What is an entity in AI search and why your business is invisible without one

The search result your business is missing, and why.

A quick test: ask ChatGPT to describe your business right now.

Before you read another word, open a new tab and type this into ChatGPT, Perplexity, or Google's AI Mode: "What do you know about [your business name]?"

What comes back? If you run a well-known brand, you might see a confident, accurate summary: your industry, your location, what you offer, maybe even a founder's name. 

However, if you're like most small and mid-size business owners, one of three things probably happened: the AI described a completely different company with a similar name, it gave you a vague non-answer hedged with "I don't have reliable information on this," or it drew a complete blank.

That result is basically where you stand with AI right now and honestly, it's starting to matter more than your Google ranking.

If ChatGPT drew a blank on your business, it's probably not a content thing or even an SEO thing,  AI just doesn't know you exist. Before AI can recommend anything, it needs to actually understand what it's recommending.

What you probably saw (and why it's a problem)

AI search tools don't work the way Google used to, when someone asked Google a question in 2018, your well-optimised page had a real shot at appearing in the results, the user clicked, landed on your site, and you had your chance.

More and more searches just return an answer, no list of links, no page two. If your business isn't one of the sources AI already understands and trusts, you don't appear in that answer. There's no close enough, you simply don't exist in that moment.

The businesses getting cited aren't always the biggest, they're just the ones the AI actually recognises and trusts. 

The shift from 'strings' to 'things' in plain English

For most of search's history, Google was essentially a very fast librarian matching the words you typed to the words on a page. Type 'accountant Wellington,' get pages with those words. It worked, but it had no idea what anything actually meant, it had no idea whether "Mercury” the planet and Mercury the car are different things. 

That changed when Google introduced its Knowledge Graph in 2012 with a now-famous phrase: "things, not strings." Instead of matching text, Google began building a map of the real world, cataloguing actual entities (people, businesses, places, concepts) and the relationships between them. Mercury the planet orbits the Sun and Mercury the car was made by Ford. These aren't just word patterns, they're facts about distinct things.

Large language models like the ones powering ChatGPT and Perplexity take this further. They don't index pages at all in the traditional sense, they've learned from vast amounts of text to understand what things are and how they relate. When they generate an answer, they're reasoning from that understanding, not searching a database of URLs.

What this means for your business is simple: if you haven't been established as a recognised entity in the systems these AI tools draw from, you're not a "thing" to them. You're just a string of words and strings don't get recommended.

What is an entity in AI search? A plain-English definition

Entity definition: person, place, organisation, concept

An entity, in the context of AI search, is any real-world thing that can be uniquely identified and distinguished from everything else.

That definition is deliberately broad because entities are broad. Google's Knowledge Graph recognises four primary types:

People: a named individual with a distinct identity. Jacinda Ardern, Elon Musk, your company's founder.

Places: a specific, locatable geography. Wellington, the Sahara Desert, your store's address.

Organisations: a structured body with a name, purpose, and persistent existence. Apple Inc., the Reserve Bank of New Zealand, your business.

Concepts: an abstract but well-defined idea. Inflation, machine learning, climate change.

The key word across all four is distinct. An entity isn't just a category or a keyword, it's a specific, nameable thing that exists independently of any single webpage or document and your business, if properly established, is one of these things. A search query like "best accountants in Wellington" is not, it's a string, a pattern of words, not a thing in itself.

This distinction is the foundation of everything that follows. AI search doesn't rank pages, it reasons about entities and decides which ones are relevant, credible, and worth mentioning.

Keywords vs. entities: what's the actual difference?

A keyword is flat, it's the surface text. An entity is three-dimensional, it has attributes, relationships, and a history that AI systems can reason about.

When you optimise a page for the keyword "Wellington accountant," you're telling Google: this page contains these words. When you establish your firm as an entity, you're telling Google: this business exists, it's located here, it employs these people, it has been mentioned by these credible third parties, it operates in this industry, and it has been consistent about all of that for years.

A keyword tells Google what words are on your page and an entity tells it who you actually are.

Keyword optimisation still matters as it's how you get found in traditional search but in AI search, it's entity recognition that determines whether you get cited, recommended, or summarised in a generated answer. A business with a modest keyword footprint but a well-established entity can outperform a heavily optimised competitor in AI results simply because the AI knows who it is.

