Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Repair Windows errors before they cause bigger problems3Fix the driver behind crashes, sound loss and screen glitchesiTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
To measure visits from ChatGPT, Gemini, Claude, Perplexity and similar assistants in GA4, start with Reports > Acquisition > Traffic acquisition and select the session source, medium, campaign or default-channel dimensions. GA4 now provides an AI Assistant default channel for recognized chatbot referrals. When native classification does not capture a source, create a custom channel group, tag links you control with UTMs, and evaluate key events and revenue rather than sessions alone. For reconciliation with orders or a CRM, use the Reporting API or BigQuery export.
What GA4 now counts as AI referral traffic
Google documents an AI Assistant default channel for users who arrive from recognized chatbot services. The channel includes sources such as ChatGPT, Gemini, Deepseek, Copilot and Grok. Google’s help page describes it as “the channel by which users arrive at your site from sources like ChatGPT, Gemini, Deepseek, Copilot, or Grok.”
Google documented this behavior in its May 13, 2026 release notes. Recognized visits are assigned the following values during collection and processing:
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →| GA4 field | Value for recognized AI-assistant referrals | What it means |
|---|---|---|
| Default channel | AI Assistant | The channel used in channel-group reports. |
| Medium | ai-assistant |
The medium value used to classify the visit. |
| Campaign | (ai-assistant) |
The campaign value shown for the recognized traffic. |
| Source | The referring assistant or source recorded by GA4 | The individual source used for source-level analysis. |
Google’s current default-channel documentation is at Google Analytics default channel group, and the dated change is listed in the Google Analytics release notes.
This classification does not mean every visit influenced by an AI system will appear in the channel. Google specifically excludes Google AI Overviews and AI Mode from the AI Assistant channel. A user who reads an answer, remembers your brand and later types your address, uses a bookmark, or searches separately has no assistant referrer for GA4 to classify.
Use the native AI Assistant channel first
Open the Traffic acquisition report
- Sign in to the relevant GA4 property.
- Open Reports > Acquisition > Traffic acquisition.
- Set the date range you want to evaluate.
- Use the table’s dimension selector to choose a session-scoped dimension such as Session default channel group, Session source, Session medium or Session campaign.
- Filter for AI Assistant, or locate the channel in the table and open its source and medium detail.
- Compare sessions with engaged sessions, engagement rate, key events, key-event rate and revenue for the same date range.
Traffic acquisition is session-oriented: its dimensions describe how the session began. Google documents the report and its session-scoped dimensions in Traffic acquisition reporting. Do not substitute a session source for event attribution; those are different scopes and answer different questions.
Break the channel down by assistant
After identifying the AI Assistant row, use the table’s secondary dimension or a detail view to inspect the session source, medium and campaign values. This lets you distinguish, for example, traffic attributed to ChatGPT from traffic attributed to Gemini or Copilot, provided the referrer is available and recognized. Keep the original source values in your analysis instead of treating every AI visit as one undifferentiated audience.
Check both volume and business outcomes
A large session count can be misleading if visitors do not engage or complete an outcome. For each date range, record at least:
- Sessions and engaged sessions.
- Engagement rate or another agreed quality measure.
- Key events, such as a signup, lead submission, purchase or subscription start.
- Revenue where ecommerce or subscription revenue is implemented.
Use the same date range and conversion definitions when comparing AI Assistant with organic search, referrals, email or direct traffic. This avoids calling a high-volume but low-value source successful.
Build a custom channel group when native classification is insufficient
Native classification depends on Google’s maintained recognition rules and on a usable referrer. If a source is missing, newly launched, or important to your reporting but not grouped as AI Assistant, Google documents a custom-channel fallback.
Create the channel
- Open Admin > Data display > Channel groups.
- Copy an existing channel group rather than changing the default group directly.
- Add a new channel named AI assistants (or another name your team will use consistently).
- Add conditions that match the relevant source hostnames or source values with a regular expression.
- Move the new channel above Referrals in the channel order. GA4 evaluates channel rules in order, so the AI rule must run before the broad referral rule.
- Save the group and use it in User acquisition and Traffic acquisition reports.
