How to Track ChatGPT, Perplexity and Gemini Traffic in GA4

Yes, you can attribute ChatGPT, Perplexity and Gemini traffic in GA4, but the answer differs by platform. ChatGPT sessions are the most reliably classified: Google names ChatGPT explicitly in its own documentation, and the platform appends UTM parameters to some outbound links. Perplexity and Gemini are less certain out of the box. Google's 11 June 2026 Source Group note names Perplexity as a recognized AI source, but Perplexity does not appear on the AI Assistant help page's own example list, and Gemini traffic depends on Google's unpublished referrer roster the same way every other platform does. The only way to know for certain which of your own Perplexity or Gemini sessions GA4 already classifies correctly is to check your own property: open Reports > Acquisition > Traffic acquisition, set the dimension to Session source / medium, and look at how sessions from perplexity.ai or gemini.google.com are labeled. A medium of ai-assistant means GA4 already recognizes that referrer for you; a medium of referral means it does not, and the custom channel group in this guide is what closes the gap.

What Google documents about the AI Assistant channel. Google does not publish the list of referrers its AI Assistant channel recognizes — its help page names "sources like ChatGPT, Gemini, Deepseek, Copilot, or Grok" as examples and defines the rule only as "if the referrer matches a list of AI Assistants" — so no one can confirm a platform's inclusion or exclusion from the documentation. Google's own 13 May 2026 release note names Claude, and its 11 June 2026 release note names Perplexity as a built-in AI source grouping. The recommendation to build a custom channel group rests on the reason given below.

Method note. Every statement on this page about what Google does and does not publish regarding the AI Assistant channel was checked on 21 September 2026 against two Google-owned pages and nothing else: the Analytics Help article "Default channel group" at https://support.google.com/analytics/answer/9756891 (channel description, and the channel definition table entry for AI Assistant) and "What's new in Google Analytics" at https://support.google.com/analytics/answer/9164320 (the 13 May 2026 and 11 June 2026 release entries). Those two documents are the complete public surface Google maintains for this channel; "Google does not publish X" on this page means X does not appear in either of them on that date. Google can change either without notice, which is why the page tells you to verify in your own property rather than trust this list.

AI traffic attribution in GA4 works best when you combine Google's native AI Assistant channel with a custom AI Search channel group. The native channel classifies referrers Google recognizes, and Google does not publish that list. Your custom rule gives you an explicit, auditable roster of your own plus the ChatGPT UTMs and referrer domains you want counted your way.

This guide gives you the exact setup: native channel verification, custom regex filters for major AI platforms, per-platform referrer patterns, UTM handling, and a GA4 dashboard that shows which pages AI engines send buyers to. AI traffic attribution is the measurement layer of Machine Relations. Without it, you cannot connect earned media placements to the AI sessions they generate or calculate ROI on citation architecture.

Key Takeaways

• Google does not publish which referrers GA4's native "AI Assistant" channel recognizes. The help page gives ChatGPT, Gemini, DeepSeek, Copilot and Grok as examples, not as a roster; Google's own release notes separately name Claude and Perplexity. Because the list is unpublished, you cannot verify coverage from the documentation — only in your own property, with the check described below.
• ChatGPT dominates AI referral traffic at 76.85% global share (April 2026 Statcounter data). Gemini holds 9.0%, Perplexity 7.73%, Copilot 3.76%, Claude 2.66%.
• AI-referred sessions now account for 0.9% of all web traffic, up 5x year-over-year, according to Similarweb's 2026 AI visibility analysis.
• The average AI-search visitor is 4.4x as valuable as the average traditional organic visit, measured on conversion rate across more than 500 digital marketing and SEO topics (Semrush, July 2025) — which is what makes accurate attribution essential for ROI measurement on earned media and content investment.
• 35 to 70% of AI referral sessions arrive without referrer headers and default to "Direct" in GA4. No referrer rule can recover those — native or custom — which is the real ceiling on referrer-based AI attribution and the reason UTM and server-log layers matter.
• Server logs see AI use that GA4 cannot. On authoritytech.io over 26 days in 2026, IP-verified ChatGPT-User requests outnumbered page loads referred by chatgpt.com by several hundred to one. Count user-triggered AI fetchers by verified IP address beside your GA4 sessions.

