The Invisible Metric: Why Ahrefs & SEMRush Can’t Track Your “Al Traffic” Yet

You are constantly refreshing your analytics dashboard. Your direct traffic is spiking. A significant uptick, but no corresponding increase in known organic or paid channels. You are stumped.

We’ve all seen it. You monitor the search queries, the brand mentions. Then you look at GA4.

The problem? Most of that AI referral traffic is currently invisible. Or, worse, it’s mislabeled.

An Ahrefs study pegged LLMs at approximately 0.1% of traffic in 2025, yet that figure feels low when 89% of B2B buyers now use AI for research. Something doesn’t add up.

You know your content is getting consumed by AI models like ChatGPT and Perplexity. It has to be, given the sheer volume of queries. But your analytics dashboard often shows a flatline for those crucial referrals.

That’s because AI platforms are changing how much referral traffic actually hits publisher sites. Google’s AI Overviews, for instance, significantly reduced click-through rates across more search queries in 2025. Your content is getting cited, maybe summarized, but not always clicked directly. This makes tracking genuine engagement a nightmare.

This misclassification is a blind spot. Most of that potential AI referral traffic is mislabeled in GA4 as ‘Direct’ or ‘Unassigned’ because key UTM parameters get stripped. You’re effectively operating in the dark.

This article pulls back that curtain. We’ll show you how to unmask these hidden data streams and optimize your content for what’s now called Generative Engine Optimization (GEO). We will define GEO as the strategic process of optimizing content for visibility and favorable citation within AI-generated responses, helping brands achieve higher authority and engagement in AI search environments. (For a broader view, check out our guide on What Is Generative Engine Optimization). You can’t fix what you can’t see.

What is AI referral traffic?

AI referral traffic is the flow of engaged traffic that reaches your website when a user clicks a citation link within an AI-generated response from platforms such as ChatGPT, Claude, Gemini, perplexity or Google AI Overviews. Think of AI citations as the new, evolving form of backlinks.

Instead of a traditional backlink within a blog post, consider it a “mention with a link” inside an AI’s answer. A user asks a question, the AI synthesizes information, and cites your content as a source, providing a clickable link.

The game has changed.

The challenge is tracking this traffic. As highlighted earlier, AI platforms are shifting the landscape of referral traffic, often stripping UTM parameters or misclassifying it in analytics dashboards.

Tracking AI referral traffic is doable. You can set up a new channel and source in GA4. But that’s only half the battle. To truly capitalize on this traffic, your content needs to be optimized not just for traditional search results, but for AI-generated responses. This is where Generative Engine Optimization (GEO) comes into play. (You can learn more about optimizing your content for AI within our guide on AI SEO and AEO in 2026).

Next up, how do you actually see this so-called invisible traffic?

How referral traffic evolved before AI

Referral traffic wasn’t always about AI citations. Before the AI era, it was a simpler game of links and clicks.

Think back to the early internet. Referral programs hinged on basic directory links. Sites like Yahoo! and DMOZ were kingmakers. Getting listed meant eyeballs, and eyeballs translated to traffic. The intent was pure discovery: find something new.

Then came the social media boom. Facebook, Twitter, and later Instagram and TikTok became the new referral hubs. Sharing content became effortless, but the intent shifted subtly. It wasn’t just about discovery anymore; it was about social validation.

And the shift has accelerated with the rise of AI-driven “answer engines.” Now, users aren’t necessarily discovering new content through referrals. They’re validating AI’s answers.

The game is no longer dependent on pure SEO.

• Directory Era: Manual submissions, broad reach.
• Social Era: Easy sharing, social validation.
• AI Era: Algorithmic citations, authority validation.

The real issue is this: are you optimizing for validation as much as discovery? If not, your brand is losing out.

Why standard SEO tools fail to track LLM visits

Standard SEO tools miss LLM visits because of fundamental design limitations. Ahrefs and SEMRush primarily rely on web crawlers to discover and index content. But LLM interactions happen within closed, often API-driven chat environments.

Think of it this way: Ahrefs’ bot can’t “ask” ChatGPT a question and then follow the outbound links in the AI’s response. It can crawl the destination website later. But that’s not real-time LLM tracking.

