What is Dark traffic (AI referral)?
AI-sourced visits that analytics mislabels as “direct.”
Dark traffic is real referral traffic that arrives with no referrer, so analytics files it under “direct.” Most AI chat apps open links this way, which means a large share of the visits AI sends you are invisible in GA4, making the citations you earn look like they did nothing.
CiteTrack recovers these sessions server-side using a first-party cookie plus referrer/UTM and known-AI-host detection, then labels them correctly, see AI referral attribution and the explainer on why AI traffic shows as direct.
Why — and when — does it matter?
Dark traffic prevents reliable channel reporting when referral information is stripped. Recovery must use transparent rules and remain probabilistic where proof is unavailable.
How it works
Capture known referrers and campaign parameters first.
Preserve first-party source data during the session.
Identify plausible stripped-referrer visits with conservative rules.
Separate confirmed from inferred AI traffic.
Attribute conversions once and disclose methodology.
Worked example
A ChatGPT mobile-app click arrives as direct to a long comparison URL. A first-party handoff marker confirms the source; without it, classifying all direct visits to that page as AI would overstate the channel.
How to track and report it
Reporting rule: State the engine, prompt set, market, time window and denominator. A number without scope cannot be reproduced or compared.
Practical playbook
- Capture known referrers and campaign parameters first.
- Preserve first-party source data during the session.
- Identify plausible stripped-referrer visits with conservative rules.
- Separate confirmed from inferred AI traffic.
- Attribute conversions once and disclose methodology.
Common mistakes
- Relabeling all direct traffic
- Combining confirmed and inferred traffic
- Ignoring privacy and consent requirements
Questions people ask
What is Dark traffic (AI referral)?
AI-sourced visits that analytics mislabels as “direct.” Dark traffic prevents reliable channel reporting when referral information is stripped. Recovery must use transparent rules and remain probabilistic where proof is unavailable.
How does Dark traffic (AI referral) work?
Capture known referrers and campaign parameters first. Preserve first-party source data during the session. Identify plausible stripped-referrer visits with conservative rules. Separate confirmed from inferred AI traffic. Attribute conversions once and disclose methodology.
How do you measure Dark traffic (AI referral)?
Confirmed and inferred recovered sessions. Conversion rate, revenue and confidence tier. Recovery rate = recovered attributable sessions ÷ previously unattributed candidate sessions × 100
What is a common mistake with Dark traffic (AI referral)?
Relabeling all direct traffic. The scope and methodology should be stated before interpreting the result.
Hidden AI Revenue Calculator
Use the free tool to turn this reference into a concrete check on your own site.
Open the tool →