A Website Isn't the Same as an Entity: Real examples

Search for "Apple" in Google and something interesting happens before you even finish typing. As soon as you enter "app", Google begins suggesting Apple as a distinct entity, complete with its logo and classification as a technology company.

This isn't just autocomplete, Google isn't predicting letters it's recognising an entity.

Search for "New Zealand" and the same thing happens, Google understands it as a country with a population, government, geography, cities, and relationships to countless other entities.

Now search for your business name, what appears?

For many businesses, Google can find their website, but it doesn't necessarily understand the business as a distinct entity. It sees pages, profiles, and listings, but has far less confidence about who the business is, what it does, and how it relates to other entities across the web.

You're described by your pages rather than known in your own right.

The difference isn't simply size or popularity, it's the depth and consistency of information available about the entity. The more clearly your business is understood, the easier it becomes for search engines and AI systems to identify, connect, and recommend it.

To Google and AI systems, there's a significant difference between being recognised as an entity and simply having a website.

The three characteristics that make something an entity

Not everything qualifies as a recognised entity, and understanding why helps you build toward it strategically. AI systems and knowledge graphs tend to recognise entities that demonstrate three core characteristics.

Uniqueness. The entity can be distinguished from all others. Not just by name, but by a combination of attributes. If two businesses share a name, the one with more distinct, consistent signals (location, founding date, specific services, associated people) will be the one the AI treats as the canonical entity.

Persistence. The entity exists across time and across multiple independent sources. A business mentioned once in a single directory isn't persistent but a business consistently described across its own website, a Google Business Profile, industry publications, a Wikidata entry, and customer review platforms… that's persistent.

Relationships. The entity is connected to other known entities. Your business is located in a known city, it's run by a named person, it operates in a defined industry and it has served clients who have their own presence online. These relationships aren't just metadata, they're how AI systems place you on the map of the real world and decide whether citing you makes sense.

Get all three right and AI systems stop treating you like a search term and start treating you like a real business.

How AI systems identify and use entities

Inside Google's Knowledge Graph: 500 billion facts about 5 billion entities

Google's Knowledge Graph is, at its core, a colossal database of real-world things and the relationships between them. As of the last published figures, it holds over 500 billion facts about approximately 5 billion entities and it grows every day as Google crawls the web, processes structured data, and cross-references new information against what it already knows.

What makes the Knowledge Graph different from a search index is that it doesn't just store documents, it stores understanding. It knows that Wellington is the capital of New Zealand, that New Zealand is a country in the South Pacific, that the South Pacific is part of the Pacific Ocean, and that the Pacific Ocean borders dozens of other countries it also has entities for. Every fact is called a node, every relationship is a connection and the more connections an entity has, the more confidently the Knowledge Graph can describe it, place it in context, and surface it in response to relevant queries.

For your business, getting into the Knowledge Graph isn't about submitting a form or paying a fee, it's about generating enough consistent, credible, cross-referenced signals that Google's systems can confidently say: this is a real organisation and here's what we know about it.

How AI systems decide who to mention

Google's Knowledge Graph is one of the most established entity databases, but it isn't the only system AI tools rely on.

ChatGPT, Perplexity, Gemini, and other AI search platforms each use different combinations of training data, structured knowledge, web content, and live retrieval. The details vary from platform to platform, but the underlying requirement remains remarkably consistent: AI needs enough reliable information to recognise your business as a real, distinct entity.

Whether an answer is generated from a knowledge graph, retrieved from live web sources, or inferred from training data, businesses with stronger entity signals are easier for AI systems to identify, understand, and include in responses.

The lesson isn't to optimise for one platform, it's to build a clear, consistent entity that can be recognised across all of them.

Why being the main topic matters more than being mentioned

AI systems don't treat every mention of your business equally.

A page that's entirely about your business sends a much stronger signal than a page where your name appears briefly alongside dozens of others. AI systems look at context, not just mentions, sometimes referred to as entity salience by AI researchers, it is a measure of how central your business is to a piece of content.

