Skip to content
WeProms Digital
Menu

GA4 Setup to Track AI Chatbot Traffic (ChatGPT, Gemini, Claude)

Track ChatGPT, Gemini and Claude visits in GA4 with channel groups, historical reports and practical attribution guidance for Pakistani businesses.

AI Chatbot Traffic Share for Top Pakistani Websites (2026) (%)
OLX Pakistan OLX Pakistan: 0.1% 0.1% Daraz Daraz: 0.08% 0.08% Typical SME Typical SME: 0.02% 0.02% High-authority blog High-authority blog: 0.35% 0.35% News portal News portal: 0.15% 0.15%
  • OLX Pakistan: 0.1% — ~8,000 ChatGPT visits/month
  • Daraz: 0.08% — Estimated from Semrush data
  • Typical SME: 0.02% — Most see negligible AI traffic
  • High-authority blog: 0.35% — Education/finance niche
  • News portal: 0.15% — Dawn, Express Tribune range

If you run an ecommerce store in Lahore or manage a service business in Karachi, your GA4 dashboard probably shows Organic Search, Direct, Social and Paid Search. Visitors who discover you through ChatGPT, Gemini, Claude or Perplexity deserve attention too. Their visits can carry an identifiable source, but some arrive without enough information to tell you where they came from.

Google Analytics now includes AI Assistant as a channel in its default channel group. Google’s channel definitions explain that recognized assistant referrers receive the medium ai-assistant and campaign (ai-assistant). Google AI Mode and AI Overviews are excluded from this channel.

That distinction matters when you judge your content marketing. Flat organic traffic does not tell you whether chatbot referrals are contributing useful visits. Equally, a rise in Direct traffic does not prove that AI recommendations are working. You need identifiable sources and reliable purchase or enquiry tracking before drawing conclusions.

This walkthrough helps you check the default channel, organize identifiable referrals and build a report you can use in monthly reviews. It cannot reveal the origin of visits whose source information was never recorded.

First — Check If GA4 Already Shows AI Assistant Traffic

Open Reports > Acquisition > Traffic acquisition. Select Session default channel group and look for AI Assistant in the selected date range. Check that a comparison or report filter is not hiding relevant traffic.

The absence of a row does not mean your property is too old or needs the channel enabled. Google maintains the default channel definitions across properties. There may simply be no qualifying traffic in your current report.

Switch the dimension to Session source / medium and look for recognizable assistant domains. This helps you distinguish visits already classified as AI Assistant from older or differently classified referrals.

If you are unsure whether the underlying installation is reliable, start with a GA4 audit for AI traffic tracking. A channel report is only useful when the site is collecting the visits and business actions you intend to measure.

What this tells you: The default channel provides a starting point. Use a custom group if you need your own source definitions or a consistent view of identifiable historical referrals.

Then — Set Up a Custom Channel Group for AI Chatbot Sources

A custom group is optional; it does not activate the default AI Assistant channel. You need Editor or Administrator access to create or edit one.

Go to Admin > Data display > Channel groups. Choose Create new channel group, name the group “AI Assistant — Custom”, and add or edit a channel for the assistant sources you want to include. Avoid leaving overlapping AI channels in the same group without deciding which should take priority.

Use these domains as a starting point, checking them against your recorded sources:

Source Pattern Matches
chatgpt.com ChatGPT referrals
chat.openai.com Older ChatGPT referrals
gemini.google.com Google Gemini referrals
claude.ai Claude referrals
perplexity.ai Perplexity referrals
copilot.microsoft.com Microsoft Copilot referrals

Match the Source dimension using exactly matches, joining the domain conditions with OR. Add observed subdomains separately if needed. Exact matching makes the rule easier to inspect than a broad condition that captures unrelated domains containing similar text.

Place your channel above Referral and any other channel whose rules overlap. GA4 assigns traffic to the first matching channel in the group. Save the channel and group, then select the corresponding session custom channel group dimension in Traffic acquisition.

