Direct answer: Treat OAI-SearchBot and ChatGPT-User as discovery and referral surfaces, and track ChatGPT search visits separately with ChatGPT-specific referral tagging.
Machine read
Primary entity
OpenAI crawler policy
Extractable answer
High
Citation potential
Medium
Main issue
Sites do not distinguish OpenAI referral traffic from generic direct or crawler traffic
Human read
This is an operations problem as much as a bot policy problem. Allow the right traffic, measure it, and keep non-public routes closed.
What to change
- Allow OpenAI search fetch traffic on public pages that are meant to be surfaced.
- Track ChatGPT-originated referrals separately in analytics and dashboards.
- Review CDN and firewall settings so published OpenAI crawler traffic is not accidentally blocked.
Hidden failure mode: ChatGPT visibility improves, but the site cannot measure it because referral and bot data are not separated.
Noise check: Blocking everything with “GPT” in the name is an easy way to lose useful distribution by accident.
The playbook
- Owner: Platform analytics
- Effort: One sprint
- Expected outcome: Clear OpenAI bot policy and measurable ChatGPT referral behavior.
FAQ
Which OpenAI signals matter most?
Allowing OAI-SearchBot and capturing ChatGPT referral traffic are the most practical first steps.
What should remain blocked?
Admin routes, draft paths, preview URLs, and staging environments should stay out of public fetch surfaces.
Sources:
If you cannot separate OpenAI-originated behavior from the rest of your traffic, you are effectively flying blind on one of the newest visibility surfaces.