How AI Search Chooses Which Businesses to Name

Rodgers Intelligence Group builds AI workflow systems for small businesses across the United States, from Denver. Its founder Mike Rodgers writes this guide for owners who want to know why some businesses get named in AI answers and others get skipped.

Someone opens ChatGPT and asks who to call when the water heater fails. The answer names a few companies. When your company is not one of them, the job goes to whatever name the answer supplied and you never hear about the request.

This guide explains what those engines read before they name a business. It also gives you five checks you can run on your own site, without buying anything. Every claim below links to the document it came from, and every source was read on 2026-09-26.

Nothing here promises a ranking or a citation. The engines decide what they repeat. What you control is whether they can read your pages at all and whether those pages say the same things as the rest of your public facts.

What the engines read

Google's AI features. AI Overviews and AI Mode pull links out of the same index as ordinary search results. To be eligible as a supporting link, a page must be indexed and eligible to be shown in Google Search with a snippet. Google states there are no additional requirements (Google Search Central, read 2026-09-26). Google also says the features may use a query fan-out technique, issuing related searches across subtopics while a response is built, which widens the set of pages that can end up linked (Google Search Central, read 2026-09-26).

ChatGPT search. OpenAI reads the web for search with a crawler named OAI-SearchBot. OpenAI's documentation says sites opted out of OAI-SearchBot will not be shown in ChatGPT search answers, though they can still appear as navigational links (OpenAI documentation, read 2026-09-26).

Perplexity. Perplexity reads the web with PerplexityBot, which its documentation describes as the crawler used to surface and link websites in search results on Perplexity, and not to crawl content for its foundation models (Perplexity documentation, read 2026-09-26).

What these products share is a habit worth knowing before you spend any money. They repeat sources they can reach and parse. The checks below cover being reachable, then parseable, then consistent.

Check 1: can the crawlers reach your pages

Open your robots.txt file and look for the user-agent names. OpenAI recommends allowing OAI-SearchBot in robots.txt to help a site appear in ChatGPT search results, and notes a robots.txt change can take roughly 24 hours to show up in search results (OpenAI documentation, read 2026-09-26). Perplexity makes the same recommendation for PerplexityBot and says its systems may take up to 24 hours to reflect a change (Perplexity documentation, read 2026-09-26).

Two details cost businesses traffic and stay invisible until someone checks. First, a web application firewall can block a crawler even when robots.txt allows it. Perplexity documents this and publishes its own IP ranges for allowlists (Perplexity documentation, read 2026-09-26). Second, a user-triggered fetch is a different thing from an automatic crawl. Perplexity-User supports user actions inside Perplexity and generally ignores robots.txt rules because a user asked for the fetch (Perplexity documentation, read 2026-09-26). ChatGPT-User works the same way, and OpenAI states it is not used to decide whether content appears in search (OpenAI documentation, read 2026-09-26).

For Google, robots.txt rules for Googlebot are the control for Search itself, and Google asks that crawling be allowed in robots.txt and by any CDN or hosting infrastructure (Google Search Central, read 2026-09-26).

Check 2: is the page eligible to be quoted

Reachability is only the first gate. Google's stated requirement for AI Overviews and AI Mode is the same one it uses for ordinary results: the page must be indexed and eligible to be shown with a snippet (Google Search Central, read 2026-09-26). A page blocked by a noindex tag or one that never got indexed cannot be a supporting link, whatever else you do to it.

Google points owners at Search Console for finding these problems, and its troubleshooting notes say a crawl can take anywhere from several days to several months after a change (Google Search Central, read 2026-09-26). That last number is the honest frame for the whole field. The engines move on their schedule, not yours.

Check 3: do the machine-readable facts match the visible ones

Structured data is a shared vocabulary your pages carry so machines do not have to guess. Schema.org is the vocabulary maintained by an open community and founded by Google, Microsoft, Yahoo and Yandex, and it ships in encodings that include JSON-LD (schema.org, read 2026-09-26). Google's guidance for AI features is short: make sure your structured data matches the visible text on the page (Google Search Central, read 2026-09-26).

For a business with a location, the useful type is LocalBusiness. Google's documentation says this markup can tell the search engine about business hours, departments and reviews (Google Search Central, read 2026-09-26). The work is not writing the markup. The work is making the hours in your markup equal the hours on your page and the hours on your Google Business Profile.

One reassurance straight from Google: you do not need to create new machine readable files, AI text files or markup to appear in its AI features, and there is no special schema.org data required for them (Google Search Central, read 2026-09-26).

Check 4: do your business facts agree everywhere

Google's guidelines for representing a business say to represent it as it is consistently represented in the real world, across signage, stationery and branding, and to make the address or service area accurate and precise (Google Business Profile Help, read 2026-09-26). The same page says including unnecessary information in your business name is not permitted and could result in suspension of the profile (Google Business Profile Help, read 2026-09-26).

