To check if ChatGPT recommends your brand, run the questions your buyers ask (for example "best CRM for startups") and record whether your brand is named, where it appears in the answer, and which pages are cited. Doing this by hand in one chat window is unreliable because answers vary by run, engine and location. An AI visibility tool runs the same prompts across ChatGPT, Claude, Gemini, Perplexity and Google's AI answers and reports mentions and citations for you.
This guide shows how to run a proper check, what to look for in the results, and which tools fit on-demand checks versus ongoing tracking.
Why can't I just ask ChatGPT if it recommends my brand?
You can, but the answer won't tell you much. A few reasons:
- Your chat is personalized. Memory, past conversations and custom instructions can shape what ChatGPT says to you.
- Answers vary. The same prompt can name different brands on different runs, so one check is a sample of one.
- Engines differ. ChatGPT, Gemini, Claude, Perplexity and Google's AI Overviews each search and cite differently. Showing up in one says little about the others.
- Asking about yourself is leading. "Do you recommend Acme?" isn't what buyers type. They ask category questions and see who gets named.
A useful check runs neutral, buyer-style prompts across several engines and records the results in a consistent way.
What should an AI visibility check measure?
For each prompt and engine, capture:
- Mentioned or not: is your brand named in the answer text?
- Position: if named, where does it appear relative to other brands?
- Cited or not: does the engine link to your site as a source?
- Who else is named: which competitors appear instead of you?
- Which sources are cited: the third-party pages (reviews, listicles, Reddit threads, docs) that the engine relied on.
The last one is the most actionable. If an engine keeps citing the same comparison article and you're missing from it, getting included there is a concrete next step.
How do I check AI visibility step by step?
Here's the workflow using Manifold MCP, a hosted MCP server that lets Claude, ChatGPT or Cursor run these checks on demand. It's hosted, needs no API keys of your own, and you pay per call.
1. Connect Manifold MCP to your AI app
Add https://mcp.manifoldmcp.com/mcp as a custom connector in Claude (Settings, then Connectors) or as an MCP app in ChatGPT developer mode, then sign in. New workspaces get 500 free credits with no card.
2. Check that AI crawlers can read your site
Before measuring answers, rule out the basics. Manifold MCP's site readiness check is free. It looks at robots.txt access for known AI crawlers, CDN or firewall rules that turn those crawlers away, whether your content is in the raw HTML without JavaScript, noindex and snippet directives, sitemap, structured data and llms.txt.
Use Manifold MCP to check whether example.com is ready for AI search engines.
Tell me if we block any AI crawlers in robots.txt or at the CDN, whether
our content loads without JavaScript, and list every failing check with
how to fix it.
3. Pick the prompts that matter
Start with 5 to 10 questions a buyer would ask before choosing a product like yours. Mix category prompts ("best X for Y"), comparison prompts ("X vs Y") and problem prompts ("how do I fix Z"). If you're unsure what people ask, Manifold MCP can search an index of prompts observed on ChatGPT and Google's AI Overviews for a topic or domain.
What prompts do people ask ChatGPT about "crm for startups"? Show the
top 30 with which brands get mentioned and which domains get cited.
Mark any prompt where example.com is cited.
4. Run the prompts across engines
Now run your shortlist across engines with your brand and competitors listed for detection. Manifold MCP's AI answer tools run up to 10 prompts per call across ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews and AI Mode, and check up to 10 brands. ChatGPT and Gemini results are read from the answers their consumer apps show a person; Claude and Perplexity come from their model APIs with web search on.
Run these prompts on ChatGPT, Claude, Gemini, Perplexity and Google AI
Overviews in the US: "best crm for startups", "hubspot alternatives for
small teams", "simplest crm for a 5 person sales team".
Detect these brands: example.com, HubSpot, Pipedrive, Attio.
Make a table of who is mentioned, in what position, and who is cited,
per engine.
5. Read the citations and act
Ask for the sources behind the answers, grouped by domain.
From those results, list every cited URL grouped by domain, and flag the
pages that mention our competitors but not us. For the top 5, suggest how
we could earn a mention: an update request, a guest post, a review, or a
better page of our own.
Common fixes include unblocking AI crawlers, publishing clear comparison and pricing pages, answering the exact question in the first paragraph of a page, and earning mentions on the third-party pages engines already cite.
Should I use an on-demand check or a scheduled tracker?
It depends on how often you need the answer.
On-demand checks suit audits, launches, pitches and "where do we stand right now?" questions. With Manifold MCP you pay per prompt and engine instead of a monthly fee for a fixed set of tracked prompts. Its pricing page lists one prompt on ChatGPT, Claude, Gemini and Perplexity at 16 credits, and the readiness check is free. AI answers are never cached, so every run reflects what the engines say now.
Scheduled trackers suit teams reporting AI visibility over time. Because answers vary day to day, daily runs give you trend lines rather than snapshots. Two well-known options:
- Peec AI runs your prompts across AI platforms like ChatGPT, Gemini and Copilot daily, according to its docs, and reports visibility, position and sentiment against competitors. It also offers a Looker Studio connector, an API and its own MCP.
- OtterlyAI monitors your prompts daily across ChatGPT, Google AI Overviews, Google AI Mode, Perplexity, Gemini, Claude and Microsoft Copilot, per its help center, and reports brand mentions, website citations and competitor comparisons. It also includes a GEO audit of your site.
Many teams use both approaches: a tracker for the weekly report, and on-demand checks in Claude or ChatGPT when they need to dig into a specific prompt, competitor or engine.
Questions people ask
Does ChatGPT mention my brand?
The only reliable way to know is to test it with the prompts your buyers use, outside your personal account, and repeat across runs. An AI visibility tool does this systematically and shows whether you're mentioned, where, and which sources are cited.
What are AI visibility tools?
AI visibility tools run prompts on AI engines like ChatGPT, Gemini, Claude and Perplexity and report which brands are mentioned and which pages are cited. Some, like Peec AI and OtterlyAI, track a set of prompts on a schedule. Others, like Manifold MCP, run checks on demand from inside your AI assistant.
How do I get ChatGPT to recommend my brand?
Make sure AI crawlers can read your site, publish pages that answer buyer questions directly, and earn mentions on the third-party pages engines already cite, such as comparison articles, reviews and community threads. Then re-check the same prompts to see what changed.
Why do AI answers change every time I check?
AI engines generate each answer fresh, may search the web differently each time, and some vary by location. That's why one manual check isn't enough, and why trackers run prompts repeatedly.
Is AI visibility tracking different from SEO rank tracking?
Yes. Rank tracking records your position in a list of links. AI visibility tracking records whether an engine names you in its answer, how prominently, and whether it cites your pages. Strong SEO helps, since many engines search the web, but the two don't always match.