Glossary · Beginner
What Is AI Referral Traffic? Definition and Example
AI referral traffic is visits to your website that arrive from AI assistants and answer engines, such as ChatGPT, Perplexity, Gemini, Claude or Copilot, usually when a user clicks a cited link. It's the most direct traffic signal of AI visibility, but it undercounts AI's real influence, because many answers don't lead to a click.
On this page
- How do you track it?
- Why is it undercounted?
- How should you use it?
- From MentionX data: how many answers can send a click
- Worked example: Ledgerline's AI channel
- What should the channel rule match?
- Common mistakes
- Common questions
- Related
Checked on 31 October 2026. "Our view" marks MentionX opinion. Ledgerline is a fictional company used for illustration.
How do you track it?
In GA4, AI referrals usually appear under the default Referral channel, mixed with other sites. Google's documentation explains how to create custom channel groups, including an example of an AI assistants channel defined by a source regex, placed above Referral. Custom channel groups apply to historical data too.
| Step | Detail |
|---|---|
| 1 | Admin → Data display → Channel groups |
| 2 | Copy the default group and add a channel such as "AI assistants" |
| 3 | Add a source-matching rule for the AI domains you see |
| 4 | Move the channel above Referral |
| 5 | Review it in Traffic acquisition reports |
See the full walkthrough in GA4 custom channel group for AI referrals.
Why is it undercounted?
- Some apps and links don't pass referrer information, so visits show as Direct.
- Many users read the answer and never click.
- Users may search for your brand later, which shows as organic or direct.
How should you use it?
Treat it as one signal among several:
| Signal | What it shows |
|---|---|
| AI referral traffic | Clicks from AI answers |
| Mention / recommendation rate | Visibility in answers |
| Self-reported attribution | Buyers who say AI influenced them |
| Key events from AI referrals | Business value of those visits |
From MentionX data: how many answers can send a click
Method. Read-only aggregate queries over MentionX scans from 22 August to 29 October 2026 (UTC): 2,217 engine runs (2,154 completed) across 18 anonymised projects and six engines.
A click needs a link. How often engines gave one varied widely:
| Engine | Brand mentioned | Answers with no citations |
|---|---|---|
| Gemini | 47.3% | 50.7% |
| OpenAI | 42.9% | 30.2% |
| Perplexity | 31.5% | 4.1% |
| AI Mode | 31.3% | 12.6% |
| Copilot | 26.0% | 47.0% |
| DeepSeek | 24.0% | 99.0% |
Where links exist, they don't always point straight to you. 69% of Gemini citation URLs in our data were Google redirect links, and OpenAI's links usually carried a UTM tag. OpenAI says ChatGPT adds utm_source=chatgpt.com to links (OpenAI); the API links we logged mostly used utm_source=openai instead.
Our view: a GA4 rule that only matches referrer domains will miss some of this. Match on source as well, and include both OpenAI tags.
Limits. A convenience sample of mostly B2B software projects, run through engine APIs, which can behave differently from the consumer apps. Perplexity, Copilot and DeepSeek figures come from a single day (26 October 2026), so treat them as a snapshot, not a trend.
Worked example: Ledgerline's AI channel
Ledgerline is a fictional accounts-payable tool. Figures are illustrative.
Ledgerline's GA4 shows a few dozen sessions a month from chatgpt.com and perplexity.ai, all filed under Referral. The team adds an "AI assistants" channel above Referral with a source regex covering chatgpt.com, openai, perplexity, gemini.google.com and copilot.
The new channel shows more sessions than the old referral count, because visits tagged utm_source=openai had been landing in other channels. It's still small next to organic search, but demo requests per session are above the site average, so the team reports both numbers.
What should the channel rule match?
Match the source, not just the referrer, and keep the list in one place so it's easy to update:
| Signal | Example | Why |
|---|---|---|
| Referrer domain | chatgpt.com, perplexity.ai | Classic referral visits |
| UTM source | chatgpt.com, openai | Tagged links from OpenAI |
| Landing pages | Pages you know are cited | Sanity check on the numbers |
Custom channel groups apply to historical data too, so you can see the trend from before you set them up (Google).
Common mistakes
- Reading a low AI referral count as proof that AI answers don't matter to your buyers.
- Matching only referrer domains, so tagged visits fall into other channels.
- Comparing months after changing the channel rule without noting the change.
- Reporting sessions without conversions; small volumes can still carry valuable visits.
Common questions
Is AI referral traffic growing for everyone? We don't have cross-site traffic data, so we can't say. Track your own trend monthly.
Do Google AI Overviews show up as AI referrals? No. Google counts clicks from AI Overviews and AI Mode as Google organic traffic in the Web search type (Google).
Should I report AI referrals as a KPI? Yes, alongside visibility metrics. Alone it misses buyers who never click.
Related
Questions
How do I see AI referral traffic in GA4?
Create a custom channel group with a rule matching AI assistant referrer domains, placed above Referral so those visits are classified correctly.
Does GA4 capture all AI-driven visits?
No. Some visits arrive without referrer information and appear as Direct, so AI's influence is usually undercounted.
Is AI referral traffic a good GEO KPI?
It's useful but incomplete. Many AI answers satisfy the buyer without a click, so pair it with visibility metrics and self-reported attribution.