This reliance on crawler data creates a critical blind spot.

The challenge lies in the fact that AI platforms often don’t pass traditional referrer information or UTM parameters. The AI summarizes, and links to, your content. But it does so in a way that evades standard web analytics protocols.

Here’s why this matters:

• Crawlers mimic typical user behavior.
• AI platforms behave differently.
• Standard SEO tools can’t see traffic coming from these interactions.

SEO tools aren’t useless. They still provide value in traditional search rankings. But they fail to show you the full picture of how your content is being consumed and cited by AI.

The bigger issue is misinterpreting this data. That spike in “direct traffic” might not be users typing in your URL. It could be the echo of an AI citation. Without proper tracking, you’re making decisions based on incomplete data.

We built FlipAEO to fill this gap optimizing content for AI answers. We saw the problem firsthand. Our clients’ high-traffic pages were underperforming in AI-generated outputs. They needed a clear view of where, how, and how much referral traffic they were getting. You can track AI traffic in GA4 by setting up a new channel and source.

Ultimately, the future of SEO lies in adapting to these new realities. You must go beyond traditional metrics and learn to measure your brand’s presence within the AI ecosystem. Next step, understand how referral traffic evolved. (Hint: It wasn’t always about AI citations). For further reading, our guide on SEO vs AEO vs GEO might help.

How AI platforms strip and withhold referrer data

AI platforms strip and withhold referrer data by several technical methods. One common tactic is failing to pass UTM parameters in citation links.

These parameters (like utm_source, utm_medium, utm_campaign) are how analytics tools like GA4 attribute traffic to specific sources. When they’re missing, the traffic defaults to “Direct” or “Unassigned”. (It’s like showing up to a party without a name tag.)

Another method involves the removal of the HTTP referrer. The HTTP referrer is a header in the web request that tells the destination website where the user came from. AI platforms can suppress this header, either intentionally or due to privacy configurations. The result? More “Direct” traffic in your reports.

• Missing UTM parameters
• Stripped referrer data

But why do they do this?

Well, the reasons are varied. It could be an effort to protect user privacy, or perhaps an attempt to consolidate data within their own ecosystems. Whatever the cause, the impact is clear: You’re left with an incomplete picture of your AI referral traffic. A big problem because as Ways To Own Your Brands Authority In 2026 becomes more critical, understanding your traffic source is paramount.

The bigger issue: The traffic is happening. Your content is getting cited. But you’re not getting credit where it’s due. You need to start looking at other data sources. And stop trusting GA4 to tell the whole story.

What data sources? I’ll show you…

The GA4 misclassification problem

GA4’s default channel grouping often dumps AI traffic into buckets where it doesn’t belong. Think of it like trying to fit a square peg into a round hole.

The problem? The standard GA4 setup struggles to differentiate AI-driven visits from other sources, causing traffic misclassification. This is a massive issue. You’re essentially flying blind when trying to understand the impact of AI on your content strategy.

ChatGPT referral traffic nearly doubled in 2025, yet remains hidden for most publishers because GA4’s default settings don’t account for the nuances of AI referrals. That traffic gets lumped into “Direct” or “Referral” without proper attribution.

This means missed opportunities. You might be underestimating the value of your AI citations and failing to optimize your content accordingly. You’re missing out on critical data that could inform your content strategy and improve your AI visibility. According to recent trends regarding how AI platforms change referral traffic flow to publisher sites, as reported by Digiday, this issue isn’t going away anytime soon.

Worse, it gives you a false sense of security. You think your content is performing fine, when in reality, a significant portion of your traffic is being misattributed. You’re making decisions based on incomplete data, and that can have serious consequences for your brand.

FlipAEO saw this gap and built features to make the GA4 set up process simple and accurate. To solve this, you can create a new channel in GA4 specifically for AI referral traffic. And you need to train GA4 to recognize the unique patterns and sources of AI referrals. Don’t just “set it and forget it”; regularly monitor and adjust your GA4 settings to adapt to the ever-changing AI landscape.

But, what alternative data sources should you look at?

