Brand name mismatch is the quiet killer

Your title tag says one name, your schema another, your LinkedIn a third. To a machine that is three companies, none of them well described.

This one costs a fixed ten points on our scan and it is almost always news to the person reading the result. Nobody sets out to have three names. It happens the way most entity problems happen — one decision at a time, each of them locally reasonable.

How it happens

The title tag gets written for search: Northstar | Product Analytics for SaaS Teams. The JSON-LD gets written by a developer copying from the legal docs: Northstar Analytics, Inc. The LinkedIn page was created before the rebrand: Northstar Labs. The footer says © 2026 Northstar Analytics Inc. The X handle is @northstarhq with the display name Northstar 🚀.

Every one of those is a defensible individual choice. Together they present a machine with four candidate names and no statement of which is canonical.

What a machine does with it

It has to decide whether these are one entity or several, and it decides on evidence. Shared URL, shared logo, a declared sameAs link — those merge the records. Nothing but name similarity, and it hedges.

A hedge looks like one of these, and you will recognise them:

• "There are a few companies operating under this name…"
• "I don't have specific information about Northstar Analytics."
• A confident description of a completely different company with a similar name.
• Being left out of a shortlist entirely, with no explanation, because the model was not confident enough to name you.

That last one is the expensive one, and it is invisible. You do not get told you were omitted for low confidence.

The canonical set

Decide these five things once, write them down somewhere your team will find them, and treat them as fixed:

The row people get wrong is the second one. Legal name belongs in legalName and in your terms of service. Putting , Inc. in your title tag and your schema name creates a mismatch with the profiles that sensibly omit it.

The title-tag question

"But I need keywords in my title tag." You can have both. Put the brand name in a consistent position with a consistent separator:

Product Analytics for B2B SaaS Teams | Northstar Analytics

Brand last, one separator, same on every page. Extraction handles this fine — what it handles badly is Northstar on the homepage, Northstar Analytics Inc on the pricing page, and NSA on the blog.

Fixing it in an afternoon

1. Write down the canonical set above. Twenty minutes, and it is the actual decision.
2. Grep your templates for every hardcoded variant of the name. Site title config, footer, email templates, OG tags, the PDF export nobody remembers.
3. Update the schema so name is the brand and legalName is the legal entity.
4. Update every profile to the same name, description, URL and logo.
5. Declare the links by putting each profile URL in sameAs. This is what converts "probably the same company" into "stated to be the same company."

Step 5 is what makes the rest hold. Without it you have consistency and a machine still has to infer the connection; with it you have consistency and an assertion.

The rebrand case

If you renamed in the last couple of years, consistency alone is not enough — the old name is in training data and will keep surfacing. Do not try to erase it. State the relationship:

• Keep the old name as an alternateName in your schema.
• Say "formerly Northstar Labs" in the About page, once, near the top.
• Keep the old domain 301-redirecting rather than expired or parked.
• Update the profiles, and where the platform allows it, keep a line about the previous name.

You want a machine that encounters the old name to arrive at the current record. A dead end is how you end up with two half-known companies and neither of them recommended.

Checking

Ask any assistant, in a fresh session: "What is <brand name>?" A crisp, correct, category-accurate answer means the record resolves. A hedge means it does not — and now you know what the ten points are for.