Custom channel groups can apply retroactively. They regroup available source data without rewriting the underlying records. If a historical visit has no identifiable assistant source, adding a rule cannot recover it from Direct.

Keep a short note of the domains and conditions you include. That makes later reporting changes easier to explain when a new assistant appears or your team changes the grouping.

Next — Build an Exploration Report to See Historical AI Traffic

Explorations let you inspect recorded assistant sources in more detail. Open Explore > Blank and import Session source, Session medium and Landing page + query string as dimensions. Import Sessions, Total users, Engagement rate and Key events as metrics.

Use Key events for important business actions in this report. Google’s terminology guidance distinguishes these actions from conversions used for advertising measurement and optimization.

Add Session source and Session medium to Rows, and your selected metrics to Values. Filter Session source using matches regex with ^(chatgpt\.com|chat\.openai\.com|gemini\.google\.com|claude\.ai|perplexity\.ai|copilot\.microsoft\.com)$. This combines the domains with OR logic. Add other verified source values if your report contains them.

Select a date range within your property’s available retention window. Data retention settings limit historical exploration data; extending retention does not restore records already deleted. Standard aggregated reports are not subject to the same retention limit.

You now have a view of identifiable assistant visits, rather than every visit influenced by AI. Check whether the key events represent meaningful outcomes, such as submitted enquiries or completed purchases. A channel label cannot compensate for missing or duplicated events.

Save the exploration with a descriptive name and record its source filter. When someone else reviews it, they should be able to see what is included and why its totals may differ from the default channel.

Infographic: GA4 AI chatbot traffic setup process showing five steps from channel group creation to monthly reporting dashboard

After That — Connect AI Traffic Data to Your SEO Strategy

Once you can see identifiable assistant visits, ask which pages receive them and what visitors do next. Add Landing page + query string to the exploration’s rows and sort by sessions.

These are entry pages for recorded visits. They are not a complete list of pages cited in AI answers, because a citation without a click does not generate a website session.

There are three patterns to investigate:

  1. Informational blog posts receiving referrals. Review the questions these pages answer and whether visitors continue to relevant products or services. Use successful pages as ideas for further testing, rather than assuming their format guarantees citations.

  2. Service pages with no identifiable AI visits. Check whether their services, location and supporting evidence are clearly explained. Missing referrals alone do not establish that a page has never appeared in an AI answer.

  3. Product pages receiving visits but few purchases. Inspect availability, pricing clarity, delivery information and the checkout journey. Where possible, compare the page with the recommendation that brought the visitor there. Avoid assuming all assistant visitors share the same purchase intent.

Ahrefs’ guide to AI Chatbot Traffic: What It Is, and How to Get More provides further context on referral tracking and content opportunities. Keep its broader findings separate from conclusions about your own Pakistani audience.

At This Point — Add AI Traffic to Your Monthly Reporting

Give identifiable assistant traffic its own line in your dashboard alongside your established channels. Label the view clearly: a default AI Assistant report and a custom domain-based report may include different visits.

Create a dedicated GA4 report or Looker Studio dashboard with these metrics:

Metric Why It Matters
AI Assistant sessions Tracks identifiable assistant visits
Top AI-referred landing pages Shows where recorded visitors enter
Key events from AI traffic Connects visits with business actions
Purchase revenue in PKR Shows recorded ecommerce value
AI sessions as a share of total Puts the channel in context

Compare the same definitions and metrics with the previous month. Include the absolute session and outcome counts so a small change in traffic does not look more significant than it is. Note any changes to source rules or event collection alongside the report.

Before presenting revenue, confirm that purchases and currency are recorded correctly. For a service business, keep enquiries separate from confirmed sales unless your reporting connects them reliably.

WeProms Digital’s GA4 setup and custom configuration service can help make these reports part of your regular performance review.

