← Back to blog

How to Actually Check Whether ChatGPT and AI Overviews Are Citing Your Business

13 October 2026·5 min read
Quick answer: There's no reliable analytics tool yet that tells you every time ChatGPT, Gemini or Google's AI Overviews mention your business — so the honest method right now is manual, repeated prompt testing across a set of templates, logged over time. Run the same prompts logged out, vary the phrasing, and track whether you're mentioned, how, and what's cited as the source. It's imperfect and time-consuming, but it beats guessing. ✨

Every client conversation about "AI SEO" eventually gets to the same question: "can you just check if ChatGPT mentions us?" And the honest answer is — sort of, but not with the confidence a Google Search Console report gives you. 💖 The tools claiming to track "AI visibility" right now are mostly scraping sampled queries with opaque methodology, and results can differ between the same prompt run five minutes apart. So until something more reliable exists, a manual audit is genuinely the most defensible method — here's how we run one.

What most businesses get wrong

The biggest mistake is running one query, in one tool, one time, and treating the result as the verdict. AI answers aren't static like a search results page — they're influenced by your account's memory settings, your approximate location, how the question is phrased, and plain randomness in how the model generates a response. The second mistake is assuming "not mentioned" means "invisible to AI," when often it just means your phrasing didn't match what that tool searches for. And the third is chasing a shiny third-party "AI visibility score" without understanding what it's actually sampling — treat those numbers as a rough directional signal at best, not a dashboard you can trust at face value.

The manual AI citation audit — a copy-paste prompt set:

Run each of these across ChatGPT (signed out or with memory off), Google (checking if an AI Overview appears), Gemini, and Perplexity. Log the results in a simple spreadsheet: tool, date, prompt, mentioned (Y/N), position in the answer, and source cited.

  • Direct brand query: "What do you know about [business name]?" — tests baseline recognition and factual accuracy.
  • Category + location query: "What's a good [category] in [suburb/region]?" — tests whether you surface in a genuine discovery scenario.
  • Comparison query: "What are some alternatives to [a known competitor]?" — tests whether you're part of the competitive set AI tools reach for.
  • Problem-based query: "I need help with [the problem your business solves] — who should I use?" — tests whether your content maps to the actual problem, not just your brand name.
  • Follow-up query: Ask "why did you recommend them?" after a mention — this often reveals which page or review source the model is actually drawing from.

Repeat this set monthly, not once. A single snapshot tells you almost nothing about a trend.

A local service business checking visibility in its own suburb: Running the category + location prompt across a few neighbouring suburb variations often reveals the business is cited for its exact suburb but drops out entirely one suburb over — useful intel for where local content and citations still need work.
A B2B firm with a long sales cycle: The comparison query matters most here, since B2B buyers research "alternatives to X" extensively before ever reaching out. If a firm never shows up in that answer, it's a sign their comparison or positioning content simply doesn't exist in a form these tools can read and summarise.
A business that recently rebranded or changed its name: The direct brand query is the most revealing test here — it's common to find an AI tool confidently citing the old business name, old address, or outdated details for months after a change, because it's drawing from older indexed content.

How to run the audit without kidding yourself

Control your variables as much as the tool allows. Use a logged-out or fresh session in ChatGPT where possible, and turn off memory/personalisation if you're signed in, since a model that remembers your previous chats about your own business will obviously be more likely to mention it — that's not a real result. Vary the phrasing across each test round rather than repeating the exact same sentence, since real customers won't phrase it identically either. Run the full set over at least three separate days before drawing any conclusion, because outputs genuinely do shift between sessions. When a citation does appear, click through and check whether the source it's drawing from is accurate and current — it's common for an AI answer to cite a review, directory listing, or old page that no longer reflects reality.

💡 Write content in clear question-and-answer form. FAQ-style pages with direct, plainly worded answers are consistently easier for these tools to extract and cite than dense marketing copy — it's one of the few levers you actually control.

Mistakes to avoid

  • Drawing conclusions from a single prompt in a single tool on a single day.
  • Testing while logged in with memory on and mistaking personalised results for genuine visibility.
  • Paying for a third-party "AI visibility" tool without understanding what it's actually sampling.
  • Ignoring what source the AI is citing — the citation itself matters less than whether it's accurate.
  • Assuming a good result once means you can stop checking — these answers shift over time.

Frequently asked questions

Is there a reliable tool that tracks AI citations automatically?

Not yet, honestly. Several tools market themselves this way, but none of them have transparent, verifiable methodology comparable to how Google Search Console reports actual search impressions — treat any such tool's numbers as a rough directional indicator, not ground truth.

How often should I re-run the audit?

Monthly is a reasonable cadence for most small businesses — frequent enough to catch meaningful change, not so frequent that you're chasing noise in results that shift session to session anyway.

Does being cited by AI tools actually drive traffic or leads?

Sometimes, but it's genuinely hard to attribute cleanly, since these tools rarely pass clear referral data the way a search click does. Treat AI citation as one signal among several, not a metric to optimise in isolation.

Can I influence what these tools say about my business?

Indirectly — clear, accurate, well-structured content and up-to-date listings help, but there's no guaranteed mechanism to force a mention the way there is with paid search. Be wary of anyone promising guaranteed AI citations.


Keep reading 🤍

Share
Written by
Kate, founder of Chronically Online

I help Gold Coast and Brisbane businesses grow with branding, websites and marketing that actually works.

Work with me ✦