Skip to main content

RankSenseAi

How to Track ChatGPT, Perplexity and Gemini Traffic in GA4 (with a Looker Studio Template)

By Irfan Dar, SEO, AEO & GEO Strategist at RankSenseAI · Last updated October 10, 2026 · 12-minute read · Applies to GA4 and Looker Studio

Short answer You can track ChatGPT, Perplexity and Gemini traffic in GA4 in three layers. First, check the native AI Assistant channel in Reports → Acquisition → Traffic acquisition. Second, add a custom channel group using a regex on Session source to catch platforms and dates the native channel misses. Third, report on both in Looker Studio. Expect an undercount: copied links, mobile apps and some AI browsers pass no referrer and land in Direct. Google AI Overviews and AI Mode stay in Organic Search.

How does GA4 see traffic from ChatGPT, Perplexity and Gemini?

GA4 attributes a session using the referrer the browser sends when someone clicks a link, plus any UTM parameters. If an AI tool passes a referrer, GA4 can name the source. If it doesn't, the visit looks like Direct or Unassigned. That one mechanic explains almost every gap below.

PlatformTypical source in GA4What to expect
ChatGPTchatgpt.com (older: chat.openai.com)Clicks from cited links are usually identifiable. App, copied-URL and Atlas-browser visits often show as Direct or (not set).
Perplexityperplexity.aiCommonly appears as a referral. MarTech's testing found the Comet browser typically passes referrer data.
Geminigemini.google.comGemini app clicks can be identified. AI Overviews and AI Mode are not split out; they remain in Organic Search.
Claude, Copilot, othersclaude.ai, copilot.microsoft.comIdentifiable when a referrer is sent. Google hasn't published the full list of platforms in its native channel.

Before Google added a dedicated channel, much of this traffic was buried in Referral or Unassigned. Cardinal Path reported that roughly 40% of the LLM traffic it saw across client data was falling into Unassigned. For a refresher on why that happens, see our guide to fixing Unassigned traffic in GA4.

Step 1: Check the native "AI Assistant" channel in GA4

1Google added an AI Assistant channel to the Default Channel Group in May 2026, with no setup required. Recognized visits are labeled:

  • Session default channel group: AI Assistant
  • Medium: ai-assistant
  • Campaign: (ai-assistant)

To find it: open GA4 → Reports → Acquisition → Traffic acquisition → set the primary dimension to Session default channel group → look for the AI Assistant row. Add Session source as a secondary dimension to see ChatGPT, Gemini and others separately.

Read this before trusting the number. The channel isn't backfilled, so older data keeps its old classification. Cardinal Path observed it appearing from around June 4, 2026, even though Google's announcement is dated May 13, so check the first date it appears in your property. Google also hasn't published a full list of qualifying sources, and MarTech noted that coverage for Perplexity and Copilot was unconfirmed. Treat the channel as a floor, not a ceiling.

Step 2: Build a custom AI channel group (regex)

2A custom group gives you control, lets you classify data from before the native channel existed, and doesn't change if Google quietly edits its backend list. Related: how custom channel groups work in GA4.

  1. Go to Admin → Data display → Channel groups and click Create new channel group.
  2. Name it Default + AI. Click Add new channel and call it AI Traffic.
  3. Set the condition to Source → matches regex, and paste the pattern below.
  4. Drag the AI channel above Referral (and above Unassigned). Rules evaluate top to bottom, and the first match wins.
  5. Save. New reports can use it as a dimension. Optionally set it as your primary channel group.
Regex: all major AI platforms (Source matches regex).*(chatgpt|openai|perplexity|gemini\.google|bard\.google|claude\.ai|copilot\.microsoft|edgeservices|deepseek|grok\.com|meta\.ai|mistral|poe\.com|phind|you\.com|kagi).*

GA4's matches regex requires a full match, which is why the pattern starts and ends with .*. Several competing guides use patterns with broad terms like gpt or bard, which can over-match unrelated sources. Review your own Source list and tighten as needed. If you'd rather split traffic by behavior, one agency case study separated AI Assistants (ChatGPT, Claude, Gemini, Copilot) from AI Search (Perplexity, You.com, Kagi), placing the more specific rule first.