Tracking AI referral traffic in GA4 vs Ahrefs

GA4 and Ahrefs offer very different views of the AI traffic landscape. One measures clicks; the other, share of voice.

GA4 is about retroactive measurement. It’s your post-game analysis. You tweak the settings, set up custom channels, and try to piece together where that “Direct” traffic really came from.

Ahrefs, specifically Ahrefs Brand Radar, is about prospective monitoring. It doesn’t tell you exactly who clicked from an AI citation. What it does tell you is when and where your brand is being mentioned by AI platforms. Think of it as competitive intel.

The core difference comes down to methodology:

• GA4: Analyzes website traffic AFTER it arrives, attempting to classify the source.
• Ahrefs Brand Radar: Scans the web for brand mentions (including in AI outputs) BEFORE traffic hits your site.

But Ahrefs Web Analytics itself falls short. Like SEMrush, it relies on crawling, which, as we covered, misses the nuances of AI-driven interactions. It can tell you about backlinks from traditionally indexed web pages, but not necessarily about a citation in a ChatGPT response.

Ahrefs Brand Radar steps in, and you can use the data to gauge your overall “share of voice” in the AI space. Are you being cited more or less than your competitors? Are certain AI platforms mentioning you more frequently? Use those insights to refine your content strategy.

There’s a catch, however.

Ahrefs Brand Radar isn’t cheap. The plans start high, and the interface can feel overwhelming. It can take hours just to set up proper alerts. GA4, meanwhile, is “free” (though the devil is in the data limits).

Here’s the real issue. Neither is a complete solution on its own. GA4 tells you where the traffic landed. Ahrefs Brand Radar tells you where your brand is being cited. They work best in tandem. FlipAEO was built specifically to track AI visibility.

Next step: alternative data sources to track that traffic…

Creating a custom AI traffic channel in GA4

Creating a custom AI traffic channel in GA4 isn’t just “nice to have”; it’s essential for understanding where your AI referral traffic originates. Without it, you’re guessing.

Here’s how to set it up, step-by-step.

1. Navigate to Admin: In your GA4 property, click “Admin” (the gear icon) at the bottom left. Find the “Data Streams” section. And pick your web data stream.
2. Configure Tagging Settings:
3. Scroll down and select “Configure tag settings”.
4. Then, choose “List unwanted referrals”.
5. Define the Regex: This is where the magic happens. You’ll create a regular expression (regex) to capture traffic from key AI platforms. Here’s a starting point that we use:

Feature	Details
Regex Pattern	`openai
Target Platforms	OpenAI (ChatGPT), Perplexity AI, Anthropic (Claude)
Function	Instructs GA4 to identify referral traffic coming from any domain containing these specific keywords.
Maintenance	High Frequency. The list must be updated manually as new AI platforms are released.
Recommended Tool	Ahrefs Brand Radar (suggested for early detection of new AI traffic sources).

1. Create Custom Channel Grouping: Now create an AI Referrals channel grouping.
2. In the Admin section, navigate to “Channel groups.”
3. Click “Create new channel group.”
4. Name it something descriptive: “AI Referrals.”
5. Add a Condition: This tells GA4 when to classify traffic as “AI Referrals.”
6. Set the dimension to “Referral source.”
7. Set the match type to “matches regex.”
8. Paste your regex from step 3 here.
9. Save…But Don’t Walk Away: Click “Save”. Now, here’s the crucial step most people miss: Test it.
10. GA4 data isn’t real-time. You’ll need to wait 24-48 hours to see data populate in your new channel.
11. After 48 hours, check your reports. Go to “Reports” > “Acquisition” > “Traffic Acquisition.”
12. Select your new “AI Referrals” channel as a secondary dimension. Do you see the traffic you expected?

This setup isn’t perfect. GA4 relies on patterns. AI platforms can, and will, change their referral methods. Your regex filter will need constant tweaking. As the way B2B buyers research purchases continues to change, with 89% now using AI, staying on top of these sources is critical. But here’s the hard truth: This is a cat-and-mouse game. They hide the data; you find it.

What are some alternative data sources to track it?

Why GEO is more than just traditional SEO

GEO isn’t just SEO 2.0. Instead, it pivots from ranking in a list to becoming the definitive source within an AI’s synthesis.