Infographic: GA4 assistant traffic reporting comparison; identifiable referrals can be grouped, but missing sources cannot be recovered

From Here — What This Produces

A carefully configured report gives you practical benefits, with clear limits:

  1. Clearer attribution for identifiable visits. You can separate recorded assistant sources from broader referral reporting. Visits without usable source information may still appear as Direct; the grouping does not reconstruct their origin.

  2. Content strategy signals. You learn which landing pages attract assistant-referred visitors and which lead to useful actions. Use that evidence to guide content improvements without treating referral volume as a complete measure of citation visibility.

  3. A starting point for competitive research. Your own GA4 property does not reveal competitors’ citations. Combine referral findings with separate checks of relevant AI answers. The CITE framework for Pakistani businesses provides a structured approach to that work.

Consider a Pakistani fashion store publishing detailed fabric guides. If visitors arrive from identifiable ChatGPT links, the report can show which guides they enter through and whether purchases follow. The store can then improve product links and buying information on those pages. It should not assume every resulting order is additional revenue caused by AI, or that unclassified orders came from chatbots.

Read next: AI search visibility tracking for Pakistani brands and server-side tracking setup.

If your source reporting or event collection needs attention, WeProms Digital can help scope an audit and configuration project. Bring the questions you want the dashboard to answer, alongside your existing property and checkout or enquiry flow. Reach out at hello@weproms.com or message on WhatsApp +92 300 0133399.

Frequently Asked Questions

How do I know if ChatGPT sends traffic to my website?

Check Traffic acquisition using Session source / medium for chatgpt.com and chat.openai.com, alongside the default AI Assistant channel. Use an exploration to inspect available historical records. This shows identifiable visits, not everyone who encountered your business in ChatGPT. Some visits, particularly from apps, may lack a usable referrer.

Does GA4 track Google AI Mode traffic automatically?

GA4 can record visits reaching your tagged website, but Google explicitly excludes AI Mode and AI Overviews from the AI Assistant channel. Do not use that channel’s total as a measure of all Google AI search traffic. An older property does not need a special activation setting for the default AI Assistant channel.

How much AI chatbot traffic do Pakistani websites get?

There is no reliable Pakistan-wide benchmark established by the sources here. Volume will depend on the site, audience, subject matter and identifiable referrals. Measure your own sessions and business outcomes over consistent reporting periods. A small recorded share can still merit attention, but it does not reveal visits whose source information is missing.

Can I see which of my pages ChatGPT and Gemini cite?

GA4 shows landing pages for recorded visits from those sources. It cannot show citations that receive no clicks or establish how often your brand appears across all answers. For a broader view, combine analytics with repeatable manual checks or a dedicated visibility monitoring service. Keep observed citations and referred visits as separate measures.

What does a GA4 AI traffic setup cost in Pakistan?

Ask for a quote based on your existing installation, event tracking, checkout integrations and reporting requirements. A channel grouping task has a different scope from repairing purchase tracking or building a dashboard. Confirm whether the work includes implementation, validation and documentation, and request a handover showing how to maintain the source rules.

Sources & References

  1. Ahrefs — AI Chatbot Traffic: What It Is, and How to Get More
  2. GoodFirms — SEO Statistics: AI Search, Rankings, Zero-Click Trends
  3. Semrush — OLX Pakistan Traffic Overview
  4. Ahrefs — AI Agents for SEO: What They Are and How They Work

Additional reading from industry feeds:

  • Search Engine Journal — Google Search Central Blog: New Resource for Optimizing Generative AI
  • Search Engine Journal — Google’s New AI Search Guide Calls AEO and GEO ‘Still SEO’
  • MarTech Series — AIEthos Debuts GEO Platform to Measure Brand Visibility in ChatGPT, Claude, and Gemini
  • Search Engine Roundtable — Google-Agent: The Web’s New Visitor Just Got An Identity

Watch WeProms

Watch on YouTube