What Dark Traffic means
Causes include stripped referrers on HTTPS-to-HTTP transitions, links in messaging apps and email clients, and increasingly, AI assistants where users read an answer and navigate separately.
It's always existed. AI-assisted research has made it a much larger share of the total for many sites.
Why it matters for AI search
Dark traffic is where AI influence hides. A buyer researching in ChatGPT and arriving days later by typing your URL shows up as direct, and gets credited to nothing.
Rising direct traffic alongside falling non-branded clicks is a signature worth investigating rather than dismissing.
See it in action
Where 'direct' traffic actually comes from.
Two thirds of 'direct' wasn't direct.
None of this is visible in a standard report. Decomposing it takes self-reported source data and correlation with branded search — imprecise, but far better than treating it as unknowable.
How to get it right
Reducing and understanding dark traffic
- Tag every campaign properly so untagged traffic stops inflating direct
- Ensure your site is fully HTTPS to preserve referrer data where possible
- Add a self-reported source question at conversion — the only way to hear 'I asked ChatGPT'
- Correlate direct traffic trends with branded search and citation share
- Report direct as a composite rather than treating it as a single channel
Common questions
Is all direct traffic dark traffic?
No. Some is genuine type-ins and bookmarks. The rest is referral data that was lost, which is the part worth decomposing.
Can we track AI referrals directly?
Partially. Some platforms pass referrer data and appear in reports; many don't, and users often navigate separately rather than clicking a citation.
How do I estimate the AI share?
Self-reported source questions at conversion plus correlation with branded search lift. It's directional, and it's far better than assuming zero.
Related concepts
These come up alongside Dark Traffic constantly.