In our guide to Generative Engine Optimisation (GEO) we made the case that being cited inside AI answers is becoming as important as ranking on Google. This is the hands on companion. Before you spend a cent on GEO, you should know exactly where you stand today, and the good news is you can find out yourself in about half an hour, with nothing but a browser and a coffee.
Here is the short version. Open ChatGPT, Perplexity and Gemini, ask each the questions your customers actually ask, and write down whether your business appears, whether a competitor appears instead, and what each engine says about you. That simple exercise tells you more than most paid dashboards, and it usually surfaces at least one uncomfortable surprise.
Two facts make this audit more important than it first looks, and they shape how you read the results.
Winning one AI engine does not win the others. Analysis of 680 million AI citations found that only about 11 percent of domains are cited by both ChatGPT and Perplexity. An independent study of 118,000 responses reached the same number. The brand each engine “sees” is almost a different internet. So you cannot test ChatGPT, see yourself, and assume you are covered. You have to check each platform separately.
ChatGPT runs largely on Bing, and almost nobody optimises for Bing. ChatGPT Search pulls primarily from Microsoft Bing’s index, a partnership OpenAI confirmed at launch. One study found that 87 percent of ChatGPT citations matched Bing’s top organic results for the same query. If your site is not indexed in Bing, ChatGPT struggles to find you, no matter how well you rank on Google. Most SEO teams have only ever touched Google Search Console and have never opened Bing Webmaster Tools. That gap is both the problem and the opportunity.
Let us run the audit.
Step 1: Build Your Query List
You are not testing your brand name. Searching “tell me about [your business]” only proves the engine has heard of you. The real test is whether you appear when someone describes a problem you solve without naming you.
Write down ten to fifteen questions a genuine prospect would ask. Mix three types:
- Category questions. “What is the best [your product or service] in Brisbane?”
- Problem questions. “How do I fix [the problem your customer has]?”
- Comparison and buying questions. “[Competitor] versus [alternative], which is better for a small business?” These matter most, because comparison and high intent queries trigger live web search far more often than general questions, which is exactly where citations are won or lost.
Keep the list realistic and local where it counts. A Brisbane plumber should ask about Brisbane, not the world. Save the list, because you will run the identical questions through every engine, and you will rerun them in a few months to track progress.
Step 2: Audit ChatGPT
Open ChatGPT and make sure web search is active for each query, since that is what pulls live citations rather than memory. Then work through your list and, for each question, note three things: does your business appear, who appears instead, and how are you described if you do show up.
A detail worth knowing as you go: ChatGPT is selective. It tends to cite only around five sources per response, where Perplexity will cite twenty or more. Fewer slots means the bar to get mentioned is higher, so do not be discouraged if you are thinner here than elsewhere. It also leans heavily on third party sources rather than your own website. A large share of ChatGPT citations come from directories, review platforms and reference sites such as Wikipedia, which is your first clue about where the work needs to happen.
Record each result before moving on. The pattern across the whole list matters more than any single answer, because AI responses vary from run to run.
Step 3: Audit Perplexity
Run the exact same questions through Perplexity. You will almost certainly see a different cast of characters, and that is the point of the 11 percent overlap figure made real in front of you.
Perplexity cites densely and shows its sources inline, which makes it the easiest engine to audit. Watch in particular for community sources. Perplexity draws heavily on Reddit and similar discussion sites, so a brand with an active, authentic presence in the right communities often shows up here while being invisible on ChatGPT. Note where your competitors are getting their Perplexity citations from. That tells you which conversations you are missing.
Step 4: Audit Gemini (and a Note on Claude and Copilot)
Run the list a third time through Google’s Gemini. Gemini grounds its answers in Google’s index rather than Bing’s, so this is where your existing Google SEO is most likely to carry over. If you appear in Gemini but not ChatGPT, that is a strong signal of the Bing gap described above.
If you want a fuller picture, two more are worth a quick pass, though neither is essential for a first audit. Claude is the one to add if any: it retrieves through Brave Search, which makes it a fourth distinct source pool, so a brand can surface there independently of the other three. Microsoft Copilot is Bing based and behaves much like ChatGPT, so it largely confirms what your ChatGPT results already showed. For most businesses the three engines above give you the core picture across the source universes that matter most, Bing led, community led and Google led, with Claude’s Brave led pool as a useful optional fourth.
Your Audit Scorecard
Keep it simple. A table like this, one row per query, is all you need. Mark whether you appear, who the top competitor is, and a quick note on how you are described.
| Query | ChatGPT | Perplexity | Gemini | Top competitor cited | Notes |
|---|---|---|---|---|---|
| Best [service] in [city] | You / No | You / No | You / No | ||
| How do I [solve problem]? | You / No | You / No | You / No | ||
| [Competitor] vs [alternative] | You / No | You / No | You / No |
When you have worked through the list, count your appearances per engine. That gives you three visibility scores out of fifteen, and an instant map of which platform is your strongest and which is your blind spot. If you ran the optional Claude check, add a fourth column and score it the same way.
How to Read Your Results
A few common patterns and what they usually mean.
- You appear on Gemini but not ChatGPT. Classic Bing gap. Your Google SEO is working, but Bing has not indexed you well, so ChatGPT cannot find you. This is one of the most fixable problems on the list.
- You appear on Perplexity but not ChatGPT. You likely have some community or editorial presence Perplexity values, but you are thin on the third party listings and reference sources ChatGPT favours.
- A competitor appears everywhere and you appear nowhere. They have built citation authority across multiple source pools. This is the compounding first mover advantage we wrote about in the GEO guide, and it is harder, though not impossible, to claw back the longer you wait.