Why GA4 Misclassifies AI Search Traffic

GA4's default configuration categorizes most AI referral visits as direct or unassigned traffic. When a visitor clicks a link inside ChatGPT, Perplexity, or Gemini, the referrer header is often stripped by the AI platform's embedded browser, mobile app, or link-handling behavior. GA4 sees no referrer and classifies the session as "direct." The same bucket as someone typing your URL manually.

This creates a measurement blind spot at scale. Semrush's July 2025 AI search study found the average AI-search visitor 4.4 times as valuable as the average visit from traditional organic search, measured on conversion rate across more than 500 digital marketing and SEO topics.

Three specific failure modes cause GA4 to misclassify AI traffic:

• Stripped referrer headers. ChatGPT's mobile app and Perplexity's in-app browser frequently omit referrer data when opening external links. The visit registers as direct.
• Unrecognized source domains. GA4's default channel definitions carried no AI channel at all until 13 May 2026. Even now, Google classifies only referrers on a list it does not publish, so any AI source it has not added — and you cannot tell which those are without checking your own reports — lands in generic "Referral."
• UTM parameter inconsistency. ChatGPT appends utm_source=chatgpt.com to some outbound links, but Perplexity, Gemini, and Claude do not consistently use UTM parameters. Relying on UTMs alone misses the majority of AI traffic.

The net result: your GA4 reports undercount AI search traffic, often by more than half. Gartner predicted that traditional search engine volume would drop 25% by 2026 as AI chatbots absorb query volume. That shift is happening. AI-referred sessions now account for 0.9% of all web traffic, up 5x year-over-year. Most marketing teams still cannot measure the traffic this shift generates.

AI Referral Traffic Market Share in 2026

ChatGPT's share of AI referral traffic dropped to 76.85% in April 2026, its lowest level on record. Gemini holds second place at 9.0%, Perplexity third at 7.73%. These shifts matter for attribution setup because each platform handles referrer headers differently, and GA4's native channel covers only the referrers on Google's unpublished list.

According to Statcounter's global AI chatbot referral data for April 2026:

AI Platform	Referral Share (April 2026)	Change from April 2025
ChatGPT	76.85%	Down from 84.21%
Google Gemini	9.00%	Up from 2.31%
Perplexity	7.73%	Down from 12.07%
Microsoft Copilot	3.76%	Stable
Claude	2.66%	Up from ~1%
DeepSeek	<1%	New entrant

Two dynamics matter for attribution setup. First, Gemini's surge from 2.31% to 9.0% year-over-year means Gemini referrer patterns now warrant dedicated tracking. Second, Claude's growth to 2.66% (with weekly peaks above 3% in mid-March) makes Anthropic's platform a material traffic source for B2B content.

For B2B-specific referrals, the distribution skews differently. ChatGPT's share drops to roughly 63% while Claude reaches approximately 18% of measurable B2B AI referrals, according to Goodie's 2026 AI Search Traffic Report. B2B buyers use Claude and Perplexity for longer research queries at higher rates than the general population. Similarweb's 2026 AI visibility analysis confirms that AI-assisted sessions are growing at over 80% year-over-year, even as they still represent a single-digit share of total web traffic.

GA4's Native AI Assistant Channel

As of May 2026, Google Analytics automatically assigns an "ai-assistant" medium value and groups recognized AI chatbot traffic under a new "AI Assistant" default channel. The channel reached broad availability across GA4 properties on June 7, 2026. No configuration is required.

Here is the detail most guides get wrong: Google does not publish the list of referrers the AI Assistant channel recognizes. Google's Default Channel Group documentation describes the channel as covering "sources like ChatGPT, Gemini, Deepseek, Copilot, or Grok" and states the rule as: the medium is set to ai-assistant "if the referrer matches a list of AI Assistants." That list is nowhere in the documentation. The five names are illustrative, and treating them as a roster is the error that produced the widely repeated claim that Perplexity is excluded.

