy.
Connect them through sameAs once they resolve, not before.
Only then consider knowledge bases with notability standards.

the collision problem nobody reports

If something else shares your name, a model may confidently describe that instead of you, and no SEO tool will tell you. We hit this on our own domain: the string collides with a betting brand, an anime series, a music project and two unrelated agencies, and engines were repeating the wrong one.

There is no clever fix, only an explicit one: qualify the name everywhere (never the bare string), and publish a disambiguation section in your llms.txt naming what you are not. It is four lines and it is the highest-leverage entity work available to a young brand.

how to check where you stand

Ask an engine to describe your company, then verify every fact it states. Ask it who your competitors are and see whether the list makes sense. Search your exact brand string and count how many of the results are actually about you. If the answer is a small fraction, your entity problem is bigger than your content problem.

The markup side is covered in schema and JSON-LD for AI search; the measurement side is the audit.

entity SEO and the knowledge graph