Entity SEO for B2B brands: the knowledge graph and AI answers

An entity is a thing a search engine can identify on its own: a company, a product, a person, a place. Google describes its Knowledge Graph as a database of billions of facts about people, places and things, compiled from public sources, data it licenses and facts that content owners provide. Knowledge panels are created automatically from it.

AI assistants answer in entities. When a buyer asks for vendors in a category, the answer is a list of named companies with short descriptions. If the system is not sure which company your name refers to, or holds conflicting facts about it, you are either left out or described wrongly. Bing's February 2026 guidelines say the same thing in plainer terms: keep entity names consistent.

B2B brands are especially exposed. Names are often generic words or acronyms, product names change with each launch, and acquisitions leave old brand names on partner sites for years.

We start every GEO engagement with this table. Each row is a fact engines need to get right, and where they read it.

Google's Organization documentation, updated in September 2026, says the markup helps Google understand your administrative details and disambiguate your organization from others, and can influence the logo and knowledge panel. None of the properties are required; you add the ones that apply. It belongs on the home page or a single page that describes the organization, such as About, not on every page.

1. Add Organization markup in JSON-LD, which Google recommends, on the home page or About page.
2. Include name, alternateName if people use one, legalName if different, url, logo, description and foundingDate.
3. List sameAs links to the profiles you control: LinkedIn, Crunchbase, your G2 or marketplace listings, YouTube and Wikidata if an item exists.
4. Add address and contactPoint for real locations and real support channels only.
5. Add identifiers through iso6523Code (Google lists prefixes for DUNS, GLN and LEI) or vatID where you have them.
6. Make sure every value matches what visitors can see on the page. Google warns against marking up information that is not visible.

Structured data is a clarifying signal, not a shortcut. Our article on whether schema markup helps AI search covers what the evidence says about citations.

Buyers rarely ask an assistant about a company in the abstract. They ask about a product for a use case, or about the people behind it. Give each product a stable page with the current name in the title, the category in the first sentence and links to documentation and pricing. When you rename a product, say so on the page ("formerly X") and redirect the old URLs, so engines can connect the two names.

For executives who speak for the company, a profile page on your site that matches their LinkedIn headline helps engines link the person to the organization. Google supports Profile page markup for pages that focus on one person or organization affiliated with the site.

If Google shows a knowledge panel for your company, an official representative can claim it and suggest changes. You can check whether Google already holds an entity for you with the Knowledge Graph Search API, which returns schema.org typed results; Google notes the API is not suitable as a production-critical service, so use it for checks, not for systems.

Wikidata and Wikipedia are strong confirmation sources, but both have notability and conflict-of-interest rules. We do not write promotional entries for clients. Where an item already exists, we help correct facts through the proper channels with published sources. Our guide to fixing wrong AI answers about your company walks through that process.

An entity becomes trustworthy when independent sources agree on it. That means partner directory pages, association listings, event speaker pages, customer case pages on partner sites and media coverage that use your current name and category. Our AI shortlist authority work maps and fixes those sources.

Consistency beats volume. Ten profiles that all say the same thing help more than fifty that describe five different companies.

Entity work is slow to show in rankings, so we track direct signals.

• The Knowledge Graph Search API returns your company, with the right type and description.
• Assistants describe you accurately when asked "What does [company] do?" in a fixed monthly prompt set.
• Branded queries show the right logo, links and panel in Google and Bing.
• Category prompts start to name you alongside the competitors you meet in deals.
• Old product and brand names stop appearing in AI answers.

Write the canonical description first, then fix your own site, then the profiles you control, then third-party sources. Most of the work is editorial and can be done in four to six weeks.

Our generative engine optimization service includes the entity audit and the schema work.

For companies based in Quebec, our page on GEO for Montreal companies covers bilingual entity consistency in English and French.

If assistants confuse you with another company or describe an old version of your business, talk to our team.