Where Does ChatGPT Get Its Information? A Business Owner's Guide

ChatGPT draws on four sources: the static data it was trained on, live web browsing through Bing's search index, third-party plugins it can query mid-conversation, and whatever context you type into the prompt. Ahrefs has mapped this breakdown in detail. For a business, the browsing layer is the one that matters most, because that is what decides whether you get named when someone asks ChatGPT for a recommendation today, not months-old training data.

Where does ChatGPT actually get its information from?

Ahrefs' review of how ChatGPT sources answers breaks the system down into four layers. The first is training data, the large fixed dataset the model learned from before a cutoff date. The second is live browsing, where ChatGPT sends a query out to Bing's index and reads back current web pages. The third is plugins and connectors, which let ChatGPT pull structured data from specific tools or APIs during a conversation. The fourth is the context window, the information you paste or type directly into the chat. Most consumer searches for local recommendations, such as best letting agent in Basingstoke, trigger the second layer, live browsing, rather than relying on frozen training data.

Why won't static training data get a business mentioned by name?

Training data is frozen at a point in time and it is built from broad web content, not a directory of every local business. A small property agency in Hampshire is extremely unlikely to appear as a named entity in that training set with any detail about its services, fees or current availability. Even where the training data has heard of a business, the model has no way to confirm the details are still accurate, so it is generally pushed toward citing something it can verify through live browsing instead. This is why a business cannot improve its ChatGPT visibility by simply existing online for years. It needs to be present and well structured in the sources ChatGPT retrieves from right now.

How does ChatGPT choose which businesses to mention?

When ChatGPT browses live, it pulls from Bing's index and weighs signals that overlap with classic SEO but are not identical to it. It favours pages with clear, specific answers to the question asked, consistent business information repeated across multiple credible sources, and structured data that makes it easy to extract a name, location and service. It also appears to favour sources that are easy to quote directly, meaning plain sentences stating facts rather than vague marketing copy. A business mentioned consistently, with the same name, address and service description, across its own site, directories and review platforms, is easier for the model to trust and cite than one with inconsistent listings.

Why can a Hampshire letting agent rank well on Google but disappear on ChatGPT?

Picture a letting agent in Andover that has spent years on local SEO. They rank on page one of Google for letting agent Andover and pick up steady traffic from organic search. Ask ChatGPT or Perplexity to recommend a letting agent in Andover, Winchester or Basingstoke, though, and that same agency is often missing entirely, replaced by national portals or a competitor with weaker Google rankings but clearer AI-readable content. This gap is common and it catches experienced agency owners off guard, because they assume a strong Google position means strong visibility everywhere.

The reason is that Google's classic ranking algorithm and ChatGPT's retrieval layer value different things. Google rewards years of backlinks and domain authority. ChatGPT's browsing layer rewards content that answers a specific question clearly and can be lifted out and quoted, plus consistent entity signals across the web. In Antek Automation's audits of Hampshire property businesses, a recurring pattern shows up: most have strong Google rankings but return zero mentions when the same buyer intent is tested against ChatGPT, Perplexity and Google AI Overviews. That gap is the whole reason generative engine optimisation exists as a separate discipline from SEO.

What is GEO, generative engine optimisation, and how is it different from SEO?

Generative engine optimisation, or GEO, is the practice of structuring a business's content and listings so that AI systems like ChatGPT, Perplexity, Copilot and Google AI Overviews can find it, extract clear facts from it, and cite it by name in an answer. SEO targets Google's classic search index and ranking signals. GEO targets the browsing and retrieval layer that sits behind AI assistants, which draws on a different mix of sources including Bing's index, structured data, review platforms and directories. The two overlap but are not the same job, and a business can rank well in one and be invisible in the other.

Antek Automation runs our GEO service for property agents, trades businesses and professional services, treating AI visibility as a separate workstream from traditional SEO rather than an afterthought.

How do you get cited by ChatGPT and AI search engines?

Getting cited starts with making sure the facts about a business, such as name, location, services and pricing structure, are stated plainly and consistently everywhere they appear online. Antek Automation's GEO audit process checks this in four steps: a crawl and indexing check to confirm AI systems can actually access and read the site, a citation and consistency check across directories and review platforms, a content gap analysis against the questions real buyers ask AI assistants, and a structured data review to confirm schema markup is in place and accurate. Running through those four steps on a letting agency's website usually surfaces the same handful of gaps: missing schema, inconsistent business details across listings, and a total absence of content written to directly answer buyer questions.

What can a property business do this month to improve AI search visibility?

Improving AI visibility does not require a rebuild of the website. A short, focused set of changes made this month will move the needle for most property businesses.

• Add schema markup for the business, its services and its locations so AI crawlers can extract clean facts instead of guessing.
• Make sure the business name, address and phone number are identical across the website, Google Business Profile, property portals and local directories.
• Publish short, direct answers to the exact questions buyers ask, such as what does a letting agent in Hampshire charge for a fully managed service, rather than generic marketing pages.
• Collect and display recent reviews, since review platforms are a common source ChatGPT and Perplexity draw on when naming businesses.
• Put an AI receptionist for property management in place so enquiries that do land, whether from Google, ChatGPT or a phone call out of hours, actually get captured instead of going to voicemail.

These changes help the many businesses across Hampshire competing for the same local searches close the gap between their Google ranking and their AI visibility.

Antek Automation offers a free AI Visibility Check that shows whether a business currently shows up when ChatGPT, Perplexity or Google AI Overviews answer questions in its trade and area. Running a GEO audit alongside it maps the specific gaps, in order of impact, so the fixes can be prioritised rather than guessed at.

Frequently asked questions

Where does ChatGPT get its information about local businesses?

ChatGPT pulls local business information mainly from its live browsing layer, which queries Bing's index and reads current web pages, directories and review platforms, rather than relying on its original training data. This means a business's current website content, directory listings and reviews matter more for AI visibility than anything baked into the model months or years ago.

How does ChatGPT choose which businesses to mention in an answer?

It favours businesses with consistent, verifiable information across multiple credible sources and content that states facts plainly enough to be lifted out and quoted. Structured data, accurate directory listings and clear service descriptions all make a business easier for the model to retrieve and cite with confidence.

What is the difference between SEO and GEO?

SEO optimises a website to rank in Google's classic search index, while GEO, generative engine optimisation, optimises a business's content and listings so AI assistants such as ChatGPT and Perplexity can find, verify and cite it directly in an answer. A business can perform well in one and barely register in the other, which is why they need separate strategies.