Using AI for Search: How to Check If It Recommends You

More customers now use AI for search when they look for local services, software and suppliers. To check whether AI assistants recommend your business, ask ChatGPT, Gemini, Perplexity and Google the questions your customers ask. Then record which businesses get named and which websites get cited, and repeat the check on a schedule. A first pass with a short question list takes about 20 minutes and a spreadsheet.

Here's the thing: AI answers vary between engines, between accounts, and even between two runs of the same question. A single screenshot proves very little. This guide gives you a six-step self-audit, a copyable tracking template, and the technical checks that decide whether AI crawlers can read your site at all.

The Six Steps at a Glance

1. List the questions your customers actually ask. Write down real buying questions in plain language.
2. Ask ChatGPT, Gemini, Perplexity and Google those questions. Use clean sessions so your personal history does not skew results.
3. Record which businesses get named and which sites get cited. Log every answer in one table.
4. Check whether AI crawlers can read your website. Confirm robots.txt and security settings do not block them.
5. Fill the gaps with clear, answer-first pages. Publish the facts AI tools need before they can recommend you.
6. Re-check on a schedule or automate the tracking. Compare results month over month, not answer by answer.

Step 1: List the Questions Your Customers Actually Ask

Start with buyer questions, not your brand name. Asking an AI tool about your own company only tests whether it knows you exist. Asking what a customer would ask tests whether it recommends you.

Pull questions from sales calls, contact-form messages, reviews and the search terms in your analytics. Start with five to ten for a first pass, then grow the list toward 20. Write them the way people talk to a chatbot: longer and more specific than a typed Google query.

Cover four types of question:

• Category + location: best family dentist in [your city] for nervous patients
• Problem: how do I stop my online store's order emails going to spam
• Comparison: [your service type] vs [alternative approach] for a small business
• Brand check: is [your business name] a good choice for [service]

The first three show whether you get recommended to people who have never heard of you. The fourth shows what AI tools say when someone already has your name.

Step 2: Ask ChatGPT, Gemini, Perplexity and Google Those Questions

Run every question through the four places your customers are most likely to use: ChatGPT, Google Gemini, Perplexity, and Google Search, where an AI Overview or AI Mode answer may appear above the regular results. Each engine draws on different sources. A strong showing in one says little about the others.

Reduce personalization before you start. ChatGPT's memory and chat history can shape answers, so use a logged-out session or Temporary Chat, which OpenAI says does not use or create memories. For Google, use a private browser window. For local questions, put the city in the question instead of relying on your device's location.

Ask each question twice, in separate sessions. If your business appears in one run but not the other, record it as partial. Partial visibility is normal and worth tracking.

It also helps to understand how Google AI Overviews work before you read the Google results. The overview draws on pages already in Google's index, so your regular search presence feeds directly into it.

Step 3: Record Which Businesses Get Named and Which Sites Get Cited

Log every answer in one table so patterns become visible. Separate two outcomes. A mention means the answer names your business. A citation means the answer links to a page as its source; this short explainer on what AI citations are covers the difference in more detail.

Copy this template into a spreadsheet:

The URLs-cited column is often the most revealing. When answers keep citing the same directory, review site, forum thread or industry roundup, that page is shaping the recommendation. Getting accurately listed there can matter as much as changing your own website, because AI citations frequently point to third-party pages rather than business sites.

Note factual errors too: wrong prices, old addresses, services you no longer offer. Each error points to a page you need to fix or publish.

Step 4: Check Whether AI Crawlers Can Read Your Website

An AI engine cannot cite a page its crawler is blocked from reading. Open yourdomain.com/robots.txt in a browser. Robots.txt is the plain-text file that tells crawlers which parts of a site they may visit, standardized as RFC 9309, the Robots Exclusion Protocol. Look for Disallow rules aimed at the crawlers below.

The table summarizes each operator's own documentation. OpenAI's crawler documentation states that sites opted out of OAI-SearchBot will not be shown in ChatGPT search answers. Google's crawler documentation says Google-Extended does not affect a site's inclusion or ranking in Google Search. Perplexity describes PerplexityBot as the crawler that surfaces and links websites in its search results.

