Google Analytics 4: How to measure AI assistant traffic on your website

Why Google Analytics 4 Created a Dedicated Channel for AI Assistants

Since May 2025, Google Analytics 4 displays a new channel in its acquisition reports: AI Assistants. This evolution isn't merely cosmetic. It addresses a reality that B2B marketing teams have been observing for months: a growing share of qualified traffic comes from ChatGPT, Perplexity, Claude, and other generative engines.

Until now, this traffic was buried in the "Direct," "Referral," or worse, "Unassigned" categories. It was impossible to distinguish a visitor arriving via a Perplexity recommendation from direct access through bookmarks. This opacity made managing GEO (Generative Engine Optimization) strategies particularly frustrating for executives investing in this emerging channel.

The new AI Assistants channel group solves this problem by automatically identifying the following sources: ChatGPT (chat.openai.com), Perplexity (perplexity.ai), Claude (claude.ai), Google Gemini, Microsoft Copilot, and several other conversational assistants. You finally have a clear view of what generative engines actually bring to your website.

How to Activate and Verify the AI Assistants Channel in GA4

The good news: the AI Assistants channel is activated by default for all GA4 properties since the May 2025 update. No manual configuration is required. Google has added this group to the default channel list, alongside "Organic Search," "Paid Search," and "Social."

Verify the Channel Appears in Your Reports

To confirm that your GA4 property is properly capturing this traffic, follow these steps:

• Log into Google Analytics 4 and select your property
• Go to Reports > Acquisition > Acquisition overview
• In the channels table, look for the "AI Assistants" row
• If it doesn't appear, it simply means you haven't yet received identified traffic from these sources

Understanding Classification Rules

GA4 identifies AI Assistants traffic based on two main criteria: the referrer (source domain) and, in some cases, UTM parameters. Specifically, a visitor arriving from perplexity.ai will be automatically classified as AI Assistants, without any action on your part.

However, be aware that some AI assistants, particularly mobile applications or API integrations, may not transmit referrer information. This traffic will continue to appear as "Direct." Google is progressively improving detection, but a portion of AI traffic remains invisible in analytics today.

Which Metrics to Track for Evaluating GEO Impact

Having a dedicated channel isn't enough. The challenge is transforming this raw data into decision-making indicators. Here are the priority metrics for B2B executives who want to measure the return on investment of their GEO strategy.

Session Volume and Trends

The first indicator to monitor is the evolution in the number of sessions from the AI Assistants channel. At AISOS, we observe that B2B companies well-positioned on business queries see this traffic grow 15 to 40% per quarter. Stagnation may signal a visibility problem in generative responses.

Create a custom report with a comparison period (previous month or quarter) to identify trends. Regularly growing AI traffic validates that your content is effectively being cited by assistants.

Engagement Rate and Session Duration

AI traffic has a unique characteristic: visitors often arrive with very specific intent. They've asked a question to ChatGPT or Perplexity, received an answer mentioning your company, and click to learn more. This context generally generates more qualified sessions.

Compare these metrics between the AI Assistants channel and Organic Search:

• Engagement rate: percentage of sessions with meaningful interaction
• Average engagement time: time spent actively on the site
• Pages per session: number of pages viewed

An engagement rate higher than organic search indicates that your GEO positioning attracts the right profiles. Conversely, a low rate may reveal a disconnect between the promise made by the AI assistant and your page's actual content.

Conversions Attributed to the AI Channel

This is the metric that interests executives most: how many leads or contact requests come from AI traffic? To obtain this, you must have configured conversion events in GA4 (form submitted, resource download, demo request).

Go to Reports > Acquisition > Traffic acquisition, then add a secondary dimension "Conversion event" to isolate conversions by channel. You can also create a custom exploration to cross-reference acquisition channel with business objectives.

Creating a GEO Dashboard in GA4

To effectively manage your GEO strategy, centralize key indicators in a dedicated dashboard. GA4 allows you to create custom explorations that you can save and share with your team.

Step 1: Create a Free-Form Exploration

In GA4, go to Explore > Create exploration and choose the "Free form" format. This format offers the flexibility needed to build a multi-dimensional report.

Step 2: Configure Dimensions and Metrics

Add the following elements to your exploration:

• Dimensions: Session default channel group, Session source, Session medium, Landing page
• Metrics: Sessions, Users, Engagement rate, Average engagement duration, Conversions, Revenue (if applicable)

Step 3: Filter on AI Assistants Channel

Apply a filter to display only AI Assistants channel data. You'll get a detailed table showing specific AI sources (ChatGPT vs Perplexity vs Claude), the most visited landing pages from these sources, and associated performance.

Step 4: Add Temporal Visualization

Complete your dashboard with a line chart showing the evolution of AI Assistants sessions over the last 90 days. This view allows you to quickly identify traffic spikes and correlate them with your actions: publishing a new article, updating a strategic page, PR campaigns.