For example, a detailed profile on an industry publication where your business is the focus can contribute more to entity recognition than dozens of directory listings that simply include your name, address, and phone number.

This is one reason old-school citation building has become less effective, being listed everywhere helps establish that your business exists, but it does little to establish what makes your business notable, credible, or worth recommending.

When you're building entity signals, ask a simple question: is my business the subject of this content, or is it simply being mentioned?

The more often your business is the subject, the easier it becomes for AI systems to understand who you are and when you should be included in an answer.

Schema markup and structured data: the signals AI trusts most

What schema markup actually does (without the developer speak)

If entity recognition is about helping AI understand who your business is, schema markup is the most direct way to have that conversation.

Your website contains a lot of useful information: your business name, address, phone number, the services you offer, your opening hours, your founding date. That information is written for humans, a sentence like "We've been serving Wellington businesses since 2009" is perfectly clear to a person reading it but to an AI system crawling your page it's just text, words that might mean something, in a format that requires interpretation.

Schema markup is a standardised vocabulary you add to your website's code that translates that human-readable information into machine-readable facts. The information itself is called structured data, schema is simply the framework that helps search engines and AI systems understand it correctly. 

Instead of hoping the AI correctly infers that you're a locally operating business founded in 2009, schema markup states it explicitly, in a structured format that AI systems and search engines are designed to read.

Think of it as a formal introduction, your website talks to customers, your schema talks to the machine, in terms it actually understands.

Good schema helps search engines understand your business more accurately, pages with comprehensive, accurate schema markup are more likely to be understood correctly, more likely to have their information included in Knowledge Graph entries, and more likely to appear as cited sources in AI-generated answers. It doesn't guarantee visibility, but it removes one of the most common barriers to it.

The 5 schema types every local business should have

There are many different schema types available, but most businesses only need a handful. It can feel overwhelming to get started, but in practice, most small and mid-size businesses need to focus on five core types that cover the vast majority of what AI systems want to know.

Organisation (or LocalBusiness). This is the foundational schema type for any business. It tells AI systems your official name, your website, your logo, your founding date, your physical address, and your contact details. If you only implement one schema type, make it this one, it establishes you as a distinct, named entity with a real-world presence.

LocalBusiness. A more specific extension of Organisation, LocalBusiness adds location-specific attributes like your geographic coordinates, your service area, your opening hours, and your price range. For any business that serves a specific city or region, this schema type is what connects your entity to a place, which is one of the three core characteristics of entity recognition.

FAQPage. If your website includes a frequently asked questions section, and it should, FAQPage schema tells AI systems that this content is structured question-and-answer information. This feeds directly into the kind of content AI tools draw from when generating answers to user queries, and it significantly increases the chance of your content being cited verbatim.

Product or Service. Depending on what your business offers, Product or Service schema gives AI systems a structured understanding of what you actually do including descriptions, pricing where relevant, and associated reviews. This is the schema type that helps AI match your business to "what should I use for X" queries rather than just "who is this company" queries.

Review and AggregateRating. Social proof isn't just for humans, when AI tools reason about which businesses to recommend, credibility signals matter and structured review data is one of the clearest credibility signals you can provide. AggregateRating schema surfaces your overall rating and review count in a format AI systems can read and factor into their understanding of your reputation.

How structured data gets your brand into AI answers

Implementing schema markup doesn't flip a switch that immediately puts you in AI answers, what it does do is remove friction from a process that's already happening.

Search engines and AI systems are constantly crawling, reading, and attempting to understand the web and structured data gives those systems a clearer, more confident signal to work with.

When you add Organisation and LocalBusiness schema to your website, you're making it easier for search engines and AI systems to understand your business. Instead of guessing from the content on the page, they can read key details such as your name, location, contact information, and services in a structured format.

The clearer that understanding becomes, the easier it is for AI systems to recognise your business and include it when it's relevant to a user's question.

One practical note: schema markup is only as useful as it is accurate. Outdated information, inconsistencies between your schema and your visible page content, or schema types that don't reflect your actual business can actively confuse AI systems rather than help them. Implement it carefully, keep it current, and audit it whenever your business details change.

