How ChatGPT, Gemini and Perplexity decide which businesses to name

Do AI assistants search the web or answer from memory?

They can do both. OpenAI’s and Google’s developer docs say the model chooses which: OpenAI’s model “can choose to search the web or not based on the content of the input prompt” (OpenAI, Web search), and Gemini “analyzes the prompt and determines if a Google Search can improve the answer” (Gemini API, Grounding with Google Search, 23 Sep 2026).

Without a search, the answer comes from training, and training stops on a fixed date. OpenAI publishes that date for each model: for the most capable one in its developer docs, the knowledge cutoff is 30 April 2026 (OpenAI, Models). A branch you opened, a service you added or a phone number you changed after that date isn’t in what the model learned. Google says grounding answers in Google Search lets Gemini cite sources “beyond its knowledge cutoff” (Gemini API, Grounding with Google Search).

“Which clinic in Dubai treats sports injuries?” is a question about who to hire today, the kind a search improves. We plan around the search path, because it’s the one your work can change.

Which sources does each assistant read?

Each company documents its own crawler: a program that visits web pages and copies what they say. Being in Google’s index, with Google-Extended left allowed, covers Gemini, and ChatGPT and Perplexity each need their own crawler let in.

• ChatGPT finds pages with OAI-SearchBot. OpenAI says sites that block it “will not be shown in ChatGPT search answers” (OpenAI, Overview of OpenAI crawlers).
• Gemini reads from Google’s Search index. Google describes Gemini’s grounding as “providing content from the Google Search index to the model at prompt time” (Google for Developers, Google’s common crawlers, 14 Jul 2026). When Gemini gives sources, a Sources button lists them, and some answers have none (Gemini Apps Help, View related sources from Gemini Apps).
• Perplexity uses PerplexityBot to “surface and link websites in search results on Perplexity.” When someone asks a question, a second agent, Perplexity-User, may visit a page “and include a link to the page in its response” (Perplexity, Perplexity crawlers).

A September 2025 study compared AI search engines with Google across several industries and languages. It found the AI engines show “a systematic and overwhelming bias towards Earned media (third-party, authoritative sources)” over a brand’s own pages and social posts, and that they differ a lot from each other in which sites they draw on (Chen et al., Generative Engine Optimization: How to Dominate AI Search, 10 Sep 2025).

How do Google’s AI Overviews choose which links to show?

They start from Google’s ordinary rankings. Google says its AI features rely on its “core Search ranking systems to retrieve relevant, up-to-date web pages,” then link to the pages that support the answer. They can also run a query fan-out: several related searches at once, across the parts of the question (Google Search Central, Optimizing your website for generative AI features, 10 Jul 2026).

To be one of those links, a page must be indexed and eligible to show in Google Search with a snippet. Google says there are no extra requirements and no special optimizations (Google Search Central, AI features and your website, 10 Dec 2025).

For local questions, Google adds business listings. It says a Google Business Profile can help a business be visible “in both AI responses and other Google Search results” (Google Search Central, Optimizing your website for generative AI features). Its local results weigh relevance, distance and prominence, and prominence draws on how many websites link to you and how many reviews you have (Google Business Profile Help, Tips to improve your local ranking on Google).

What does this mean for your business?

Our reading is that an assistant recommends from a shortlist that search builds for it. A strong reputation among your customers reaches the answer only once it’s written down on pages the search can find. You can drop off that shortlist in three places.

• The crawler can’t get in. If your robots.txt file or your firewall blocks OAI-SearchBot, ChatGPT search won’t show your pages in its answers. Perplexity asks sites to allow PerplexityBot so they appear in its results (Perplexity, Perplexity crawlers).
• Your pages don’t come up for the question. AI Overviews start from Google’s rankings, and the other assistants run searches of their own, so ordinary search work still decides whether your page is in the pool.
• Nobody else mentions you. The assistants lean toward third-party sources, so a business that only describes itself gives them little to cite.

You also won’t see one stable answer. In a study published in January 2026, SparkToro had 600 volunteers run 12 questions through ChatGPT, Claude and Google’s AI, 2,961 times in all, and found less than a 1 in 100 chance that ChatGPT or Google’s AI would give the same list of brands twice in 100 tries (SparkToro, 27 Jan 2026). Judge by how often you’re named across many runs, never by one answer.

What should you do first?

1. Ask ChatGPT, Gemini and Perplexity the question your best customer would ask, and open the sources each answer links to. Our guide has the full two-minute test.

2. Open yourdomain.com/robots.txt and check that it doesn’t block OAI-SearchBot, PerplexityBot or Googlebot. If you want Gemini to use your pages, leave Google-Extended allowed as well. That one setting covers both training future Gemini models and grounding answers in the Gemini app (Google for Developers, Google’s common crawlers).

3. Make your services page say, in its text, what you do, for whom and where. Then complete every field of your Google Business Profile.

4. Get mentioned on the sites the answers already cite: the directories, review sites and articles you found in step one. The guide covers each fix in order.