Two of Google's own release notes cut against that claim directly. The 13 May 2026 launch note describes the channel as covering "chatbots like ChatGPT, Gemini, and Claude" — Claude, which does not appear on the help page's example list. The 11 June 2026 Source Group note says the update "includes built-in grouping for emerging traffic sources like ChatGPT (OpenAI) and Perplexity."

So the honest position is: nobody outside Google can confirm or deny a given platform's inclusion from the public documentation, and any guide that gives you a definitive roster is reading examples as a specification. Check it in your own property instead. Open Reports > Acquisition > Traffic acquisition, set the dimension to Session source / medium, and look at how sessions from perplexity.ai, claude.ai or any other AI source are actually labelled. If the medium reads ai-assistant, Google recognizes that referrer for your property today. If it reads referral, it does not. That check takes a minute, survives Google changing the list without announcement, and is the only answer that is true for you.

What the native channel does:

• Automatic classification. Sessions from recognized AI chatbot referrers are tagged with medium ai-assistant and grouped into the "AI Assistant" channel. No configuration required.
• Forward-only application. The channel applies to new data from the rollout date forward. Historical data before May 2026 remains classified under old channel assignments.
• Zero setup. Every GA4 property gets the channel automatically. It parallels Google's 2022 addition of "cross-network" as a default channel for Performance Max traffic.

What the native channel does not do:

• No coverage for stripped referrers. Traffic arriving without a referrer header, which accounts for 35 to 70% of AI referral sessions depending on the platform and month, still lands in Direct.
• No published roster. Google does not tell you which referrers it recognizes, so coverage for any specific platform is unverifiable from the documentation and can change without an announcement.
• No historical reclassification. Sessions prior to May 2026 stay wherever GA4 originally bucketed them.
• No Google AI Overviews. Clicks from Google's own AI Overviews and AI Mode count as Organic Search, not AI Assistant. Google states this on both sides: the AI Assistant channel "excludes Google's AI Overviews and AI Mode," and Organic Search explicitly includes them. There is no native way to separate an AI Overview click from a blue-link click in GA4.

The native channel is a useful baseline, but it is not sufficient for complete AI traffic attribution. You still need a custom channel group — for an explicit roster you control rather than one you cannot read, and to handle the 35 to 70% of sessions with stripped referrers that no referrer rule can classify.

How to Create a Custom AI Search Channel Group in GA4

A custom channel group with a regex filter gives you an AI channel whose membership you can read, rather than one governed by a list Google does not publish. It also catches AI sources Google has not added yet. This setup takes under 10 minutes and works alongside the native AI Assistant channel.

Step-by-step setup:

1. Open channel settings. Go to Admin > Data display > Channel groups in your GA4 property.
2. Create a new channel group. Click "Create new channel group" and name it something distinct from GA4's native channel. "AI Search (Custom)" works well to differentiate from the built-in "AI Assistant" channel.
3. Add a new channel. Click "Add new channel" and name it "AI Search."
4. Set the rule. Choose "Source" matches regex, then enter the regex pattern from the next section.
5. Position the channel above Referral. GA4 evaluates channel rules top-down. If your AI channel sits below Referral, AI visits with valid referrers get classified as generic referrals before your AI rule is checked.
6. Save and verify. Apply the channel group to your reports. New sessions will begin classifying under the custom channel immediately.

You can run both the native AI Assistant channel and your custom channel group simultaneously. The custom group gives you explicit control over which platforms are tracked and lets you include AI sources before Google adds them to its recognized list.

AI Platform Referrer Patterns and Regex Filters for 2026

Each AI platform uses different referrer domains, and the pattern set has expanded significantly since 2025. A production-ready regex filter must cover the current major platforms plus recent entrants like DeepSeek and Grok.

Current referrer patterns by platform:

The last column records what Google has said in public, not what its classifier does — because Google publishes no roster, only examples. "Named by Google" means the platform appears in Google's own help page or release notes; "not named" means it does not, which is not the same as excluded. Verify your own property with the Traffic acquisition check above.

