On October 31, 2024, OpenAI announced ChatGPT Search. The feature lets ChatGPT pull live information from the web during a conversation instead of relying only on its training data. It rolled out to Plus and Team users that day, expanded to logged-in free users in December 2024, and went fully public in February 2025.

For local businesses, this changed the AEO playbook. ChatGPT can now query directories, review sites, and your own pages in real time and cite them by URL in the answer it returns. The optimization signals are different from training-only ChatGPT, and they are different from Google AI Overviews, Claude, and Perplexity. Here is what actually moves the needle.

What ChatGPT Search is, and how it differs from training-only ChatGPT

Classic ChatGPT answers from a frozen training set. If your business opened last month, classic ChatGPT does not know you exist. ChatGPT Search runs an active web crawl, fetches recent pages, and assembles an answer with inline citations back to the source URLs.

The two modes share a chat window but use different retrieval paths. With 800 million weekly active users as of October 2025, ChatGPT is now the second-largest search surface in the US after Google. A meaningful share of those queries trigger the search path, which means real-time retrievability has become a ranking factor in its own right.

How ChatGPT Search retrieves content

OpenAI runs a dedicated crawler for search called OAI-SearchBot. It is separate from GPTBot (which collects training data) and ChatGPT-User (which fetches a page when a user pastes a URL into a chat). Each can be controlled independently in robots.txt.

Blocking OAI-SearchBot removes your site from the index ChatGPT Search reads from. Most businesses block it by accident because their robots.txt denies all bots that are not Googlebot or Bingbot. That one line of config is enough to make you invisible in ChatGPT Search even if everything else on your site is perfect.

From what we have observed in our scans, ChatGPT Search heavily favors content that is recent, structurally clean, and cross-referenced by at least one third-party source. Pages that read like marketing copy get skipped. Pages that directly answer a question, with a clear heading and a concise paragraph, get cited.

The five highest-leverage moves for ChatGPT Search

  1. Allow OAI-SearchBot in robots.txt. Audit the file at yourdomain.com/robots.txt. If you have a blanket User-agent: * with a restrictive Disallow, add an explicit allow rule for OAI-SearchBot, ChatGPT-User, and GPTBot. Pair it with allowlist entries for PerplexityBot, ClaudeBot, and Google-Extended while you are in the file.
  2. Publish an llms.txt. A plain text summary of your business at yourdomain.com/llms.txt gives AI crawlers a clean, authoritative description in your own words. It takes ten minutes to write and removes ambiguity about who you are and what you do.
  3. Add JSON-LD schema for LocalBusiness, FAQPage, and Service. ChatGPT Search reads structured data at the speed of a millisecond parse. Unstructured pages require inference, which the model often skips on. Schema is the fastest way to make your facts machine-legible.
  4. Keep review sentiment fresh. ChatGPT Search reads recent reviews when forming an opinion. A practice with twelve reviews from this quarter outperforms one with two hundred reviews from three years ago. Ask recent clients to mention the specific service they received.
  5. Write answer-format pages for the questions clients ask. Headings phrased as questions, followed by a direct two to three sentence answer, get cited far more often than long marketing pages. Cost questions, comparison questions, and process questions are the highest-yield formats.

How ChatGPT Search differs from the other AI engines

Each AI engine has a different retrieval architecture, and the blind spots are not the same.

  • Google AI Overviews uses Gemini layered on top of the regular Google index. If you rank well in classic Google search, you have a head start here. AI Overviews rewards the same signals as traditional SEO plus strong schema.
  • ChatGPT Search uses Bing as one input but blends it with its own crawl from OAI-SearchBot and partner data feeds. A site that ranks poorly on Google can still appear in ChatGPT Search if the structured data and review signals are strong.
  • Perplexity is the most aggressive at fetching live pages mid-answer and tends to cite four to six sources per response. It rewards depth and freshness more than any other engine. A single timestamped, comprehensive page often beats ten thin pages.
  • Claude with web search enabled is the most conservative. It cites fewer sources and prefers authoritative domains. Industry directories, association sites, and well-known publications carry disproportionate weight.

Optimizing for one does not automatically optimize for the others. The foundation is the same (schema, llms.txt, clean robots.txt, fresh reviews, answer-format content) but the weighting differs. A business that ranks well in ChatGPT Search but not in Claude usually has a directory authority gap. A business that ranks in Claude but not in Perplexity usually has a freshness gap.

How to test if you appear in ChatGPT Search today

Open ChatGPT, make sure search is enabled (the small globe icon in the prompt box), and ask the same kind of question a prospective client would ask. Examples:

  • “Best [your service] in [your city]”
  • “Who does [specific procedure] in [your neighborhood]”
  • “Top rated [your category] near [landmark]”

Click the source icons under the response. ChatGPT Search shows exactly which URLs it pulled from. If your domain is not in that source list, you have a retrieval problem to solve. If your domain is there but the summary is wrong, you have a schema and content problem.

Run the same three queries every two weeks. ChatGPT Search re-crawls and re-ranks often enough that you can measure changes in days, not months. That feedback loop is what makes AEO work faster than traditional SEO.

The compounding effect

Every quarter, more of the people searching for your services use an AI engine instead of Google. BrightLocal's 2026 Local Consumer Review Survey puts the figure at 45% of consumers using AI for local business recommendations, up from 6% the year before. ChatGPT Search is the single largest channel inside that shift.

The businesses that fix their crawler access, add structured data, and publish answer-format content this quarter will be the ones AI cites for the next two years. The signals compound. Late entrants will be optimizing against incumbents who have already been indexed and cited dozens of times.

To see how ChatGPT Search currently describes your business and where the gaps are, run a free AEO Grader scan. We query ChatGPT, Claude, Gemini, and Perplexity with 60 prompts about your brand and return the prioritized fix list in about sixty seconds.

Related reading: the broader BrightLocal 2026 data deep-dive for the consumer-side numbers driving the AI shift, and which schemas AI actually reads for the JSON-LD implementation specifics referenced above.