You already suspect something is off. Maybe demo requests have slowed down even though your website traffic looks steady. Maybe a competitor keeps coming up in conversations with prospects who say "I saw this mentioned when I asked ChatGPT." Whatever brought you here, the next step is the same. Before you can fix your AI visibility, you need to know exactly where it stands right now.
An AI search visibility audit does that. It tells you whether ChatGPT, Perplexity, Gemini, and Claude know your product exists, what they say about it when asked, and where competitors are winning ground you did not even know was being contested. This guide walks through how to run that audit yourself, step by step.
Why this audit matters more than it used to
The reason this audit has become necessary is simple. Buyers are not waiting until they visit your website to form an opinion about your product anymore. Forrester's 2026 B2B Buyer Journey report found that 72% of B2B software buyers use ChatGPT to evaluate vendors at some point in their decision. If you have not checked what those buyers are seeing when they ask, you are marketing blind to a large part of your funnel.
The stakes are also uneven across the industry. Crackle PR's Q2 2026 AI Citation Benchmark found that 51% of B2B tech brands have zero citations across ChatGPT, Perplexity, and Gemini. Half the industry is invisible to AI-driven research. An audit tells you which half you are in, and if it is the wrong one, exactly why.
Step one: write down the questions your buyers are actually asking
Start before you open any AI tool. Make a list of the real questions a buyer at each stage of their research might ask. Early stage questions are broad and category based, like "what are the best project management tools for a 50 person marketing team." Later stage questions get specific, comparing your product by name against one or two named competitors, or asking about a particular feature, integration, or price range.
Aim for at least 15 to 20 questions across both stages. This list becomes the backbone of your entire audit, so pull directly from real sales calls, support tickets, and G2 reviews if you have access to them. Generic questions produce generic results. Specific questions, written the way your actual buyers talk, produce an audit that reflects reality.
Step two: run each question through every major AI platform
Take your question list and run it through ChatGPT, Gemini, Claude, and Perplexity individually. Do not just check one platform and assume the results generalize. G2's report on ChatGPT usage found that ChatGPT remains the dominant AI chatbot for B2B software research at 63%, but the competitive landscape between platforms is shifting fast enough that a product missing from Perplexity's answers could still be showing up well in ChatGPT's, or the reverse.
For each question, log three things. Did your product appear in the answer at all. If it did, was the description accurate and positive, or outdated and vague. And which competitors appeared alongside you, or instead of you.
Keep this log in a simple spreadsheet with one row per question and one column per platform. It feels tedious at first, but the pattern that emerges is usually the most useful part of the entire audit. Some SaaS products show up consistently on Perplexity but rarely on Claude. Others get mentioned constantly for broad category questions but drop out completely once a buyer asks a more specific comparison question. You will not spot these patterns without the log, because any single answer can look fine in isolation while the overall trend tells a different story.
It also helps to run each question more than once, since AI tools do not always return the exact same answer twice, even with identical wording. A product that shows up in three out of five attempts for the same question is in a meaningfully different position than one that shows up in zero out of five, even though a single spot-check might have caught either result and assumed it was the norm.
Step three: check whether AI is citing your actual content
When an AI answer does mention your product, dig one level deeper and check where that information came from. Some AI tools show source links directly in the answer. Others require asking a follow up question like "where did you get that information" to reveal the sources being used.
This step matters because being mentioned and being cited accurately are two different outcomes. Ahrefs' analysis of ChatGPT's citation behavior found that 65.3% of ChatGPT's top cited pages come from domains with a Domain Rating of 80 or higher, meaning authority and trust signals still matter for which sources AI models lean on. If AI is describing your product using outdated pricing or a discontinued feature, that is a signal your own content has not kept pace with what AI models are pulling from.
Step four: benchmark against your top two or three competitors
Run the exact same question list against your closest competitors' names instead of yours. This step often reveals the real story behind your numbers. A product that appears in 40% of relevant AI answers might feel disappointing in isolation, but if your top competitor only appears in 25%, you are actually ahead of where you assumed.
