Entity SEO: making ChatGPT and Google know exactly who your company is

Entity SEO is the work of making sure Google and AI assistants recognise your business as one clearly defined company, with the same facts attached to it on your own site, in official records and in independent sources. Get it right and ChatGPT, Gemini and Google describe you accurately and feel safe recommending you. Get it wrong and they repeat old addresses, confuse you with a namesake, or leave you out.

Ask ChatGPT who your company is. If you’ve been trading for 20 years, there’s a fair chance the answer gets something wrong: the founding date, the head office you left in 2019, a director who retired, or a mix-up with a firm of the same name in another county. That’s an entity problem, and in AI search it’s expensive, because assistants recommend businesses they can identify with confidence.

Entity SEO is how you fix it. This guide covers what it is, why it matters more now, and the eight steps we’d take for an established UK business, including the question everyone asks: should you have a Wikipedia page?

Want ChatGPT and Google to describe your business correctly? Castle’s GEO team audits how AI tools see your company and fixes the sources they rely on.

Explore our GEO services

TL;DR

• Entity SEO makes sure search engines and AI models recognise your business as one distinct thing with consistent, verifiable facts attached.
• Assistants name businesses they can describe accurately and check against more than one source, and entity sources such as Wikipedia and LinkedIn are among the domains AI tools cite most.
• Start with one signed-off fact sheet, then copy it into your About page, Organization schema and every official record you control.
• A factual Wikidata entry is achievable for many established firms. A Wikipedia page usually isn’t, and paying someone to create one is a false economy.
• Check what ChatGPT, Gemini, Perplexity, Copilot and Claude say about you every month and fix the sources behind each error.

Table of Contents

• What is entity SEO?
• Why entities matter more in AI search
• The eight-step entity checklist
• How to tell if it’s working
• Common entity mistakes
• Our perspective
• Is AI describing your business correctly?
• FAQ
• Sources

What is entity SEO?

Entity SEO is the work of making sure search engines and AI models recognise your business as one distinct, well-defined thing, with consistent facts attached to it everywhere they look. Google calls these things entities in its Knowledge Graph. Language models build a similar picture from everything they’ve read and retrieved about you.

Traditional SEO asks “does this page rank for this keyword?” Entity SEO asks “does the machine know who we are, what we do and why we’re credible?” You need both, but the second one decides whether an assistant is comfortable naming you.

Why entities matter more in AI search

An assistant answering “best commercial property solicitors in Leeds” has to name three or four firms and stand behind them. It picks businesses it can describe accurately and verify from more than one source.

Where those sources come from matters. Wikipedia is ChatGPT’s second most-cited domain in Ahrefs Brand Radar data compiled by Obsurfable, and LinkedIn and Wikipedia both sit in the top four of Peec AI’s analysis of 30 million AI citations. Those are entity sources: they describe who an organisation is, not what it sells this month.

Established businesses have an advantage here. You have history, coverage, people and records a start-up can’t fake. The problem is usually that it’s scattered and inconsistent.

The eight-step entity checklist

1. Write down your canonical facts

Before changing anything, agree one fact sheet: registered name, trading name, company number, year founded, head office address, phone number, leadership team, core services, sectors served, accreditations and awards. Every later step copies from this sheet, so get it signed off by someone senior.

2. Put those facts on your own site, in words and in markup

Your About page should state the facts plainly. “Founded in 2006 in Chester, 60 staff, specialists in industrial roofing for warehouses and factories” gives a model something to work with. “Passionate about delivering excellence” gives it nothing.

Then add Organization (or LocalBusiness) structured data to your homepage, following Google’s organisation markup guidance. The properties that matter most for entity work are name, legalName, url, logo, foundingDate, address and sameAs, the list of your official profiles elsewhere. A minimal version looks like this:

{
  "@context": "https://schema.org",
  "@type": "Organization",
  "name": "Example Roofing",
  "legalName": "Example Roofing Ltd",
  "url": "https://www.exampleroofing.co.uk/",
  "logo": "https://www.exampleroofing.co.uk/logo.png",
  "foundingDate": "2006",
  "sameAs": [
    "https://find-and-update.company-information.service.gov.uk/company/01234567",
    "https://www.linkedin.com/company/example-roofing"
  ]
}

Make sure the markup says exactly what the page says. Schema that contradicts the visible content does more harm than no schema.

3. Align the official records

Next, make every record you control match the fact sheet: Companies House, Google Business Profile, Bing Places, Apple Business Connect, LinkedIn, and any register your sector has (the SRA for solicitors, the FCA register for regulated firms, Gas Safe, the ICO data protection register). Old addresses after an office move are the most common error we find, and assistants repeat them for years.

4. Get a factual Wikidata entry

Wikidata is the structured database behind Wikipedia, and its notability policy is far lower than Wikipedia’s: an item needs to be clearly identifiable and described by serious, public sources. For many established firms that’s achievable. Keep the entry strictly factual, reference every statement, and link it to your official website.