- You appear, but the description is wrong or dated. Visibility without accuracy. The engines are pulling stale or incorrect information about you from somewhere, and that source needs correcting.
One more reassurance. Different engines disagree on which brands to recommend more often than not. Benchmarks have found the major engines name different brands around 62 percent of the time. So an inconsistent result across platforms is normal, not a mistake in your method.
The Technical Checks Most People Skip
Content is only half the story. If the engines physically cannot reach or read your pages, no amount of good writing will get you cited. These are the checks most SEO teams have never run.
- Get into Bing Webmaster Tools. Because ChatGPT leans on Bing’s index, Bing indexing is effectively a prerequisite for ChatGPT visibility. You can import your settings straight from Google Search Console in a click, submit your sitemap, and check that your key pages are actually indexed. If they are not in Bing, they are invisible to ChatGPT.
- Check you are not blocking the AI crawlers. This is the most common own goal in AI search. A robots.txt rule added a year or two ago to “stop AI stealing our content” can quietly remove you from AI answers entirely. To be citable you need to allow the search crawlers, in particular OAI-SearchBot for ChatGPT, PerplexityBot for Perplexity, and Bingbot. Note that the search crawler and the training crawler are separate, so you can allow citation while still blocking training if that is your preference.
- Read your server logs. Search the access logs for OAI-SearchBot. If it appears regularly, ChatGPT’s crawler is visiting you. If it never appears, something is blocking it, sometimes at the firewall or CDN level rather than in robots.txt, which is a different fix.
- Make sure your content is in the HTML, not just the JavaScript. Many AI crawlers do not run JavaScript. If your main content only appears after scripts load, the crawler may see an empty shell. Server side rendering, or at least ensuring core content is in the initial HTML, is the single technical requirement shared by every platform.
- Use IndexNow to speed things up. Submitting updated URLs through Bing Webmaster Tools or the IndexNow protocol prompts a faster crawl, so new and changed pages reach ChatGPT sooner.
If that list reads like another language, that is exactly the kind of thing solid website development takes off your plate.
What to Do About the Gaps
Once you can see your three scores and your technical checklist, the priorities usually sort themselves.
- If the technical checks failed, fix those first. There is no point optimising content the crawlers cannot reach.
- If ChatGPT is your weak spot, the work is mostly off your own website: getting indexed in Bing, and building presence on the third party directories, review platforms and reference sites ChatGPT trusts.
- If Perplexity is your weak spot, look at authentic participation in the communities and editorial sources it favours.
- If you are invisible everywhere, start with original, genuinely useful content that answers your top questions directly, then earn the third party mentions that signal credibility to every engine.
None of this requires a big budget. It rewards expertise, accurate information and consistency, which is precisely why a focused small business can outpace a larger but complacent competitor right now.
There is a reason this is a “test it yourself” article. When you run the audit and see your own gaps in black and white, the next step becomes obvious. That is the point where a lot of business owners decide they would rather hand the fix to someone who does this every day. If that is you, we are happy to help.
Frequently Asked Questions (FAQ)
How often should I run this audit?
Quarterly is a sensible rhythm for most businesses, using the same query list each time so the results are comparable. AI engines update their sources and behaviour constantly, so a result from six months ago is not reliable today. Run it more often after you make a significant change, to see whether it landed.
Why do I appear in one AI engine but not another?
Because the engines draw from largely separate source pools. Only about 11 percent of domains are cited by both ChatGPT and Perplexity, and each platform weighs different signals. ChatGPT leans on Bing’s index and third party listings, Perplexity leans on community and editorial sources, and Gemini grounds in Google. Appearing on one is no guarantee of the others.
My SEO is good, so why is ChatGPT ignoring me?
Most likely the Bing gap. ChatGPT Search relies heavily on Bing’s index, and many businesses that rank well on Google have weak Bing coverage because no one has ever set up Bing Webmaster Tools. A second common cause is a robots.txt or firewall rule that blocks the AI search crawlers without anyone realising.
Do I need paid tools to audit my AI visibility?
No. The manual audit in this article costs nothing but time and is genuinely informative. Paid platforms automate tracking across many queries and engines, which is useful at scale, but you should run the manual version first so you understand what the dashboards are measuring.
Is the result reliable if I only ask each question once?
Treat patterns, not single answers, as your signal. AI responses are non deterministic, so the same question can return different sources on different runs. Ask each question once or twice, focus on what shows up consistently across your whole list, and you will get a fair picture.
Should I bother with Bing if almost no one uses it to search?
Yes, but not for Bing’s own traffic. Bing matters because it is the index feeding ChatGPT, and to a large extent Copilot. Getting indexed in Bing is one of the highest leverage, lowest effort moves for AI search visibility, precisely because so few competitors have done it.
Conclusion
You cannot improve what you have not measured, and AI search visibility is the rare thing you can measure yourself this afternoon. Build your query list, run it through ChatGPT, Perplexity and Gemini, score where you appear, and check the handful of technical settings that quietly decide whether the engines can see you at all.
Do that and you will know two things most of your competitors do not: exactly where your brand stands in AI answers today, and which specific gaps to close first. The businesses acting on that now are the ones AI will be recommending by name a year from now. The window is open, and the first move costs nothing but thirty minutes.
Related Reading
- A Guide to GEO (Generative Engine Optimisation): The Sequel to AEO, the strategy this audit puts into practice.
- A Guide to AEO (Answer Engine Optimisation) and How It Differs From SEO, where the answer engine story began.
- AI Revolution: Your Competitive Edge Is Disappearing (What to Do), why moving early on this matters.
- AI in Online Marketing: Unlocking New Possibilities, the wider shift AI is driving across marketing.
- Website Development and Online Marketing services, the technical and strategic help behind closing the gaps.
