Get your business into ChatGPT and AI search: a 1-hour audit

To get your business into ChatGPT and other AI search tools, make sure their search crawlers are allowed to read your site, check that Google can index your pages with a snippet, then measure where you already appear before you change anything. Each assistant uses its own crawler, and blocking one by mistake in robots.txt takes you out of that tool’s answers.

This playbook is the one-hour audit: four checks you can run this week with free tools, the prompts to test what each assistant says about you, and where the measurement still falls short (as of October 2026).

How do ChatGPT and other AI tools decide what to show?

When an assistant searches the web, it pulls a handful of pages, writes an answer from them and links its sources. This is a form of retrieval-augmented generation. Your job is to make sure your pages can be fetched, and that they answer the questions your customers ask in plain words.

Each tool has its own crawler for this:

Assistant	Search crawler (allow it)	Training crawler (your choice)	Fetches for a user’s question
ChatGPT	OAI-SearchBot	GPTBot	ChatGPT-User
Claude	Claude-SearchBot	ClaudeBot	Claude-User
Perplexity	PerplexityBot	none listed	Perplexity-User
Google AI Overviews and AI Mode	Googlebot	Google-Extended (other Google AI systems)	—

OpenAI says sites that block OAI-SearchBot “will not be shown in ChatGPT search answers”, and that the search and training crawlers are set independently. Anthropic says blocking Claude-SearchBot may reduce your visibility in Claude’s search results. For Google there’s no separate AI crawler to allow: if Googlebot can crawl and index a page, it can appear in AI Overviews.

Step 1: Check your robots.txt (10 minutes)

Open yourdomain.com/robots.txt in a browser. You’re looking for any Disallow: / under the search crawlers in the table above, or under User-agent: *.

If you, your web developer or a security plugin ever switched on a “block AI bots” setting, check what it actually blocked. A blanket rule can catch the search crawlers along with the training ones. If you only want to opt out of training, a file like this does it:

# Allow AI search crawlers, so pages can be cited in answers
User-agent: OAI-SearchBot
Allow: /

User-agent: Claude-SearchBot
Allow: /

User-agent: PerplexityBot
Allow: /

# Opt out of model training (optional)
User-agent: GPTBot
Disallow: /

User-agent: ClaudeBot
Disallow: /

User-agent: Google-Extended
Disallow: /

Leave Googlebot alone unless you know why it’s blocked. OpenAI and Perplexity both say a robots.txt change can take about 24 hours to take effect.

If you’re unsure what your file means, paste it into ChatGPT, Claude or Gemini with this prompt:

Here is my website's robots.txt file. For each of these user agents,
tell me whether it can crawl my homepage and my /services/ pages:
OAI-SearchBot, GPTBot, Claude-SearchBot, ClaudeBot, PerplexityBot,
Googlebot, Google-Extended.
Show your answer as a table. Quote the exact line that decides each one.
If a rule is ambiguous, say so instead of guessing.

[paste robots.txt here]

Check the quoted lines yourself. The model can misread wildcard rules.

Step 2: Check Google can index your key pages (10 minutes)

Google says there are no extra requirements and no special optimisation needed for AI Overviews or AI Mode. A page has to be indexed and eligible to show with a snippet in normal search. You don’t need an llms.txt file or new markup for Google.

In Google Search Console, paste your five most important URLs (homepage, main service or product pages, pricing, contact) into the URL Inspection bar at the top. Each should say “URL is on Google”. While you’re there, check that nobody has added nosnippet or noindex to those pages. Either one keeps a page out of AI answers as well as normal results.

Step 3: Turn on the free AI reports (15 minutes)

Three free reports now show part of your AI search traffic. None of them shows all of it, so use all three.

Report	Where	What it shows	What it doesn’t
Generative AI performance	Google Search Console	Impressions in AI Overviews, AI Mode and Discover AI features, by page, country, device and date	Clicks, click-through rate or the queries
AI Assistant channel	Google Analytics 4, Traffic acquisition	Visits from tools such as ChatGPT, Gemini, Deepseek, Copilot and Grok	AI Overviews and AI Mode, which GA4 counts as Organic Search
AI Performance (public preview)	Bing Webmaster Tools	Citations in Microsoft Copilot, Bing’s AI summaries and some partners: total citations, cited pages and the “grounding queries” used to find you	ChatGPT, Claude or Perplexity

Google rolled the Search Console report out to all websites on 31 August 2026. The GA4 channel switches on by itself, with no set-up. Bing launched its report as a public preview in February 2026, so expect it to change.

Write this month’s numbers in a spreadsheet as your baseline. Small sites may see single figures for months, which is still worth knowing.

Step 4: Ask the assistants about your business (20 minutes)

The reports show traffic. They don’t show what the assistants say about you when nobody clicks. For that, ask them.

Use ChatGPT, Gemini, Claude and Perplexity with web search on. Start a fresh chat for each prompt, and log out or use a temporary chat so your history doesn’t steer the answer. Swap in your own details:

I'm looking for a [service, e.g. "bookkeeper for a small café"]
in [town or area]. Who would you recommend, and why?
List your sources.

What does [your business name] in [town] do, what does it charge,
and what do customers say about it? List your sources.

Compare [your business name] with [main competitor] for
[the job customers hire you for]. List your sources.

Paste each answer into your spreadsheet with the date, the tool, whether you were mentioned, which of your pages were cited, and anything wrong (old prices, a service you dropped, the wrong area).

Then fix what the answers got wrong at the source. If an assistant quotes old prices, the old prices are on a page somewhere: your site, a directory listing, an old blog post. If it can’t say what you do, your service pages probably don’t say it plainly enough. Put the direct answer in the first two sentences of the page, the same way you’d answer a customer on the phone.

Where does this break?

• Answers change from run to run. Ask the same question twice and you can get two different lists. Treat your log as a rough guide, and look for patterns across tools and months rather than reacting to one answer.
• A lot of AI traffic arrives without a label. GA4 can only sort a visit into the AI Assistant channel when the browser passes on where it came from. When it doesn’t, the visit lands in Direct. Google’s help page also names example assistants rather than a full list.
• Search Console shows impressions only. You’ll see that a page appeared in an AI Overview, not which question triggered it or whether anyone clicked.
• robots.txt doesn’t control everything. OpenAI says robots.txt rules may not apply to ChatGPT-User, because a person asked for the page. Perplexity says Perplexity-User generally ignores them. Blocking those agents won’t reliably keep your pages out of an answer.
• Nobody can promise placement. Agencies selling “guaranteed ChatGPT rankings” are selling something the platforms don’t offer.

How long does this take?

Our estimate is about an hour the first time: 10 minutes on robots.txt, 10 on URL Inspection, 15 on the reports and 20 on the prompts. After that, repeat Steps 3 and 4 once a month. Fifteen to twenty minutes is enough to update the numbers and rerun the prompts.

What to do next

Most fixes from Step 4 are rewrites of your own pages, so make sure the AI drafts you use for them sound like you. The brand voice guide playbook sets that up in an hour. If you’re also deciding what your own team can paste into these tools, read how to stop ChatGPT, Claude and Gemini training on your business data. The glossary has a short entry on SEO AI if you’re weighing up tools for the rewrites.