5. Decide honestly whether you should have a Wikipedia page

Most businesses shouldn’t try. Wikipedia’s notability guideline for companies asks for significant coverage in multiple reliable sources that are independent of the company. Your own press releases, directory listings and paid features don’t count.

Paying someone to create the page is a false economy. Wikipedia requires paid contributors to disclose who’s paying them, and undisclosed promotional articles about companies tend to be deleted, sometimes with an awkward edit history left behind. If you genuinely meet the bar, an independent editor may create the page from the coverage that exists. If you don’t, earn the coverage first.

6. Make your people visible

Leaders and experts are entities too. Give your directors and specialists proper bio pages, link them to their LinkedIn profiles, attach them as authors to the content they’re responsible for, and list them consistently on speaker pages and award entries. An assistant that knows your managing director is quoted regularly in your trade press has one more reason to trust the business.

7. Earn independent corroboration

Models believe what several unconnected sources agree on. Trade press interviews, association memberships, awards, case studies published by partners, and reviews on the platforms your customers use all corroborate the facts on your own site. One accurate mention in your industry’s main publication is worth more than fifty directory listings.

8. Find and fix what the assistants get wrong

Ask ChatGPT, Gemini, Perplexity, Copilot and Claude “Who is [company]?” and “What does [company] do?” Save the answers. For each error, find the source it probably came from (an old directory, an outdated press release, a namesake’s website) and fix or update that source. Then check again next month. Corrections take time to filter through, especially for models that rely on training data rather than live search.

How to tell if it’s working

You’ll see it in four places: a Knowledge Panel that appears for brand searches and shows the right details, assistants describing you accurately and consistently, your business being named for category questions it was missing from, and fewer customers arriving with wrong information. Track the same brand and category prompts monthly so you can see the change rather than guess at it.

Common entity mistakes

• Three names for one business. “Smith & Co”, “Smith and Company” and “Smith Consulting Ltd” across different profiles look like three entities. Pick one public name.
• Leaving old offices online. Every listing with a previous address is a vote for the wrong answer.
• Ignoring namesakes. If another business shares your name, make your location and specialism unmistakable everywhere, and use sameAs to point at the records that are really yours.
• Buying a Wikipedia page. Covered above. It usually ends badly.

Our perspective

We think entity work is the most underrated part of AI search for established businesses. The firms that lose out in ChatGPT answers usually have plenty of credibility and very little consistency: twenty years of history scattered across old listings, three versions of the company name and an About page that says nothing a machine can use.

So we start with the dull part: one fact sheet, copied everywhere, then checked against what the assistants actually say. We’d steer most clients away from chasing a Wikipedia page and towards a factual Wikidata entry, visible experts and independent coverage, because that’s the route that holds up when a model or an editor checks the sources.

Is AI describing your business correctly?

If you’ve just asked ChatGPT who you are and didn’t like the answer, start with our 40-point AI visibility audit, then read our guide to generative engine optimisation and how LLM citations work.

Require AI SEO, GEO, or digital marketing and AI automation within your business? Book a free appointment with a senior consultant. Or see our GEO services.

Talk to our team

FAQ

What is an entity in SEO?

An entity is a distinct, identifiable thing, such as a company, person, place or product, that search engines and AI models recognise and store facts about. Entity SEO makes sure your business is recognised as one clear entity with consistent, verifiable facts attached to it across your website, official records and independent sources.

Does schema markup help with ChatGPT and other AI assistants?

Schema helps machines read your facts unambiguously, and Google uses Organization markup to understand your business. There’s no public confirmation that ChatGPT reads schema directly, so treat it as one part of entity work alongside consistent records, clear About pages and independent coverage, not a shortcut on its own.

Should my business have a Wikipedia page?

Usually not. Wikipedia expects significant coverage in multiple reliable sources that are independent of the company, and paid contributors must disclose who’s paying them. A factual, referenced Wikidata entry is a more realistic goal for most established firms, alongside independent press coverage.

How long does entity SEO take to work?

Fixing your site and official records takes a few weeks. Search engines usually reflect those changes within a month or two. AI assistants that rely on live search can update quickly, while answers drawn from a model’s training data may take until the next model update, so expect a gradual improvement over several months.

Sources

• Obsurfable, Top domains cited by LLMs (August 2026), compiled from Ahrefs Brand Radar
• Peec AI, Most cited domains in AI search
• Google Search Central, Organization structured data
• Companies House, Find and update company information
• Wikidata, Notability
• Wikipedia, Notability (organizations and companies)
• Wikipedia, Paid-contribution disclosure

Recommended

• The AI visibility audit: a 40-point checklist for UK businesses
• LLM Citations: What They Are and How to Earn Them
• How to rank on ChatGPT: 10 tactics that get businesses recommended