Pay attention to the specific language AI models use to describe each competitor. If a competitor consistently gets described with a specific differentiator, like "best for enterprise teams" or "most affordable option for startups," that phrase is likely pulled directly from their own content. It tells you which positioning angles are already winning in AI-generated answers, and where there might be room to claim a different angle for yourself.
Step five: look for the content gap behind the invisibility
By this point you should have a clear picture of where your product shows up, where it does not, and where competitors are winning. The next step is connecting those gaps back to your actual content. Moz's 2026 analysis of 40,000 queries found that 88% of Google AI Mode citations do not appear in the organic top 10 search results at all, which means a page that ranks well on Google is not automatically a page AI models will trust enough to cite.
Look specifically for content that clearly states what your product does, who it is built for, and how it compares to named alternatives, using plain language rather than marketing phrasing. If that kind of content does not exist on your site, or exists but is buried three pages deep, you now know exactly what to build first.
Step six: turn the audit into a repeatable process, not a one-time check
A single audit gives you a snapshot. AI models update their training and retrieval behavior constantly, and competitors are actively working to improve their own visibility, which means today's results will not hold for long. Usage of AI tools in B2B research grew from 89% in 2025 to 94% in 2026 according to Forrester's Buyers' Journey Survey, and that kind of fast movement means visibility can shift within a single quarter.
Set a recurring schedule, monthly at minimum, to rerun your question list and track whether your position is improving, holding steady, or slipping. Treat it the same way you would treat rank tracking for traditional SEO.
How Blazly makes this audit faster and easier to repeat
Running this audit manually, across four AI platforms and 15 to 20 questions, takes real time, and doing it consistently every month is where most teams fall behind. Blazly GEO is built to handle the entire process described above, automatically and on an ongoing basis.
The GEO Audit feature runs the visibility check described in steps one through three, scanning how your product shows up across ChatGPT, Gemini, Claude, and Perplexity, and flagging exactly why certain content is not being cited. AI Prompt Ranking handles the competitive benchmarking from step four, showing your position against named competitors for the specific questions your buyers actually ask, updated continuously instead of manually rerun each month. AI Competitor Research goes further, breaking down which competitors are winning which types of queries and what language AI models use to describe them.
For the content gap work in step five, the GEO Content Writer and GEO Landing Page Generator help you build the specific, clearly structured content AI models are more likely to trust and cite, based on the exact gaps the audit surfaces. And because visibility shifts over time, AI Citation Flow and GEO Reports keep the monthly tracking from step six running automatically, so you always know whether your position is improving before a competitor pulls further ahead.
Running this audit once tells you where you stand today. Running it consistently, with the right tools behind it, is what actually moves the number.
Frequently Asked Questions
How often should I audit my SaaS product's AI search visibility?
At minimum once a month, since AI models update frequently and competitors are actively working to improve their own visibility. Usage of AI tools in B2B research grew from 89% to 94% between 2025 and 2026, showing how quickly this landscape moves.
Which AI platforms should I include in an AI visibility audit?
Cover ChatGPT, Gemini, Claude, and Perplexity at minimum, since buyers use different platforms and results can vary significantly between them. G2's research found ChatGPT remains the most used platform for B2B software research at 63%, but it is not the only one buyers rely on.
What is the difference between being mentioned and being cited by AI?
Being mentioned means an AI answer includes your product name. Being cited means the AI is pulling information directly from your website or other owned content, which affects how accurate and current that description is likely to be.
Why does my product rank well on Google but not show up in AI answers?
Traditional SEO and AI citation rely on different signals. Moz's 2026 analysis found that 88% of Google AI Mode citations do not appear in the organic top 10 results, meaning a page can rank well without being structured in a way AI models trust enough to cite.
Can I run this kind of audit without specialized software?
es, the manual process is possible by asking the right questions across each AI platform and logging the results by hand. Tools like Blazly GEO simply make the process faster and repeatable, tracking changes automatically instead of requiring a manual recheck every month.