Tracking the invisible: how senior marketers can audit AI search traffic in GA4 and GSC

That's changed. Recent infrastructure updates across Google Analytics 4 (GA4) and Google Search Console (GSC) have introduced dedicated tooling to help brands explicitly measure how generative platforms interact with their web properties. The GA4 update went live on 13 May 2026. The GSC Generative AI performance report followed on 3 June 2026, initially rolling out to a subset of UK-based properties before a wider global release.

The GA4 breakthrough: the native AI assistant channel

Previously, tracking a user who clicked a link inside ChatGPT or Gemini meant building custom channel groups with complex regex patterns just to separate them from standard referral domains. Google has now removed that friction.

GA4 has rolled out a native update that isolates visits from conversational platforms into a standalone acquisition group. The update is not retroactive: sessions from before 13 May 2026 that previously landed in Referral or Direct remain under those original channel labels, which means historical month-on-month comparisons aren't possible using the default grouping alone.

Where to find it

Navigate to Reports > Acquisition > Traffic acquisition. Switch the primary dimension to Session default channel group and look for the new entry: AI Assistant.

When a user clicks through to a site from a supported AI assistant, GA4 automatically applies the following:

• Medium: ai-assistant
• Channel group: AI Assistant
• Campaign: (ai-assistant)

At launch, Google named ChatGPT, Gemini, and Claude as recognised referrers. The full list of covered platforms hasn't been published, so Perplexity, Microsoft Copilot, and others may or may not be included at this stage.

An important caveat on coverage

The channel only captures traffic where GA4 can detect a referrer header. When users click a link inside a native iOS or Android app, or copy and paste a URL, the referrer is typically stripped by the operating system. 

That session arrives looking like direct traffic. Independent analyses suggest this referrer-stripping affects somewhere between 20% and 40% of genuine AI-referred visits, so the AI Assistant channel should not be treated as the complete picture.

What the data shows

Early cross-industry benchmarks indicate that while AI referral traffic remains a small slice of total session volume (around 1% on average, per Conductor's 2026 data), the underlying user intent is exceptionally concentrated. 

According to Semrush's 2026 research, AI-referred visitors convert at roughly 4.4 times the rate of standard organic traffic. B2B-specific studies report a wider range, with some sectors seeing conversion rates between 6% and 14% for high-intent queries (Opollo, 2026). 

The rates vary considerably by industry and purchase type, so treating any single benchmark as universal isn't advisable.

With the native channel in place, analytics teams can run standard conversion, revenue, and lifetime value comparisons without relying on custom workarounds though.

The GSC evolution: the Generative AI performance report

While GA4 tracks the users who arrive at a site, Google Search Console launched a dedicated Generative AI performance report on 3 June 2026, designed to measure top-of-funnel exposure. 

The report is available under the Performance tab for eligible properties. 

At the time of writing, it's rolling out to a subset of UK-based site owners first, with a wider global release planned but not yet dated. 

The official announcement is available on the Google Search Central Blog, and Google's full documentation covers the report's specifications in detail.

What the data covers

The report provides a dedicated view of impressions across Google's AI Overviews, AI Mode, and generative features in Discover. You can filter this data by specific pages, countries, devices, and dates. 

There are two important constraints to note:

• Impressions only: the report currently shows how often your URLs appeared inside AI features, with no click or CTR data included. Google has indicated additional metrics will arrive over time, but there's no confirmed timeline.
• No historical backfill: data appears to begin from 18 May 2026. There's no access to earlier AI impression data.

Understanding which URLs Google's AI selects to ground its conversational responses helps technical teams identify which informational assets carry the highest structural authority. 

If a core commercial page has high impressions in the Generative AI report but very little movement in standard organic reports, it's a signal that Google's LLM considers that asset a trusted source for grounding its knowledge, regardless of whether users are clicking a standard link.

The regulatory driver

The June 2026 release was driven largely by regulatory pressure in the UK. 

The Competition and Markets Authority (CMA) issued a formal conduct requirement under the Digital Markets, Competition and Consumers Act 2024, which obliged Google to give publishers clearer attribution in AI-generated search results and meaningful controls over whether their content appears in AI features. 

Non-compliance carries a penalty of up to 10% of global annual turnover. The CMA's announcement and Google's response were published simultaneously on 3 June 2026.

The new GSC opt-out toggle: weighing the risk

Alongside the reporting updates, Google introduced a direct backend toggle within Search Console that lets webmasters opt out of having their content used to ground AI Overviews, AI Mode, and generative features in Discover.