Google’s procedure is described in Custom channel groups. A practical starting pattern can include the hostnames you actually observe, for example (chatgpt.com|openai.com|gemini.google.com|copilot.microsoft.com|claude.ai|perplexity.ai). Treat that expression as an editable example: inspect your property’s recorded source values, follow Google’s current example, and add or remove hostnames when assistant URLs change.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Native channel versus custom channel group
| Decision point | AI Assistant default channel | Custom channel group |
|---|---|---|
| Coverage | Recognized sources on Google’s maintained list, when a referrer is available. | Only the sources and patterns you define, including sources not yet recognized natively. |
| Naming and rules | Google controls the channel name and classification logic. | You control the channel name, conditions, regular expressions and ordering. |
| Historical reporting | Uses GA4’s native classification for the data available to the property. | Do not assume a new group changes every historical report; record its creation date and verify an earlier date range in the property. |
| Reporting scope | Available through the standard channel-group dimensions and related source dimensions. | Available wherever the selected custom channel group is supported, including User acquisition and Traffic acquisition. |
| Maintenance | Google maintains recognition, although referrer changes can still affect coverage. | You must review observed hostnames and update the regular expression when assistant URLs change. |
| Reconciliation | Fast for routine dashboard comparisons. | Useful when your organization needs a controlled taxonomy that matches CRM or warehouse reporting. |
Keep one documented definition for the native channel and one for any custom group. Otherwise, changing a regular expression can make month-over-month results look like a traffic change when it is actually a classification change.
Tag AI links you control with UTMs
You cannot add a UTM to a link shown in an assistant if you do not control that link. You can, however, tag links in your own documentation, support responses, partner pages, newsletters or campaigns that may be copied into an AI conversation. GA4 maps UTM values to campaign and traffic-source dimensions when the tagged URL is collected.
Choose a naming convention
Use lowercase values, a fixed vocabulary and a written owner for the convention. For example:
| Parameter | Example | Purpose |
|---|---|---|
utm_source |
assistant |
Identifies the broad source category you control. |
utm_medium |
ai-assistant |
Separates the traffic from other referral or campaign media. |
utm_campaign |
product-guide-2026 |
Identifies the initiative or content series. |
utm_content |
setup-link |
Distinguishes links or placements within the same campaign. |
utm_id |
guide-042 |
Provides a stable internal identifier when your reporting system uses one. |
Google explains campaign-parameter collection in Campaign URL builder and manual tagging. A tagged URL should retain every parameter through redirects, tracking hops and shorteners. Test the final destination in a browser and confirm that the landing-page URL still contains the parameters before relying on the data.
Prevent parameters from being lost
- Configure redirects to pass the complete query string to the destination.
- Check that a link shortener does not remove or rewrite UTM parameters.
- Do not use a redirect that sends visitors through an untagged intermediate page unless the parameters are preserved.
- Keep cross-domain navigation and referral exclusions configured consistently when a journey moves between domains you own.
- Document whether a tagged link is intended to represent an assistant campaign, a support article or another controlled placement.
Google notes that visits without a clear source can appear as (direct) / (none); its explanation is available at About direct traffic.
Define conversions as GA4 key events
Decide what “conversion” means before comparing assistants. In GA4, mark the event representing the business outcome as a key event. Typical examples are a completed signup, qualified lead submission, purchase or subscription start. A page view or session is not automatically a business conversion.
Separate session acquisition from event attribution
Session-scoped dimensions answer: How did this session begin? They are the right lens for Traffic acquisition. Event-scoped traffic dimensions and the property’s attribution settings answer: Which source receives credit for this key event under the selected attribution method? Google describes these dimension scopes and attribution uses in Traffic and user acquisition reports and Campaign and traffic-source dimensions.
For example, a visitor might arrive from an AI assistant, return through an email link and then purchase. The session report can show how the first session began, while event-level attribution can assign purchase credit according to the property’s configured model. Do not add session conversions and event-attributed conversions together as if they were the same measure.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchPC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Use a consistent comparison table
| Question | Recommended GA4 view | Metric to compare |
|---|---|---|
| How many visits began from an assistant? | Traffic acquisition with Session default channel group or Session source | Sessions and engaged sessions |
| Did those sessions produce outcomes? | Traffic acquisition with key-event columns | Key events and key-event rate |
| Which source receives purchase or lead credit? | Event-scoped source dimensions and the property’s attribution reports | Key events and revenue under the selected attribution method |
| Does the dashboard agree with operational records? | Reporting API or BigQuery export joined to CRM or order data | User, session and event-level totals |
Why AI visits show as referral, direct or unassigned
AI attribution is observable only when GA4 receives enough information to classify the visit. A click can be genuinely influenced by an assistant and still lack a usable AI referrer.