SEO, in its traditional form, focuses on algorithms that rank web pages based on keywords and backlinks. You optimize for a position on a search engine results page (SERP). GEO, however, aims to get your content directly cited and referenced within AI-generated content. The shift is about becoming the answer, not just appearing in a list of possible answers.

Consider this:

• SEO: Optimizes for a search engine’s ranking algorithm.
• GEO: Optimizes for a generative AI’s citation algorithm.

The difference impacts content strategy. With SEO, you’re chasing keywords. With GEO, you’re establishing authority.

One strategy involves claiming very specific niches. Are you the only site that publishes a comprehensive guide on “fixing error 3194 on iOS 9.3.3?” Generative AI is likely to reference your page. It’s no longer about broad keywords; it’s about capturing ultra-specific expertise.

And you’re not just optimizing for visibility in standard search results anymore; you’re also trying to show up in AI-generated content.

GEO requires a different toolset and mindset. FlipAEO built features to meet this different strategy head on, specifically to track AI visibility.

SEO isn’t dead. It’s evolving. The rise of GEO has changed how we think about the relationship between search engines and content.

Content types that win AI citations

Data-heavy reports, unique insights, and structured “how-to” guides attract the most brand citations from AI platforms. Think of AI as a super-powered research assistant. It craves information.

AI models prefer content that’s easily digestible and authoritative. They are biased to factual authority.

• Data-driven reports with original research.
• Step-by-step “how-to” guides with clear instructions.
• Unique insights and perspectives that aren’t readily available elsewhere.

But the magic lies in refreshing high-traffic pages.

Don’t let those pages stagnate. Audit them for AI optimization. That means updating statistics, adding new examples, and clarifying any confusing language. (By the way, don’t just change the publication date, do the work).

The bigger issue is making sure that information is easily digestible by AI. That means structuring your content with clear headings, subheadings, and bullet points. Use tables to present data and highlight key takeaways. Don’t make the AI model work harder than it needs to. Help it help your audience.

Consider this process:

1. Find your highest traffic pages in GA4.
2. Assess AI performance in Ahrefs.
3. Refresh the top 20% based on authority.

The goal? Become the definitive source. This is a game of constant refinement.

Tools to audit your AI visibility

Writesonic GEO isn’t the only game in town when you need to audit your AI search visibility, but it’s a start. Several tools can give you insights into how your brand is being perceived, and cited, by LLMs.

Tools like Ahrefs Brand Radar help you track brand mentions, but they only scratch the surface of true sentiment analysis within AI outputs. You need to know how AI is talking about you, not just that it’s talking about you. That requires diving deeper.

Here’s the rub:

• Writesonic GEO focuses on tracking AI visibility and optimizing content for AI answers.
• Ahrefs Brand Radar tracks when and how AI platforms reference your content.

But, sentiment analysis? That’s still largely a manual game.

Right now, the best approach combines automated tracking with human review. Monitor your brand mentions with tools like Ahrefs, then manually analyze the context of those mentions within AI-generated responses. Is the AI presenting your brand accurately? Is it highlighting your strengths or weaknesses? Is it recommending you alongside your competitors?

This is where FlipAEO comes in. We built it as a strategic engine for brands effectively invisible to AI. It helps you in finding out how you’re being perceived. It’s about understanding the narratives AI is building around your brand, so you can proactively shape them.

Because as AI becomes more integrated into the buying process, your brand’s AI reputation becomes everything.

What’s next? Use those citation insights to influence AI platform with authority.

How to judge AI traffic quality

AI referral traffic often exhibits higher engagement rates. But it’s not a guaranteed win. You need to dive into the data and actually see if that holds true for your content.

How do you judge AI traffic quality? By looking at session duration and conversion data for those specific users.

First, segment your AI traffic in GA4. (You did set up that custom channel, right?). Then, compare those visitors to traffic from other sources.

• Session Duration: Are AI referrals spending more or less time on your site? Are they reading entire articles, or just skimming?
• Conversion Rate: Are AI referrals more likely to sign up for a newsletter, request a demo, or make a purchase? Or are they simply validating the AI’s recommendation and bouncing?