Building your brand entity from zero: a 6-month roadmap

Month 1–2: Establish your entity footprint

The first stage of entity recognition isn't about gaming AI systems, it's about making it easy for them to reach the same conclusion: your business exists, it does a specific thing, and the information about it is consistent wherever they look.

Start with the foundations.

Your Google Business Profile

If you haven't claimed and fully completed your Google Business Profile, start there.

For many local and regional businesses, this is one of the strongest entity signals available. It helps Google understand who you are, where you're located, what services you provide, and how customers interact with your business.

Complete every field you can, add your founding year, business description, products or services, website address, opening hours, and recent photos. The more complete and accurate your profile is, the easier it is for search engines and AI systems to identify your business correctly.

Consistent business information

One of the simplest ways to strengthen entity recognition is to make sure your business is described consistently everywhere it appears online.

Your business name, address, phone number, website URL, service descriptions, and brand positioning should align across your website, Google Business Profile, social media profiles, directories, and industry listings.

AI systems compare information from multiple sources and when those sources disagree, it becomes harder to understand your business accurately and uncertainty increases.

If your business appears as "Smith Consulting Ltd" in one place, "Smith Consulting" in another, and "S. Consulting" somewhere else, you're creating unnecessary ambiguity.

Spend time auditing your online presence and standardising the information wherever possible.

Implement foundational schema markup

Once your business information is consistent, help search engines understand it more clearly by implementing Organisation and LocalBusiness schema markup.

Schema gives AI systems structured information about your business, including your name, location, website, contact details, and services. Rather than relying on interpretation, AI systems can read those details directly.

For businesses beginning their entity-building journey, schema markup is often one of the highest-impact technical improvements available.

Additional entity signals

As your online presence grows, you may choose to strengthen it further through sources such as Wikidata, industry directories, professional associations, media coverage, and other trusted third-party references.

These signals work best when they reinforce information already established on your website, Google Business Profile, and other core business assets.

The goal isn't to appear everywhere. It's to ensure the same accurate information appears wherever your business is mentioned.

Spend the first two months auditing every place your business appears online and standardising your NAP to a single canonical format. This includes your website footer, your Google Business Profile, every directory listing you can find, your social media profiles, and any press mentions where the information can be corrected. It can be boring work, but nothing else holds without it.

Month 3–4: Earn third-party mentions that AI can see

With your entity footprint established, months three and four are about expanding your presence beyond the sources you directly control. This is where most people get stuck, you can't just tick boxes anymore, you actually have to get other people to talk about you.

Not every mention of your business carries the same weight, content where your business is the primary subject tends to contribute far more to entity recognition than content where you're mentioned briefly alongside dozens of others.

Industry features, interviews, and third-party mentions

Once your business information is consistent, the next step is to earn mentions from sources you don't control. This might be an industry publication, a local business journal, a chamber of commerce article, a podcast interview, a case study, or a customer success story published elsewhere online.

The goal isn't to collect links, it's to build independent evidence that your business exists, serves a real market, and is recognised by others in your industry or community.

The more often your business is discussed as the main topic, the easier it becomes for AI systems to understand who you are, what you do, and when you should be included in an answer.

Month 5 - 6: Check what AI knows about your business and close the gaps

By month five, there should be far more information available about your business than when you started. The focus now shifts from building signals to checking what AI systems have learned.

Ask ChatGPT, Gemini, Perplexity, and Google's AI search tools a series of questions about your business:

• What does [business name] do?

• Where is [business name] located?

• What services does [business name] offer?

• Who founded [business name]?

• Would you recommend [business name]?

Pay attention to what the AI gets right, what it gets wrong, and what it doesn't know at all.

Incorrect information often points to inconsistent signals and missing information usually means you haven't provided enough evidence for AI systems to confidently understand that part of your business.

Look for gaps in your third-party coverage

Review the articles, interviews, industry features, and mentions you've earned over the previous months. Ask yourself a simple question: “are these sources actually about my business, or do they only mention it briefly?”. 

Content where your business is the main topic gives AI systems far more context than a passing mention in a directory or list article. If most of your coverage consists of brief mentions, focus on earning a few deeper pieces. A client case study, industry feature, expert article, or business profile can often contribute more to entity recognition than dozens of smaller references.

Review your schema and business information

Check that your schema still matches your business.