AI Platform	Referrer Domains	UTM Behavior	Named by Google
ChatGPT	chatgpt.com, chat.openai.com	Appends utm_source=chatgpt.com on some links	Yes — help page and both release notes
Perplexity	perplexity.ai	No consistent UTM parameters	Yes — 11 June 2026 Source Group note
Google Gemini	gemini.google.com, bard.google.com	No UTM parameters	Yes — help page and launch note
Claude	claude.ai	No UTM parameters	Yes — 13 May 2026 launch note
Microsoft Copilot	copilot.microsoft.com, bing.com/chat	Inconsistent UTM usage	Yes — help page
DeepSeek	deepseek.com, chat.deepseek.com	No UTM parameters	Yes — help page
Grok	grok.com, x.com/grok	No UTM parameters	Yes — help page
Meta AI	meta.ai	No UTM parameters	Not named
You.com	you.com	No UTM parameters	Not named

Full regex pattern for GA4 source matching:

chatgpt\.com|chat\.openai\.com|openai\.com|perplexity\.ai|claude\.ai|gemini\.google\.com|bard\.google\.com|copilot\.microsoft\.com|bing\.com/chat|deepseek\.com|grok\.com|meta\.ai|you\.com

This pattern covers the platforms responsible for over 99% of measurable AI referral traffic as of mid-2026. Review and update quarterly as new platforms emerge.

How to Track ChatGPT Traffic in GA4

ChatGPT is the easiest AI platform to track because it appends UTM parameters to some outbound links and uses consistent referrer domains. ChatGPT holds 76.85% of all AI referral traffic according to Statcounter. Accurate ChatGPT attribution is the single highest-leverage fix for AI traffic measurement.

ChatGPT sends traffic from two primary domains:

• chatgpt.com (the consumer product)
• chat.openai.com (the original domain, still active)

Additionally, ChatGPT appends utm_source=chatgpt.com to many outbound links. This means you can track ChatGPT traffic through three independent signals:

1. Source domain matching. The regex pattern chatgpt\.com|chat\.openai\.com captures both referrer domains.
2. UTM parameter. Filter for utm_source=chatgpt.com in GA4 explorations to isolate ChatGPT sessions even when the referrer is stripped.
3. GA4 native channel. ChatGPT is named by Google in both the help page and the launch note, so recognition is as well established as the public record allows. Confirm in Traffic acquisition that your ChatGPT sessions carry the ai-assistant medium.

For the most accurate ChatGPT measurement, create a GA4 exploration that combines all three signals: source domain regex OR utm_source match. This captures sessions that any single method would miss.

ChatGPT also uses OAI-SearchBot as its web crawler for search features. Check your server logs for OAI-SearchBot user-agent strings to see which pages ChatGPT's search index is crawling. This is a leading indicator of which content may generate future AI referral traffic. The requests ChatGPT makes while answering a user arrive under a different agent, ChatGPT-User, covered in the server-log section below.

How to Check Perplexity Referral Traffic in GA4

Perplexity is worth checking by hand, because its status in GA4's native channel is the one most widely misreported. Google named Perplexity in its 11 June 2026 Source Group release note, but it does not appear on the AI Assistant help page's example list, and Google publishes no roster to settle it. Look at your own Traffic acquisition report: if perplexity.ai sessions carry the ai-assistant medium, they are in the native channel; if they carry referral, they are not, and a custom channel group is the only way to see them alongside your other AI traffic.

Perplexity sends traffic from perplexity.ai. The web interface passes referrer headers reliably, but the mobile app strips them frequently. Perplexity's global referral share dropped from 12.07% to 7.73% year-over-year, but it remains the primary AI research tool for long-form queries. Perplexity users spend more time on-site than ChatGPT referrals on average, making them a higher-value audience for B2B content.

To check Perplexity referral traffic in GA4:

1. Custom channel group (recommended). Add perplexity\.ai to your custom AI channel regex. This is the only way to see Perplexity alongside other AI platforms in a unified channel view.
2. Exploration filter. In any GA4 exploration, filter Session source contains "perplexity" to isolate Perplexity sessions.
3. Traffic acquisition report. Go to Reports > Acquisition > Traffic acquisition, then search for "perplexity" in the source/medium column. Perplexity sessions appear as perplexity.ai / referral unless you have a custom channel group reclassifying them.