Q: Does blocking GPTBot remove my business from ChatGPT search?
A: No. According to OpenAI's crawler documentation, GPTBot governs model training, while OAI-SearchBot governs whether pages appear in ChatGPT search answers, and each is controlled separately in robots.txt.

Robots.txt is not the only possible blocker. Some CDN and security services, Cloudflare among them, offer settings that block AI crawlers at the network level, so ask whoever manages your hosting to confirm what is switched on. If the file needs rewriting, a robots.txt generator produces correctly formatted rules for each crawler.

What about llms.txt?

An llms.txt file is a Markdown file placed at yourdomain.com/llms.txt that summarizes a site and links to its key pages for language models. It is a proposed convention published at llmstxt.org in 2024, not an adopted web standard. The major AI search engines have not publicly confirmed that they use it.

Treat llms.txt as a low-cost optional step, not a ranking lever. A free llms.txt generator produces one in minutes, and the same set of free llms.txt and schema tools covers related markup.

Step 5: Fill the Gaps With Clear, Answer-First Pages

Your tracking table now shows where you are missing. Close those gaps with pages that answer the exact question in the first paragraph, then add detail. AI tools pull short, self-contained answers more easily than claims buried in long marketing copy.

Q: Do I need special code to appear in Google AI Overviews?
A: No. Google Search Central states that there are no additional requirements to appear in AI Overviews or AI Mode beyond being indexed and eligible to show a snippet in Google Search.

In practice, the gaps usually fall into three groups:

• Missing pages. If customers ask about a service or area and you have no page for it, create one.
• Missing facts. State service areas, price ranges, turnaround times and who the service suits in plain sentences.
• Missing third-party presence. Get listed accurately on the directories and review sites the engines cited in Step 3.

Matching intent matters at every level. The closer a page and its search snippet match what people actually search for, the more likely Google shows your version instead of rewriting it. For the wider strategy, see the AI search SEO guide and this breakdown of how SEO and GEO, short for generative engine optimization, work together.

Step 6: Re-Check on a Schedule or Automate the Tracking

AI answers shift as engines update their models and re-crawl the web, so one audit is only a snapshot. Repeat the same question set monthly, and again a few weeks after you publish significant new pages. Compare how often you were named across months rather than reacting to any single answer.

Manual checks work well for a short question list. They get harder as questions, engines and locations multiply.

Once the manual version becomes a chore, dedicated tools for automated AI search visibility tracking run the same buyer questions on a fixed schedule and log mentions and cited URLs over time. Either way, the method stays the same: fixed questions, consistent settings, and trends over single results.

Common Mistakes When Testing Your AI Search Visibility

• Testing once and drawing conclusions. Answers vary between runs. Trends across weeks are what count.
• Only searching your own name. Brand queries confirm awareness, not recommendation.
• Using a personalized account. Chat history and memory can favor businesses you have discussed before.
• Confusing mentions with citations. Being named without a link and being cited as a source are different outcomes with different fixes.
• Blocking the wrong crawler. Blocking OAI-SearchBot to avoid model training removes pages from ChatGPT search; GPTBot is the training crawler.
• Treating llms.txt as a fix. It is optional and unconfirmed. Crawl access and clear pages matter far more.
• Ignoring third-party sources. When engines cite directories and reviews, your own website is only part of the picture.

Related Articles

• What Is AI SEO? — a plain-language introduction to optimizing for AI-generated answers alongside regular search.
• How AI Search Is Rewriting the Rules of SEO in 2026 — how AI answers are changing the way businesses get found online.
• How to Read Google Search Console Like an Analyst — how to mine the search data you already have for the real questions customers ask.

Useful Links

• AI Search Visibility
• GoBorn — booked out by search & cited by AI
• Free llms.txt Generator
• What Is GEO? Generative Engine Optimization

Need help with this? GoBorn · Get started