Services change, team members change, locations, contact details, opening hours, and business descriptions can all shift over time. If your schema says one thing and your website, Google Business Profile, or directory listings say another, you create conflicting signals.

Review your schema, update anything that has changed, and make sure the structured information on your website matches what people can see on the page.

The entity bootstrap checklist: 11 things AI should know about your business

Before you move on to measuring your long-term visibility, use this checklist as a final audit. These are the eleven attributes that, when consistently established across your website, your Google Business Profile, directories, and your third-party mentions, give AI systems the complete, corroborated picture they need to confidently recognise and recommend your business.

1. Official business name is used consistently everywhere your business appears online

2. Physical address or service area is clearly stated and consistent across platforms

3. Phone number is consistent across your website, Google Business Profile, and directories

4. Website URL, there is a single primary website referenced across all profiles and listings

5. Founding date is clearly stated where relevant

6. Industry and business category is clearly defined so AI systems understand what type of business you are

7. Products or services are described consistently across your website and business profiles

8. Founder or key personnel are named individuals associated with the business

9. Geographic service area, the cities, regions, or countries you serve are correctly stated 

10. Customer reviews and ratings are authentic reviews across Google and other trusted platforms

11. Third-party coverage includes independent articles, interviews, features, or mentions where your business is the primary subject

If you can confidently check all eleven, your entity foundation is solid. If four or five are missing or inconsistent, you've found your roadmap for the next quarter. Entity Definition is the first stage of the AI Visibility Engine™ framework.

How to measure your entity visibility in AI search

The manual test: 5 prompts to run in ChatGPT, Perplexity, and Google AI Mode

Measuring entity visibility doesn't require expensive software or a technical background. The most revealing test you can run costs nothing and takes about twenty minutes and it will tell you more about your current AI search presence than any rank tracker on the market.

Open four tabs: ChatGPT, Perplexity, Google AI Mode, and Gemini. Run the same five prompts in each one, and record every response in a simple spreadsheet. You're not looking for perfect answers, you're looking for patterns.

Prompt 1: "What do you know about [your business name]?" This is your baseline identity test. A well-recognised entity will return a confident, accurate summary of your industry, location, what the business does, possibly key personnel or founding details. A poorly recognised entity will return hedged language, a description of a different business with a similar name, or a polite admission that the AI has limited information. Whatever comes back, write it down exactly.

Prompt 2: "What does [your business name] specialise in?" This tests whether AI systems understand your core offer, not just that you exist, but what you actually do. Many businesses pass the identity test but fail this one, because their entity signals establish presence without establishing expertise. If the AI describes your services inaccurately or too broadly, your schema markup and third-party content likely need more specific, service-level detail.

Prompt 3: "Can you recommend a [your service type] in [your location]?" This is the commercial visibility test, it’s the query closest to what a real potential customer might ask. Does your business appear in the recommended results? If not, which businesses do? This prompt tells you who the AI currently considers the credible, recognisable entities in your category and geography, and how far you have to go to join them.

Prompt 4: "What are people saying about [your business name]?" This tests your reputation signals. A well-recognised business will usually return a confident, accurate summary of customer sentiment drawn from reviews and third-party coverage. A weakly recognised entity will either return nothing useful or, worse, surface an inaccurate impression based on limited or unrepresentative sources. If the AI gets your reputation wrong, look at your reviews, testimonials, and third-party mentions to understand why.

Prompt 5: "Is [your business name] a reputable company?" This prompt forces the AI to make a credibility judgement about your business and the factors it draws on to make that judgement (third-party mentions, review data, longevity signals, association with known entities) map almost exactly to the entity-building work covered in this article. A confident, positive response here is one of the strongest indicators that your entity recognition is in good shape.

Run this full set of prompts once a month, track how the responses change over time as your entity-building work compounds. When the AI stops hedging and just describes your business accurately, that's when you know it's working.

Tools that can help monitor AI visibility

The most useful visibility tool is often the simplest: regularly asking AI platforms what they know about your business. Running the same set of prompts in ChatGPT, Perplexity, Gemini, and Google's AI search experiences each month allows you to track how AI systems describe your business, what information they understand, and where gaps still exist.