Perplexity also crawls your site with PerplexityBot. Checking your server logs for PerplexityBot hits shows which pages Perplexity indexes and may cite in answers.

How to Track Gemini and Claude Traffic in GA4

Google Gemini sends traffic from gemini.google.com and the legacy bard.google.com. Gemini's referral share surged from 2.31% to 9.0% in the year ending April 2026, making it the second-largest AI referral source globally. Gemini is named by Google in both the AI Assistant help page and the launch note, so tracking should be automatic for sessions with intact referrers. Validate by checking your GA4 Realtime report while clicking through from a Gemini response.

One caveat: Google AI Overviews and AI Mode clicks do not register as AI Assistant traffic. Those sessions count as Organic Search because Google treats AI features in Search as part of the search product. There is no native way to separate "clicked from an AI Overview" from "clicked from a blue link" in GA4.

Claude sends traffic from claude.ai. Claude's global referral share is 2.66%, but in B2B contexts it reaches approximately 18% of measurable AI referrals according to Goodie's 2026 AI Search Traffic Report. Claude does not append UTM parameters. Google named Claude in its 13 May 2026 launch note for the AI Assistant channel, even though Claude is absent from the help page's example list — a good illustration of why that list should not be read as a roster. Add claude\.ai to your custom regex channel group regardless, so your AI channel does not depend on an unpublished list staying the same.

For each platform, validate your tracking by opening the AI tool, searching for a query that cites your content, clicking through to your site, and checking GA4's Realtime report to confirm the session is classified correctly.

How to See the AI Use GA4 Never Records: Server Logs

GA4 records the visit; your server or CDN logs record the AI engine reading the page. Every major AI vendor fetches pages under named user agents, and each documents separate jobs for them. Only one of those jobs happens because a person asked a question.

Vendor	User-triggered fetcher (a person asked)	Search indexer	Training crawler
OpenAI	ChatGPT-User	OAI-SearchBot	GPTBot
Anthropic	Claude-User	Claude-SearchBot	ClaudeBot
Perplexity	Perplexity-User	PerplexityBot	none listed

OpenAI's crawler documentation says that when users ask ChatGPT a question, it may visit a web page with a ChatGPT-User agent, and that ChatGPT-User is not used for crawling the web in an automatic fashion. Perplexity and Anthropic describe Perplexity-User and Claude-User the same way. A user-triggered fetch is the nearest thing to an AI visit that GA4 never sees: the engine read your page to answer someone, whether or not that person clicked.

To count them:

1. Pull requests by user agent from your CDN or server logs. In Cloudflare, the GraphQL dataset httpRequestsAdaptiveGroups accepts a userAgent_like filter; on your own servers, search access logs for the names above. Keep successful (200) responses on real content paths.
2. Verify identity by IP address, not by name. A user-agent string costs nothing to fake. OpenAI publishes its ranges at openai.com/chatgpt-user.json, openai.com/searchbot.json and openai.com/gptbot.json, Anthropic at claude.com/crawling/bots.json, and Perplexity lists its two files on its crawler page. Count only requests from the published ranges.
3. Report verified fetches per page beside AI-referred sessions per page. Together they show which pages engines use, and which of those uses ever turn into a click.

What one site's logs showed

Here is the same check run on authoritytech.io's own logs for the 26 days from 19 September to 14 October 2026:

User agent	Requests from the vendor's published ranges	Requests using the name from other addresses
ChatGPT-User	7,014	563
Claude-User	2,805	2,979
Perplexity-User	0	8
OAI-SearchBot	5,331	234
PerplexityBot	8,978	650

Three readings matter for attribution:

• The fetch count dwarfs the click count. For every page load Cloudflare Web Analytics attributed to a chatgpt.com referrer in the same window, ChatGPT-User made several hundred verified requests. That ratio is not a click-through rate: one answer can fetch several pages, and fetches include automated prompt monitoring, including monitoring of this site's own AI visibility. Even if nine in ten fetches were monitoring, ChatGPT would still be reading the site dozens of times for every click analytics could record.
• Verify before you count. A name-only count would have roughly doubled Claude-User, because about half of the requests using that name came from addresses outside the ranges Anthropic publishes.
• Perplexity reads from its index. Perplexity-User made 8 requests in 26 days, all from addresses outside Perplexity's published ranges, while PerplexityBot fetched pages 8,978 times. Perplexity's use of this site appears to come from its index rather than from fetches at question time, so for Perplexity the useful log signal is which pages PerplexityBot crawls, and the perplexity.ai referral count in GA4 remains the per-click measure.