Per-platform patterns

PlatformRegex for Source
ChatGPT.*(chatgpt|openai).*
Perplexity.*perplexity.*
Gemini.*(gemini\.google|bard\.google).*
Claude.*claude\.ai.*
Microsoft Copilot.*(copilot\.microsoft|edgeservices).*

Step 3: Break it down by platform and landing page

3Use a GA4 Exploration so you can see which pages AI tools actually send people to. See also our walkthrough of GA4 Explorations.

  1. Go to Explore → Blank → Free form.
  2. Dimensions: Session source / medium, Session default channel group, Landing page + query string.
  3. Metrics: Sessions, Engaged sessions, Engagement rate, Average engagement time per session, Key events.
  4. Rows: Session source / medium. Add a filter: Session source matches regex using the pattern from Step 2, or Session default channel group exactly matches AI Assistant.
  5. Duplicate the tab and swap the row dimension to Landing page to see top AI-referred pages.

Quick alternative: in Traffic acquisition, click Add filter → Session source / medium → matches regex and paste the same pattern.

Step 4: Looker Studio template for AI traffic

4A dashboard keeps AI traffic visible to people who never open GA4. Here is the template structure, plus the fields to build it yourself. Pair it with our GA4 + Looker Studio dashboard guide.

Copy the Looker Studio template

Build it yourself in 10 minutes

  1. Create a new report in Looker Studio and add your Google Analytics data source (the GA4 property).
  2. Add a calculated field named AI Platform:
Calculated field: AI PlatformCASE WHEN REGEXP_MATCH(LOWER(Session source), ".*(chatgpt|openai).*") THEN "ChatGPT" WHEN REGEXP_MATCH(LOWER(Session source), ".*perplexity.*") THEN "Perplexity" WHEN REGEXP_MATCH(LOWER(Session source), ".*(gemini\\.google|bard\\.google).*") THEN "Gemini" WHEN REGEXP_MATCH(LOWER(Session source), ".*claude\\.ai.*") THEN "Claude" WHEN REGEXP_MATCH(LOWER(Session source), ".*(copilot\\.microsoft|edgeservices).*") THEN "Copilot" WHEN Session default channel group = "AI Assistant" THEN "Other AI (native channel)" ELSE "Not AI" END
  1. Add a page-level filter: AI Platform not equal to "Not AI".
  2. Build the pages below.
PageChartsQuestion it answers
1. OverviewScorecards (sessions, engagement rate, key events); time series by AI PlatformIs AI traffic growing?
2. PlatformsTable: AI Platform × sessions, engagement rate, avg. engagement time, key eventsWhich assistant sends the best visitors?
3. Landing pagesTable: Landing page × sessions, key events (sorted by sessions)Which content do AI tools cite?
4. BenchmarkComparison of AI vs Organic Search vs Direct (use Session default channel group)Does AI traffic behave differently?
5. Direct watchlistDirect + new users on informational pages (see Step 5), shown separatelyIs there hidden AI influence?

If you used a custom channel group in Step 2, it appears in Looker Studio as a dimension and can replace the calculated field for channel-level views.

Step 5: Estimate hidden AI traffic inside Direct

5Many ChatGPT users copy and paste links, and the ChatGPT app and AI browsers often strip the referrer. That visit looks like Direct. Backbone Media suggests a proxy: Direct + new users + landing on blog, FAQ or how-to pages, with high engagement.

  • Segment: Session source = (direct) AND New users AND Landing page contains your informational paths (for example /blog/).
  • Track it beside your AI channel, never added into it.
  • A rising trend after you gain AI visibility is a signal, not proof.
Optional controlled test: add UTMs to links you know AI tools surface, for example ?utm_source=chatgpt&utm_medium=referral&utm_campaign=llmtest. See our UTM parameters guide. Use it for experiments only, since inconsistent UTMs create messy reporting.

Step 6: Validate your setup

  • Cross-check classification. In an Exploration, show Session source against Session default channel group, and confirm each AI source lands in the channel you expect.
  • Compare native vs custom. If both exist, check overlap before reporting two different numbers.
  • Test live. Click a cited link from ChatGPT, Perplexity and Gemini, then confirm the session in Realtime.
  • Watch for bot-like behavior. Agentic AI browsers can behave like humans. Look for uniform session lengths, spikes with no engagement, or unexplained paid-click bursts.
  • Annotate dates. Note when you launched the channel group and when the native channel began appearing.