This is an important clarification on how the toggle works: opting out removes a site from AI features only. Google has confirmed the setting won't be used as a ranking signal for standard search results. Sites that opt out continue to be indexed and ranked normally in conventional search and the Discover feed.

That said, for most brands that rely on AI citation for top-of-funnel discovery, opting out carries a real strategic cost. 

The toggle is most defensible for publishers whose business model depends on driving users directly to their site (advertising revenue, subscription paywalls) and who are concerned about content being summarised without generating clicks.

It's worth noting that the opt-out applies to Google Search products only. It doesn't affect how content surfaces in the Gemini app.

So, how can you use these changes? Let's look at some strategic use cases for reporting

Bringing this data into monthly performance reviews transforms speculative AI positioning into verifiable, revenue-supporting initiatives.

Use case 1: defending and proving conversational ROI

What it shows: by blending the GA4 AI Assistant default channel grouping with specific conversion event parameters, reports will show the exact lead volume, registration count, or revenue generated by users arriving from AI platforms. Be clear on what your key events are. 

How to use it in reporting: extract these values to establish an AI ROI baseline. When validating content budgets, this demonstrates how specific high-intent informational landing pages contribute to pipeline through conversational engines, and it gives you a comparison against standard organic traffic.

Use case 2: identifying informational grounding assets

What it shows: cross-referencing the GSC Generative AI performance report against standard organic landing page reports will flag pages that hold high structural value to LLMs despite low traditional organic click metrics.

How to use it in reporting: group these high-impression URLs into a grounding asset library inside tracking dashboards. Use this list to prioritise structural site optimisations, ensuring these URLs receive schema enhancements, deep factual citations, and regular validation to safeguard their position within Google's indexing layers. Red C's on-page SEO services cover schema implementation and E-E-A-T reinforcement for exactly this kind of work.

Use case 3: mapping the conversion attribution gap

What it shows: correlating spikes in GSC Generative AI impressions with localised lifts in direct or branded organic search traffic, even when traditional CTR remains flat.

How to use it in reporting: When a commercial page experiences a significant lift in Generative AI impressions, look at branded search performance. This bridges the conversational search attribution gap, showing how top-of-funnel generative visibility prompts buyers to conduct direct branded actions later.

The native GA4 AI Assistant channel and GSC Generative AI reporting are significant leaps forward for data transparency, but marketing operations teams need to understand that they only illuminate part of the picture. There's a substantial attribution gap in generative search. If a corporate buyer asks an LLM for the best enterprise infrastructure partners in the UK, reviews the summary, and then opens a new browser window to search for that brand directly via standard organic search, that traffic logs as a standard organic branded visit. It completely hides the AI touchpoint that generated the intent. True measurement requires a hybrid model.

Upgrading your measurement infrastructure

To capitalise on these updates and defend market share as search behaviours shift, organisations need to move beyond legacy tracking. Here are the three priorities:

• Audit your GA4 property: ensure data streams are cleanly capturing the new native ai-assistant medium values and that internal dashboards segment this traffic for conversion analysis. Because the channel isn't retroactive, it's also worth retaining any existing custom channel group rules to maintain coverage for platforms not yet on Google's recognised referrer list. Red C's SEO tagging and tracking services can help configure and validate this setup.
• Cross-reference GSC impression data: map out the pages Google is heavily leveraging for AI grounding and reinforce those assets with first-hand research, original media, and schema to maintain structural authority. Red C's content marketing services and off-page SEO work are designed to build and protect this kind of structural authority.
• Bridge the tracking gap: because AI search behaviour often maps to downstream branded search lifts, align organic reporting to catch those ripples. Red C's custom Looker Studio dashboards combine GA4, GSC, and SEMrush into a single reporting environment, making this kind of cross-channel correlation straightforward.

For a broader view of how AI is reshaping the search landscape, Red C's AI in Search whitepaper introduces the principles of Generative Engine Optimisation (GEO) and covers how brands can build visibility across both traditional and AI-powered search surfaces.

Getting the reporting infrastructure right

Navigating the architecture of LLM visibility and platform analytics requires careful, data-led execution. To ensure reporting systems are built to capture every element of AI-driven intent, explore the SEO reporting services and SEO tagging and tracking services offered by Red C's technical teams. 

Visit Red C's search engine optimisation hub to explore the full range of services, or get in touch to discuss how your current measurement setup can be adapted for the new generative search landscape.