Method note. Requests come from Cloudflare HTTP request analytics for authoritytech.io, 19 September to 14 October 2026 (UTC), status 200 only, with user agents matched by name and checked against each vendor's published IP list as fetched on the collection day. Referred page loads come from Cloudflare Web Analytics, which samples page loads and counts only browsers that run its script, so the referral side is an estimate and a floor. One site and one window: read the ratio as an order of magnitude, not a benchmark.

How to Build an AI Traffic Dashboard in GA4

A dedicated AI traffic exploration in GA4 lets you monitor AI referral trends, compare platform performance, and identify which content AI engines cite most. This dashboard takes five minutes to build and becomes the operational layer on top of your channel group setup.

Create a free-form exploration in GA4 with these parameters:

Dimensions:

• Session source (to identify which AI platform sent the traffic)
• Session medium (to confirm ai-assistant vs. referral classification)
• Landing page (to see which content AI engines are citing)
• Date (for trend analysis)

Metrics:

• Sessions
• Engaged sessions
• Engagement rate
• Key events (conversions)
• Average engagement time per session

Filters:

• Session source matches regex: chatgpt|openai|perplexity|gemini|bard|claude|copilot|deepseek|grok|meta\.ai
• OR Session medium equals ai-assistant

This combined filter catches traffic from both your custom channel group and GA4's native AI Assistant channel, giving you a unified view of all AI traffic regardless of how GA4 classified it.

Add a comparison period (previous 28 days vs. current 28 days) to track whether your AI traffic is growing or declining after content changes. Pages that appear frequently in this report are the ones AI engines trust and cite. These are your highest-value assets for AI visibility optimization.

What AI Traffic Data Reveals About Content Strategy

AI traffic attribution exposes which content AI engines trust enough to cite, and which content is invisible to AI discovery despite ranking in traditional search. This data directly informs both content optimization and earned media investment.

Once your AI traffic attribution is live, three patterns typically emerge:

Pattern 1: High AI traffic, moderate organic traffic. These pages are being cited in AI-generated responses. They tend to have clear answer-first structure, specific claims with named sources, and extractable data (tables, lists, definitions). Double down on these pages. They are your citation architecture foundation.

Pattern 2: High organic traffic, zero AI traffic. These pages rank in traditional search but AI engines are not citing them. Common causes: vague headings without query-specific language, narrative-heavy prose without extractable claims, or missing structured data (FAQ schema, comparison tables). According to the Princeton GEO study (Aggarwal et al., ACM KDD 2024), adding statistics, quotations, and authoritative citations to source content boosted visibility by up to 40% in the paper's GEO-bench evaluation. That figure was measured on a research benchmark of generative engine responses, not on live ChatGPT, Perplexity, or Gemini traffic, and it does not establish a citation gain, referral gain, or revenue outcome for any specific page. These are your highest-leverage optimization targets for GEO, a distribution tactic within Machine Relations.

Pattern 3: Zero AI traffic, zero organic traffic. These pages are not visible anywhere. Before investing in optimization, verify the page has a genuine query target and source-backed claims worth citing. If not, the page may need a fundamental premise upgrade rather than technical fixes.

The gap between Pattern 1 and Pattern 2 pages is often structural, not topical. A page ranking position 5 for a competitive query but receiving zero AI traffic is almost always missing extractable claim blocks, comparison data, or FAQ schema. Those are the elements AI engines parse for citation.