But here’s the tricky part: User intent matters.

Someone clicking a link in ChatGPT might be looking for quick validation of a fact. Someone clicking from Perplexity might be seeking a deeper understanding of a topic.

You can look at scroll depth to know if people are reading your content.

To make sure that your intent is clear, review what you wrote, clarify the purpose of your information, and make sure you deliver what is promised. Do you really want that traffic? Are you providing the right experience once they arrive?

If you’re seeing low session duration and high bounce rates, it’s a sign that your content isn’t meeting the needs of those AI-driven visitors. Either you need to improve your content, or you need to refine the context of citation.

Next step? Use those citation insights to influence AI platforms with authority.

Ethical concerns regarding AI data privacy

Ethical lines blur when AI scrapes content for training data. The promise of AI platforms is increased traffic. The reality is more complex: they might replace the need to visit your site entirely.

The trade-off is this: Visibility versus brand cannibalization. They are not equivalent. Is a fleeting citation worth more than a dedicated reader?

AI models need data. Your content is that data. But the terms of that exchange are murky. Do you implicitly consent to AI scraping simply by publishing online? Or do publishers have the right to control how their content is used to train these models? A lawsuit brought by the New York Times against OpenAI attempts to establish legal ground rules.

• Traffic Promise: AI platforms offer the potential for increased referral traffic via citations.
• Cannibalization Threat: AI-generated summaries may satisfy user queries without driving clicks to the original content.

The bigger ethical issue lies in data privacy. AI models are trained on vast datasets. Those datasets can contain personal information scraped from websites, social media, and other sources. What safeguards are in place to protect that data? What rights do individuals have to control their data? As these AI models learn and evolve, are they transparent about their data sources and the potential biases those sources may introduce? We believe in visibility and control; so much so that we build our platform for that. That will be a key tenet in AI.

This is where publisher rights come into play. Should publishers be compensated when their content is used to train AI models? Should they have the right to opt out of AI scraping altogether? What if the content includes personal data or sensitive information? There are more questions than answers.

Because what are they willing to do to own your brand’s authority?

What AI traffic looks like after 2026

After 2026, prepare for the slow death of the click-through rate as we know it. The rise of AI Overviews means fewer direct clicks, period. The game is shifting from chasing clicks to capturing brand mentions.

Clicks will matter less.

The future isn’t about driving traffic directly from AI responses; it’s about influencing the AI’s perception of your brand. It’s about becoming the source of truth that AI platforms consistently cite and recommend.

Think of it as a reputation war.

The new KPI? “Share of citation.” How often does your brand get mentioned, positively, in AI-generated content compared to your competitors? This demands a shift in strategy.

We built FlipAEO to handle this shift. Where other tools show you where your traffic is now, we want you to be prepared for where it’s going.

Traditional SEO focused on ranking; GEO demands authority. This means going beyond keyword optimization. It means creating content so valuable, so insightful, and so uniquely “you” that AI can’t help but cite it.

Remember, the AI is only as good as its sources. Be the best source.

But watch out: the algorithm will evolve. AI platform’s ability to synthesize and present information will continue to improve. Meaning, your old content may not be enough. You can’t “set it and forget it.” You must continuously monitor and adapt your strategy.

The bigger challenge? Measuring the impact of those mentions. How do you tie an AI citation to real-world business outcomes? That’s the puzzle we’re all going to be solving in the coming years. But you can get ahead of this by understanding how referral traffic evolved.

Common questions about AI visibility

Think you can skip AI SEO? Think again. Many believe that AI traffic is some kind of mirage. And others assume you can’t really influence what AI answers. Both are wrong.

Those are common AI SEO FAQs. But it’s more than just SEO—it’s about positioning your brand as the definitive source that AIs choose to cite.

• Can you track AI traffic? Yes, with the right tools and setup.
• Can you influence AI answers? Absolutely, by building topical authority and providing unique insights.

The bigger issue is understanding how to do it.

You need a strategy for more than just traditional SEO. You need Generative Engine Optimization (GEO). For a more detailed definition and strategy for Generative Engine Optimization, check out our full blog post here. The game’s changed. You should, too.