Missing or stripped referrer
If the browser, assistant, privacy setting or intermediate page does not send a referrer, GA4 cannot apply the native AI Assistant rule. The visit may appear as direct, referral or unassigned instead. You cannot recover a source that was never transmitted; use controlled UTMs on links you own and monitor the proportion of direct traffic after major site or redirect changes.
Redirects and shorteners remove UTMs
A redirect that drops the query string removes the campaign information before the GA4 tag runs. Test the full click path and preserve every parameter. If the final landing page has no UTM values, fix the redirect rather than trying to repair the report afterward.
Consent, blockers and incomplete tagging
Consent choices, browser protections, ad blockers and an incorrectly deployed GA4 tag can prevent collection. Check whether the landing page sends a page-view event after consent, whether the measurement ID is the intended property, and whether the same consent behavior applies across domains.
Cross-domain and self-referral problems
When a journey moves between domains you operate, missing cross-domain configuration can create a new session or a self-referral. That can hide the original assistant source. Review the cross-domain implementation and referral exclusions, then test a complete assistant-to-site journey in a clean browser session.
Unrecognized or newly changed assistant hostnames
Native recognition depends on Google’s maintained list and the referrer value. If an assistant changes its hostname or introduces a new service, a custom channel rule can provide interim visibility. Keep the rule above Referrals and revise it when your observed source values change.
Validate totals with the Reporting API and BigQuery
The GA4 interface is convenient for routine analysis, but row-level reconciliation often requires an export. The Reporting API exposes first-user, session and event traffic-source dimensions. Its schema is documented at Google Analytics Data API schema.
BigQuery export provides traffic-attribution data at user, session and event levels, including session last-click manual-campaign fields. That lets an analyst compare the GA4 session classification with the campaign values collected at an event and then reconcile those records with order or CRM identifiers. The traffic-attribution export is described at BigQuery traffic attribution data.
Use exports for three specific checks
- Dashboard reconciliation: Pull the same date range and dimensions used in the GA4 report, then compare row counts and key-event totals.
- Session-to-event inspection: Trace a session’s acquisition source separately from the source attached to a key event, rather than assuming they are identical.
- Operational matching: Join event or transaction identifiers to CRM leads, subscriptions or orders and investigate differences caused by time zones, consent or deduplication.
Record the property time zone, date range, channel-group version and attribution settings with each extract. Without those controls, a later analyst may be unable to explain why an API or warehouse total differs from a screen in GA4.
A repeatable measurement workflow
- Write the measurement question. Decide whether you need assistant-started sessions, key-event attribution, revenue, or all three.
- Check native classification. In Traffic acquisition, inspect AI Assistant, session source, session medium and session campaign for the chosen period.
- Confirm the outcome event. Verify that signup, lead, purchase or subscription-start events are marked as key events and are firing once per intended action.
- Compare quality, not only volume. Review engaged sessions, key-event rate, key events and revenue alongside sessions.
- Add a custom channel only where needed. Copy a channel group, define the AI-assistant rule, place it above Referrals and document the regular expression.
- Tag links under your control. Apply the agreed UTM convention and test redirects, shorteners and final URLs.
- Investigate direct and referral anomalies. Check referrer availability, consent, blockers, cross-domain setup and parameter preservation before changing classifications.
- Reconcile important numbers. Use the Reporting API or BigQuery when GA4 totals must match CRM, subscription or order records.
- Review the taxonomy regularly. Assistant hostnames and URL structures can change; update custom rules and note the effective date.
What GA4 can and cannot prove
GA4 measures clicks and the key events it can attribute to those collected visits. It cannot prove every AI-influenced visit, because an assistant may not send a referrer, a user may return through another channel, a link may be shared offline, or consent and browser controls may prevent collection. Treat AI Assistant figures as measured referral traffic and attributed outcomes, not as a census of all influence from AI systems.
The most defensible report therefore labels its scope: native or custom classification, session or event attribution, date range, channel-group version and whether revenue came from GA4 or a reconciled operational system. That precision prevents a visible AI referral count from being mistaken for the total number of customers who encountered your content in an assistant.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