How Earned Media Enters AI Search Referral Paths

AI engines cite third-party publications, not just brand websites, when generating answers. Muck Rack's May 2026 Generative Pulse classified more than 25 million cited links from ChatGPT, Claude, and Gemini responses across 17 industries. It placed about 84% under a broad earned-media category, with journalism at 27%; paid and advertorial content together accounted for 0.3%. Read that category carefully before planning against it: Muck Rack's earned-media bucket aggregates journalism with third-party corporate blogs, academic and research sources, government and NGO sites, encyclopedic aggregators such as Wikipedia, and social or user-generated content. Named business and technology outlets — Forbes, TechCrunch, The Wall Street Journal, Reuters, and industry publications — fall inside the 27% journalism share, which is one component of the broader 84%. The 84% figure is not a measure of Tier 1 press coverage.

The AI search referral path works in sequence:

1. Your brand secures an earned media placement in a publication that AI engines index and trust.
2. The placement mentions your brand with context, claims, and links.
3. An AI engine (ChatGPT, Perplexity, Gemini) retrieves that placement when answering a relevant query.
4. The AI engine synthesizes the placement's content and may cite it with a link.
5. The user clicks through to your site from the AI-generated response.
6. GA4 records the visit as AI referral traffic (if properly attributed).

Without AI traffic attribution, step 6 is invisible. You cannot measure which earned media placements generate AI search traffic, which publications AI engines trust most for your category, or whether your earned media investment is producing measurable AI visibility returns.

AI traffic attribution closes this measurement loop. When you can trace a visit from ChatGPT back to a specific landing page, and cross-reference that page with your earned media calendar, you know exactly which placements drive AI search results. This is the earned media ROI measurement that traditional PR metrics (impressions, reach, AVE) never provided.

AuthorityTech guarantees Tier 1 placements in publications AI engines cite (Forbes, TechCrunch, The Wall Street Journal) or you pay nothing. Start with a free AI visibility audit to measure your current citation footprint across ChatGPT, Perplexity, and Gemini.

How GEO, AEO, and SEO Fit Within Machine Relations

These disciplines are not competing alternatives. They represent different layers of the same system. Machine Relations is the full architecture that contains each of them:

Discipline	Optimizes for	Success condition	Scope
SEO	Ranking algorithms	Top 10 position on SERP	Technical + content
GEO	Generative AI engines	Cited in AI-generated answers	Content formatting + distribution
AEO	Answer boxes / featured snippets	Selected as the direct answer	Structured content
Digital PR	Human journalists/editors	Media placement	Outreach + storytelling
Machine Relations	AI-mediated discovery systems	Resolved and cited across AI engines	Full system: authority, entity, citation, distribution, measurement

AI traffic attribution is a Layer 5 (Measurement) function in the Machine Relations stack. Without it, the entire system operates without feedback. You cannot tell which authority signals, entity associations, citation architectures, or distribution tactics actually produce AI search visibility. Machine Relations research catalogues what independent studies have measured about AI source selection; attribution is what lets you test those findings against your own pages instead of assuming they transfer.

Frequently Asked Questions

Does GA4 now track AI traffic automatically?

Partially. Since 13 May 2026, GA4 automatically classifies traffic from recognized AI assistants under a native "AI Assistant" channel, setting the medium to ai-assistant. Google does not publish which referrers it recognizes: the help page lists ChatGPT, Gemini, DeepSeek, Copilot and Grok as examples, the launch note separately names Claude, and the June 2026 Source Group note names Perplexity. Treat any published roster, including the one this guide used to carry, as unverified, and check your own Traffic acquisition report instead. The native channel also only works for sessions with intact referrer headers, and between 35% and 70% of AI referral sessions arrive without one and land in "Direct." For coverage you control, run a custom channel group alongside the native channel.

Can I attribute traffic from ChatGPT or Perplexity in GA4?

Yes. ChatGPT traffic is the easiest to attribute because it uses consistent referrer domains (chatgpt.com, chat.openai.com) and appends utm_source=chatgpt.com to some outbound links. GA4's native channel also recognizes it automatically. Perplexity is harder to pin down, because Google named it in a June 2026 release note but does not publish the AI Assistant channel's referrer list. Check whether your perplexity.ai sessions carry the ai-assistant medium, and add perplexity\.ai to your custom channel group's source regex either way.

How do I check organic traffic from Perplexity in GA4?

