If you've spent any time thinking about AI visibility, you've probably come across tools that promise to track how often your brand or content gets cited by AI assistants like ChatGPT, Gemini, Perplexity, or Claude. Most of these tools work by running a large set of human-crafted prompts repeatedly and counting how often your site shows up in the responses.
I don't love this approach. Running thousands of synthetic queries inflates Search Console impression numbers. And these prompts are not written by real users and don't necessarily reflect how people actually search.
That said, I do think there's value in spot-checking specific prompts. If I want to know whether a particular article is being cited correctly, or whether there's a factual error in how an AI describes my product, running a targeted prompt manually tells me what I need to know. The problem is that rerunning the same prompts on a schedule doesn't scale well and doesn't tell you much that changes week to week.
More importantly, many of the AI visibility tracking tools are freaking expensive. On top of your other regular SEO tech stack, adding another AI visibility tool is often not financially feasible for freelancers or smaller businesses.
So what's a more grounded approach? Here are two free methods I actually use.
Method 1: Use Microsoft Clarity AI visibility dashboard
Microsoft Clarity is a free behavioral analytics tool most people know for heatmaps and session recordings. As of May 2026, it has a dedicated AI visibility dashboard that tracks when and how your content is actually being cited inside AI-generated answers.
Unlike most AI visibility in the market, Microsoft gives you first-party insights into how an AI system retrieves and references your content as a source.
Clarity AI visibility dashboard shows:
- Grounding queries: The actual questions or prompts that AI systems used when retrieving your content. This is closer to real user intent than anything a third-party prompt-tracking tool manufactures.
- Cited pages: A page-level breakdown of which specific URLs on your domain were cited, along with citation counts and the queries associated with each.
- Share of authority: How your domain's citation presence compares to competing domains across the same query space.
- AI-referred traffic: How citation activity translates into actual referral visits over time.
- Trendlines: Track how citation patterns shift as your content evolves or as AI query behavior changes.
The grounding queries are particularly useful for the kind of spot-checking I mentioned in the intro. Instead of manually re-running the same prompts to test whether your content is being cited, you can look at the queries Clarity has already captured and verify whether the AI's interpretation of your content is accurate. It flips the workflow: rather than guessing which prompts to test, you're starting from queries that are already tied to real citation activity.
At the time of writing, Google just announced their own Generative AI performance dashboard, but they only show impressions. That’s relatively useless as you don’t have any insights into the queries at all to make the data actionable. It’s better than nothing, but so underwhelming compared to what Microsoft’s offering.
Method 2: Use Google Analytics to track AI referral traffic
Google Analytics (GA4) is the most straightforward way to see whether AI tools are sending traffic to your site at all. When a user follows a link from an AI assistant, the browser typically passes along a referrer, and GA4 captures it under the Traffic Acquisition report.
Recently, GA4 added the "AI Assistant" channel in the Default Channel Group reports. However, I would also recommend creating a custom channel group to bucket all known AI sources together, which makes it easier to track the combined trend over time and make sure you have included all relevant AI sources.
What you can learn from this:
- Which AI tools are actually sending you traffic (versus which ones mention you but don't result in clicks).
- Which pages on your site are receiving AI-referred visits — this is a strong signal that those pages are being cited.
- How AI-referred visitors behave compared to organic search visitors: engagement time, pages per session, conversions.
This means GA4 helps you benchmark the AI referral traffic. The traffic you see is real, but the actual number of times your content influenced an AI response is almost certainly higher. So treat it as a directional signal rather than a complete picture.
The only downside is that not all AI-generated traffic is captured. If the AI chatbot doesn’t append any page referrer, then the source will not be passed to GA4. When someone asks ChatGPT a question and reads the answer directly in the interface without clicking a link, that visit never happens, so it never shows up in your analytics. Some browsers and privacy extensions also strip referrers.
Turn AI visibility data into actionable SEO decisions
To be honest, neither of these methods gives you a perfect view of AI visibility, and that's fine. The goal is to understand whether real users are finding you through AI tools and whether the experience they have when they arrive is a good one.
While Microsoft’s AI visibility tells you which queries are driving that activity, GA4 tells you how often that citation activity converts into actual visits and which pages are pulling their weight. Together they cover both sides of the funnel: AI-side citation signals and human-side referral behavior.
To make this actionable, I often turn the data into a short list of pages and queries to decide my next steps. Start with the AI-referred visits in GA4, then check which landing pages are getting traffic and whether those users engage, convert, or leave quickly. Use Microsoft Clarity to identify the grounding queries behind those visits, then compare them against the page content.
If a query is sending users to a page that only partially answers the question, improve that section, add missing context, strengthen internal links, or create a more focused supporting page. This turns AI visibility from a vague reporting metric into a practical content optimization workflow.