What you can't reliably track in GA4

  • Copy-and-paste visits: no referrer, so they appear as Direct.
  • Mobile apps and in-app browsers: operating systems and webviews often strip referrers.
  • AI browsers: Perplexity Comet tends to pass a referrer; ChatGPT Atlas often doesn't, according to MarTech's testing.
  • Google AI Overviews and AI Mode: credited to Organic Search, with no reliable split.
  • Answers with no link: an assistant can cite or paraphrase you with zero click and zero session. That's brand influence GA4 can't see.

What should you do with AI traffic data?

Compare quality, not just volume. Put AI sessions beside Organic Search for engagement rate, pages per session and key events. Cardinal Path says AI-referred traffic has repeatedly converted better than comparable referral traffic in its client work, but test this on your own data before making budget decisions.

Find the topics, not just the pages. Ranko Media suggests identifying which topics pull AI traffic, then using Google Search Console queries for those pages to see related searches, since AI Overview data is merged into Search Console. Then build more depth where you already have authority. For the strategy side, read our guides to generative engine optimization (GEO) and answer engine optimization (AEO), or our beginner's guide to AI SEO.

Report honestly. Show recognized AI traffic as a floor and the Direct watchlist as a separate directional indicator.

Common mistakes to avoid

  • Putting the AI channel below Referral, so it never matches.
  • Forgetting the leading and trailing .* in a full-match regex.
  • Comparing pre- and post-May 2026 AI traffic without noting the classification change.
  • Counting all Direct traffic as AI traffic.
  • Reading a small volume as unimportant. Check engagement and conversions first.

Frequently asked questions

Can GA4 track ChatGPT traffic?

Yes, when the visitor clicks a link and the referrer is passed. GA4 groups recognized visits into its AI Assistant channel. Visits from copied links, mobile apps and some AI browsers still appear as Direct.

Does GA4 have a channel for AI traffic?

Yes. In May 2026 Google added an AI Assistant channel to the Default Channel Group, with medium ai-assistant and campaign (ai-assistant). It isn't applied retroactively.

How do I track Perplexity traffic in GA4?

Look for sessions with source perplexity.ai in Traffic acquisition, or use the regex .*perplexity.* in a custom channel group. Coverage of Perplexity in Google's native channel isn't confirmed, so a custom group is the safer approach.

Can GA4 separate Google AI Overviews or AI Mode traffic from Organic Search?

Not reliably. Google excludes its own AI search features from the AI Assistant channel, so that traffic continues to be counted as Organic Search.

Why does some AI traffic show up as Direct?

Because the referrer was lost. Copied URLs, native apps, in-app browsers and some AI browsers don't pass it, so GA4 has nothing to attribute.

Is there a free Looker Studio template for AI traffic?

Yes, use the template linked in Step 4, or build one with the AI Platform calculated field above. It connects to your own GA4 property, so your data stays in your account.

Want this set up and validated on your site?
We handle GA4 and Google Tag Manager setup, dashboards and AI visibility reporting.
Talk to our team

Related guides

GA4 custom channel groups Fix Unassigned traffic in GA4 GA4 Looker Studio dashboard GEO guide Free AI Citation Checker

Sources and further reading

Irfan
Author: Irfan

Irfan Dar is an SEO professional with 4+ years of experience in SEO, performance marketing, and digital analytics. He specializes in technical SEO, content strategy, on-page optimization, keyword research, AEO (Answer Engine Optimization), and GEO (Generative Engine Optimization). At RankSenseAI, Irfan focuses on helping businesses improve their visibility across traditional search engines and emerging AI search platforms. His work covers technical SEO, content optimization, AI search visibility, structured data, entity optimization, and strategies designed to help brands get discovered across search experiences such as Google AI Overviews, ChatGPT, and other AI-powered platforms. He also works with tools and platforms including Google Search Console, GA4, Google Tag Manager, SEMrush, Screaming Frog, Google Ads, Meta Ads, and Looker Studio. Irfan believes effective SEO is not just about rankings. It is about creating useful, technically sound content and building the signals that help search engines and AI systems understand, trust, and surface a brand.