Go to Reports > Acquisition > Traffic acquisition and search "perplexity" in the source/medium column. Sessions show as perplexity.ai / referral unless a custom channel group reclassifies them, or perplexity.ai / ai-assistant if GA4's native channel already recognizes the referrer for your property. Perplexity's web interface passes referrer headers reliably; its mobile app strips them more often, so a share of Perplexity visits will always land in Direct regardless of setup. Add perplexity\.ai to a custom channel regex to see all Perplexity sessions in one place.

How do I gauge how much traffic Perplexity sends me?

Combine two counts, because neither alone is complete: sessions where perplexity.ai already shows in Traffic acquisition (with or without the ai-assistant medium), plus a rough estimate of stripped-referrer visits from Perplexity's mobile app, which GA4 cannot attribute to any source. The first number is exact and available today; the second is not measurable from GA4 alone. In server logs, count PerplexityBot and Perplexity-User requests from the IP ranges Perplexity publishes. On authoritytech.io, Perplexity-User was nearly absent while PerplexityBot crawled heavily, so per-page crawl activity is the more useful log signal for Perplexity.

What is ChatGPT-User in my server logs?

ChatGPT-User is the agent OpenAI uses when ChatGPT visits a web page because a user asked a question, and for GPT Actions. It does not crawl automatically and does not decide what appears in ChatGPT search; OAI-SearchBot does that. Verified ChatGPT-User requests on a page mean ChatGPT read that page while answering someone, whether or not anyone clicked, which makes them the closest log-level measure of ChatGPT using your content. Check the request's IP address against openai.com/chatgpt-user.json before counting it.

What regex pattern should I use to track all AI search traffic in GA4?

Use this pattern for GA4 source matching: chatgpt\.com|chat\.openai\.com|openai\.com|perplexity\.ai|claude\.ai|gemini\.google\.com|bard\.google\.com|copilot\.microsoft\.com|bing\.com/chat|deepseek\.com|grok\.com|meta\.ai|you\.com. This covers the platforms responsible for over 99% of measurable AI referral traffic as of mid-2026. Review quarterly as the AI search market evolves.

Why does ChatGPT traffic show as direct in GA4?

ChatGPT's mobile app and some desktop interactions strip the referrer header when opening external links. Without a referrer, GA4 defaults to classifying the session as "direct." ChatGPT does append utm_source=chatgpt.com to some outbound links, which helps, but this parameter is not present on all click-throughs. A custom channel group using both source regex and UTM parameter matching captures the broadest set of ChatGPT sessions.

How do I track Gemini traffic in GA4?

Google Gemini traffic is automatically classified under GA4's native AI Assistant channel when the referrer is intact: Google names Gemini on both its help page and its launch note. Go to Reports > Acquisition > Traffic acquisition, then filter Session source for gemini.google.com or the legacy bard.google.com, or look under the AI Assistant channel directly. Add gemini\.google\.com|bard\.google\.com to a custom channel regex as a second, independent check. Clicks from Google AI Overviews and AI Mode count as Organic Search, not AI Assistant, and GA4 has no native way to separate the two.

How much does AI search traffic convert compared to organic search?

Semrush's July 2025 study put the average AI-search visitor at 4.4 times as valuable as the average visit from traditional organic search, measured on conversion rate across more than 500 digital marketing and SEO topics. That is a value multiple in one vertical, not a universal conversion rate. The likely reason it runs higher: AI search users have already asked a specific question, received a synthesized answer citing your brand or content, and clicked through with higher intent and trust than a typical SERP visitor.

Where do GEO and AEO fit inside Machine Relations?

GEO (Generative Engine Optimization) and AEO (Answer Engine Optimization) are distribution tactics within Layer 4 of the Machine Relations stack. They optimize how content is formatted and delivered for AI extraction and citation. Machine Relations contains GEO and AEO but extends beyond them to include authority building (Layer 1), entity clarity (Layer 2), citation architecture (Layer 3), and measurement (Layer 5), including the AI traffic attribution covered in this guide.

Updated 2026-10-15: added the server-log layer, with user-triggered AI fetchers verified by IP address and 26 days of authoritytech.io data, and an answer on what ChatGPT-User means in logs.