Google's AI features checklist includes checking that your Merchant Center and Business Profile information is up to date (Google Search Central, read 2026-09-26). In practice that is one long pass with your own eyes: the name, the address, the service area, the hours and the prices you publish should read the same on your website, in your structured data and on your profile. An engine that finds three versions of your hours quotes one of them at random.

Check 5: do your reviews read as earned

Reviews matter to what the engines say about you. Google states that local results rest on relevance and on distance and on prominence, and that prominence is based partly on how many websites link to your business and how many reviews you have (Google Business Profile Help, read 2026-09-26). The same page says plainly that there is no way to request or pay for a better local ranking on Google (Google Business Profile Help, read 2026-09-26).

The review rules are written down and short. Reviews must reflect a genuine experience, and offering incentives in exchange for posting reviews or changing or removing a negative review is treated as fake and misleading content and is strictly prohibited (Google Business Profile Help, read 2026-09-26). Google's Maps policy adds that merchants must not discourage or prohibit negative reviews or selectively solicit positive reviews from customers, and should not pressure customers to leave ratings while on the premises (Maps user content policy, read 2026-09-26). Ask everyone the same way and let the rating land where it lands.

llms.txt, stated honestly

You will see llms.txt sold as a way into AI answers. Here is what it actually is. The file comes from a page that describes itself as a proposal to standardise on using an /llms.txt file to provide information to help agents use a website, written by Jeremy Howard and first published in September 2024 (llmstxt.org, read 2026-09-26). It is a proposal. It is not a standard adopted by a standards body.

Adoption is real, though. The same page counts thousands of sites publishing the file and notes that OpenAI and other AI labs publish llms.txt files for their own developer documentation (llmstxt.org, read 2026-09-26). Perplexity's own documentation invites agents to fetch its documentation index at /llms.txt (Perplexity documentation, read 2026-09-26).

So publish one if you want a short, honest statement of what your business does, what it charges and what it does not do. State your facts there. Do not expect a citation for the file itself, and remember Google's line that no new machine readable file is required to appear in its AI features (Google Search Central, read 2026-09-26).

What blocking a crawler actually buys you

Blocking a crawler is a real decision, so know what each token covers. Google-Extended is a robots.txt product token that manages whether content Google crawls may be used for training future generations of Gemini models and for grounding in Gemini Apps. Google states it does not impact a site's inclusion in Google Search and is not used as a ranking signal there (Google Search Central, read 2026-09-26). For Search, the levers are Googlebot rules plus snippet controls like nosnippet or noindex (Google Search Central, read 2026-09-26).

On OpenAI's side, GPTBot is the crawler whose disallow signals that content should not be used for training its foundation models, while OAI-SearchBot governs ChatGPT search. OpenAI documents the two robots.txt tags as independent settings, so you can allow one and disallow the other (OpenAI documentation, read 2026-09-26).

What none of this promises

No document cited above says a fixed set of fixes gets a business named. Google says indexing and serving are not guaranteed even when a page meets the requirements (Google Search Central, read 2026-09-26). Perplexity and OpenAI make no promise about which sources their answers will pick. Anyone who sells you a guaranteed citation is selling something the engines do not offer.

What you can do is remove every reason for an engine to skip you. Then run the same questions again and write down what the answers say. That log is the only honest measure of this work.

Price facts

The numbers on this page, stated one more time so nothing is implied. The AI Visibility Program: 1200 USD per month, quoted in writing before any work starts (rodgersintelligence.com/ai-seo-geo, read 2026-09-26). The One-Workflow Review: 3500 USD fixed fee, with a written readout within ten business days after required access (rodgersintelligence.com, read 2026-09-26). Those are the only prices this article states, and the same values sit in this page's structured data.

Questions owners ask next

Where Rodgers Intelligence Group fits

Rodgers Intelligence Group runs these checks as paid work for small businesses across the United States, from Denver. The AI Visibility Program covers the checks in this guide plus answer pages on your own domain and a monthly log of which prompts name you and which name someone else. The price is stated here and in this page's machine-readable markup: the program is $1,200 a month, quoted in writing before any work starts (rodgersintelligence.com/ai-seo-geo, read 2026-09-26). This page states its prices in words and in machine-readable form, so any engine can quote the same numbers. If you serve one trade or one profession, the industry pages cover the buyer language in that field: law, construction and professional services.

If you want one workflow mapped before you commit to anything else, the One-Workflow Review is a $3,500 fixed fee, with a written readout within ten business days after required access (rodgersintelligence.com, read 2026-09-26). The readout stands on its own.

For the profile side of the same job, read the Google Business Profile checklist or the older AI strategy playbook for local businesses. Questions first, pitch second.