For businesses investing heavily in AI search visibility, tools such as Semrush can provide additional insight into how often your brand appears in AI-generated answers and how that visibility changes over time.

Google Search Console is also worth monitoring, while it isn't an AI visibility tool, strong performance in traditional search is often a positive signal that your website is building authority and understanding with Google's systems.

No single tool provides a complete picture, the most reliable approach is to combine regular AI audits with ongoing monitoring of your website's search visibility and brand presence online.

Red flags: what AI getting your business wrong actually means

Not all AI visibility problems look the same, and understanding the specific type of error you're seeing points directly to the fix.

AI describes a different business. This means AI is probably confusing you with another business that has a similar name. The fix is almost always NAP consistency and a more distinctive entity footprint. Adding unique identifiers to your schema markup, your founding date, your specific geographic coordinates, your Companies Office registration number gives AI systems the attributes they need to tell you apart from your namesakes.

AI describes your business but gets key facts wrong. This means AI is finding conflicting information. Somewhere in the information available about your business online, there's a source that contradicts the accurate information, maybe an old address that was never updated, a category misclassification in an early directory listing, or an outdated service description that's been superseded by your current offering. 

Track down the conflicting source, correct it, and ensure your website, Google Business Profile, schema markup, and other business profiles all contain the correct information. AI systems resolve conflicts by defaulting to the most frequently and credibly corroborated version of the facts.

AI knows you exist but won't recommend you. Your business is recognised and the AI can describe it accurately, but it doesn't yet have enough evidence to confidently recommend it in response to a commercial query.

This is one of the most common situations for businesses that have established the foundations of entity recognition but haven't yet built enough third-party coverage.

The solution is the work covered in months three and four of the roadmap: earning industry features, interviews, case studies, and other independent mentions where your business is the primary subject. These sources help AI systems understand not just that your business exists, but why it should be considered a credible option.

AI can't tell you much about your business. This is usually a foundation problem. AI systems don't have enough information to confidently understand who you are, what you do, or whether you're a credible answer to a user's question.

While this can be frustrating, it's often easier to fix than incorrect information as you're not dealing with conflicting signals or a damaged reputation, you're simply dealing with a lack of recognition. This is where the six-month roadmap comes in. Start with the foundations and build from there.

AI's description of you is accurate today but changes next month. This usually means the information AI is finding about your business is still quite thin and it's more common than most people realise. AI systems periodically retrain, recrawl, and re-evaluate their entity data and businesses with thin or inconsistently maintained entity signals can find their recognition fluctuating as a result. 

The solution is consistency, the more places that accurately describe your business, the easier it becomes for AI systems to understand who you are and trust the information they find.

Go back and run that test again in six months. The businesses showing up in AI answers aren't doing anything mysterious, they've just made it easy for AI to understand who they are. Do that consistently, and the difference in what comes back will speak for itself.

Frequently Asked Questions About Entities in AI Search

What is an entity in AI search?

An entity is a uniquely identifiable thing that AI systems can recognise and understand. Entities can include businesses, people, locations, products, organisations, and concepts. Instead of simply matching keywords, AI systems use entities to understand what something is and how it relates to other things.

Why are entities important for AI search?

AI systems can only recommend, compare, or describe businesses they can confidently identify. If your business is not clearly established as an entity, AI may struggle to understand who you are, what you do, or when you should appear in an answer.

What is the difference between a keyword and an entity?

A keyword is a phrase someone types into a search engine. An entity is the actual thing being discussed. For example, "Auckland accountant" is a keyword, while a specific accounting firm in Auckland is an entity. AI search relies heavily on entities because they provide context, relationships, and real-world meaning.

What is an example of an entity?

An entity can be a person, place, organisation, or concept. For example, Jacinda Ardern is a person entity, Wellington is a place entity, Apple is an organisation entity, and machine learning is a concept entity. In AI search, businesses are also entities because they can be uniquely identified and distinguished from other organisations.

Can a small business become a recognised entity?

Absolutely. Entity recognition is not reserved for large brands. Small businesses can strengthen their entity footprint by maintaining consistent business information, implementing schema markup, earning third-party mentions, building reviews, and creating content that clearly explains who they are and what they do.