How do AI assistants like ChatGPT choose which local businesses to recommend?

The AI companies’ public documentation gives no formula for which local business an assistant recommends. What the companies do publish is how their assistants search, which crawlers they use and what blocking those crawlers does. This page separates what is documented by the companies from what outside researchers have only observed.

What is documented, and what is only observed?

Documented means the company says it in its own help pages or developer docs. Observed means a third party measured it. Observations are useful, but they are snapshots and can change without notice.

What does OpenAI document about ChatGPT? (Documented)

OpenAI’s help page on searching the web with ChatGPT says ChatGPT “can also use location information to find local results.” It may use an approximate location from the person’s IP address. Sharing device location is optional and off by default.

The same page says ChatGPT search “sometimes partners with other search providers” and rewrites the question into targeted queries. Its example: “good restaurants near me” from San Francisco may become “top restaurants San Francisco.” It may also share location with “trusted third-party providers that help provide more local information, such as nearby business listings.”

On crawling, OpenAI’s crawler overview says OAI-SearchBot “is used to surface websites in search results in ChatGPT’s search features.” Sites that opt out “will not be shown in ChatGPT search answers, though can still appear as navigational links.” GPTBot, the training crawler, is a separate setting.

What does Google document about AI Overviews, AI Mode and Gemini? (Documented)

Google’s AI features guide says the same SEO best practices apply, with no extra requirements. AI Overviews and AI Mode may use “query fan-out,” issuing multiple related searches across subtopics and data sources. Google also lists “checking that your Merchant Center and Business Profile information is up-to-date.”

For local results on Maps and Search, Google’s local ranking page says results “are mainly based on relevance, distance, and popularity,” and describes prominence as “how well-known a business is.” It adds: “There’s no way to request or pay for a better local ranking on Google.” Google does not say these same factors decide AI Overviews.

Google compiles business information from crawled web content such as the business’s own website, licensed data from third parties, user contributions such as reviews, and its own interactions with the business.

For Gemini Apps, Google describes grounding as “providing content from the Google Search index to the model at prompt time”.

What does Microsoft document about Copilot? (Documented)

The Bing Webmaster Guidelines cover “Bing search experiences, Copilot, and grounding API results.” They say pages focused on a single topic are more likely to be selected for grounding, and ask for “clear and consistent naming for people, organizations, products, and locations.”

For local businesses specifically, Microsoft’s AI Performance announcement says “accurate business information is especially important” for location-based queries. It recommends Bing Places for Business to keep address, hours and contact information current.

What do Perplexity and Anthropic document? (Documented)

Perplexity says PerplexityBot “is designed to surface and link websites in search results on Perplexity” and recommends allowing it in robots.txt. A second agent, Perplexity-User, visits pages when a person asks a question. Perplexity’s help center says each answer includes numbered citations.

Anthropic documents three crawlers. Blocking Claude-SearchBot “may reduce your site’s visibility and accuracy in user search results.” Blocking Claude-User prevents Claude from retrieving your content in response to a user’s question.

What has been observed but not documented? (Observed)

A December 2024 analysis by Vercel and MERJ found that “none of the major AI crawlers currently render JavaScript,” naming OAI-SearchBot, ClaudeBot and PerplexityBot among others. It found Gemini uses Googlebot’s infrastructure, which does render JavaScript. If this still holds, text that appears only after scripts run may be invisible to those crawlers.

In a January 2026 study, SparkToro had 600 volunteers run 12 prompts through ChatGPT, Claude and Google’s AI a combined 2,961 times. It reported “a <1 in 100 chance” that ChatGPT or Google’s AI would return the same list of brands in any two responses.

The author suggests measuring how often a brand appears across many runs rather than its position. The research was done with a partner at Gumshoe.ai, a company in the AI tracking sector.

What is still unknown?

None of the documentation above explains how an assistant weighs reviews, distance or website quality when it names one local business over another. On its local ranking page, Google says it keeps search algorithm details confidential “to make the ranking system as fair as possible for everyone.”

On ranking, OpenAI’s help page says: “ChatGPT ranks search results using multiple factors intended to help users find relevant, reliable information. Placement is not guaranteed.” Treat any claim of an exact formula with caution.

What can a business do with this?

Work from what is documented. Allow the search crawlers, put your services and service area in plain page text, keep your Google Business Profile and Bing Places listing current, and use one name and description everywhere. Then ask the assistants your customers’ questions each month and record who they name.

The AI visibility checklist covers each step. If you would rather have it checked for you, Ultruxy offers a free visibility report.