Interpreting Data: What AI Traffic Reveals About Your Visibility

Raw numbers only tell part of the story. Here's how to interpret GA4 data to make concrete decisions.

Case 1: Low AI Traffic Despite GEO Efforts

You've optimized your content for generative engines, but the AI Assistants channel shows disappointing volumes. Several hypotheses to explore:

• Citation problem: your content may be used by LLMs to formulate their responses, but without clickable links to your site. AISOS audits reveal that 60 to 70% of citations in AI responses don't generate clicks.
• Tracking problem: some AI traffic arrives without a referrer and gets classified as "Direct." Analyze the parallel evolution of your direct sessions.
• Positioning problem: your content simply isn't being picked up by assistants for your target queries. Test manually by asking questions to ChatGPT and Perplexity.

Case 2: High AI Traffic but Low Conversions

Session volume is satisfactory, but conversion rate remains below other channels. This scenario typically indicates an alignment problem between visitor intent and your landing page.

Analyze the main landing pages of AI traffic. If visitors arrive on informational articles but your conversions are demo requests, add contextual calls-to-action in this content. Visitors from AI assistants often seek to deepen a topic: offer them a complementary resource (guide, case study) in exchange for their contact information.

Case 3: Traffic Concentrated on a Single AI Source

If 90% of your AI Assistants traffic comes from Perplexity and almost nothing from ChatGPT, you've identified an opportunity. Perplexity systematically cites its sources with links, explaining its overrepresentation in analytics. ChatGPT generates fewer clicks because its responses rarely include URLs.

This distribution doesn't mean you're invisible on ChatGPT. It reflects different platform behaviors. To evaluate your actual visibility on ChatGPT, complement GA4 analysis with manual tests or dedicated GEO monitoring tools.

Current Limitations of AI Tracking in GA4

The AI Assistants channel represents a major advancement, but it doesn't capture the full impact of generative engines on your visibility. Here are the limitations to know to avoid erroneous conclusions.

Referrer-less Traffic Remains Invisible

Mobile applications of AI assistants, API integrations, and certain browser configurations don't transmit referrer information. This traffic appears as "Direct" in GA4. Google estimates that 20 to 30% of actual traffic from AI assistants escapes automatic classification.

Citations Without Clicks Aren't Measured

An AI assistant can mention your company, product, or expertise in its response without including a link. The user retains the information but never visits your site. This brand awareness impact, equivalent to an ad impression, appears nowhere in GA4.

Attribution Remains Simplistic

GA4 attributes the session to the last identified channel. If a prospect discovers your company via a ChatGPT response, then returns three days later by typing your URL directly, the conversion will be attributed to the "Direct" channel, not AI Assistants. GA4's multi-touch attribution models can partially correct this bias, but require advanced configuration.

Concrete Actions to Improve Your Measurable AI Visibility

GA4 data gives you a factual foundation. Here's how to act to increase both your presence in AI responses and the measurable traffic that results.

Optimize for Citations with Links

Perplexity and Google AI Overview systematically cite their sources. To maximize your chances of being cited with a clickable link:

• Publish factual content with original data (studies, benchmarks, industry statistics)
• Structure your pages with direct answers to your audience's frequent questions
• Maintain high domain authority through quality backlinks

Facilitate Brand Identification

When an AI assistant mentions "according to a recent study" without citing the source, the traffic opportunity is lost. To make LLMs explicitly associate information with your brand:

• Include your company name in key passages of your content
• Create detailed "About" and "Team" pages that LLMs can use to contextualize your expertise
• Develop your presence on sources LLMs consult: Wikipedia, LinkedIn, industry media

Correlate Analytics and GEO Monitoring

GA4 measures incoming traffic. To complete this view, use tools that monitor your presence in AI assistant responses, whether there are clicks or not. Combining both data sources offers a complete picture: visibility (citations) and conversion (traffic, leads).

What This Evolution Means for Your Digital Strategy

Adding the AI Assistants channel in GA4 officially recognizes what GEO specialists have been asserting for two years: traffic from generative engines has become a distinct acquisition channel for B2B companies.

For SME and mid-market executives, this evolution significantly simplifies reporting and decision-making. You can now answer essential questions: what volume of traffic do AI assistants actually generate? Does this traffic convert as well as organic search? Do my GEO content investments produce measurable results?

Companies that leverage this data now gain a competitive advantage. They identify content that performs in AI responses, understand which assistants bring them the most value, and adjust their strategy accordingly.

Set up your GEO dashboard in GA4 this week. Analyze your first months of data. And if the numbers reveal untapped potential, now is the time to seriously invest in your visibility on generative engines. AISOS supports B2B companies in this transition, from initial audit to deploying a complete GEO strategy.