Browser analytics is incomplete for AI traffic.
Many AI crawlers and server-side agents request pages, feeds, files, APIs or structured content without executing browser JavaScript. Analytics therefore prioritizes authorized first-party/server-side observation where possible: CDN and edge logs, server/reverse-proxy data, WAF/security logs, application events and other supported sources.
Public product copy does not invent connectors. Exact supported sources are documented for the customer environment, and Analytics does not require access to private AI conversations.
Classify machine activity without forcing uncertainty into a brand name.
Detection can combine user-agent strings, verified ranges, IP/ASN patterns, reverse DNS where appropriate, authenticated agent identity, request behavior and customer classifications. Identities can remain known crawler, retrieval agent, authenticated agent, likely AI automation, unknown automation or human browser.
Core machine metrics
Requests
Raw classified machine requests over time by agent, property, path, topic or content type.
Successful retrievals
Requests that return usable content under defined response criteria.
Content objects
Unique pages, files, endpoints or other objects accessed by machine identities.
Access outcomes
Allowed, blocked, rate-limited, challenged, redirected or failed requests and their policy match.
Definitions for AI Crawler Requests, Content Retrievals and related metrics are published in the public metrics standard.
Connect machine access to human referrals — without pretending correlation is causation.
Where attribution is possible, Analytics measures AI referral sessions using first-party referrers, campaign/link metadata, known redirect patterns and authorized platform data. Downstream views can include engaged sessions, conversions, revenue and customer acquisition metrics when connected.
| Layer | Examples |
|---|---|
| Machine | Crawler requests, agent identities, retrievals and access state. |
| Human | AI referral sessions, engagement and landing content. |
| Business | Leads, purchase events, conversion rate and revenue where connected. |
| Relationship | Machine requests vs referrals, retrieval concentration vs conversions; descriptive unless causal design supports more. |
See machine demand across content and properties.
Portfolio analytics can compare multiple domains, subdomains, brands, business units and regions while preserving their boundaries. Content-level views can retain URL/object identity, topic, requests, distinct agents, latest access, policy, referrals, conversions, revenue and related Intelligence citations.
- Find assets with high machine demand but little human traffic.
- Compare properties by machine activity, referral conversion and access state.
- Identify content concentrations by agent, topic or market.
- Connect external citation changes from Intelligence to first-party retrieval patterns without claiming automatic causality.
Measure whether actual access matches intended policy.
Analytics can expose mismatches such as robots rules that allow access while a WAF blocks requests, content intended to be blocked still reaching origin, important structured endpoints failing, or rate limits creating unexpected access errors. For an independent policy audit, hand the evidence to Verify.
First-party measurement with data minimization.
Analytics should collect only the data required for the defined measurement. Enterprise controls can include configurable retention, field minimization, hashing or pseudonymization where appropriate, role-based access, export controls, tenant isolation and audit logs.
Attribution types stay explicit: referrer attribution, campaign attribution, multi-touch attribution and incrementality answer different questions and should not be collapsed.
Setup journey
- Create the organization/property and verify authorized ownership or control.
- Connect a supported first-party data source.
- Classify machine and human activity and configure access-policy expectations if used.
- Connect conversion, CRM or commerce sources where desired.
- Validate events, identities and measurement dimensions.
- Begin reporting by agents, content, properties, referrals and access state.
Intelligence measures the external answer environment; Analytics measures first-party machine and human outcomes. Together they create a multi-layer picture without claiming each step is causal.
Connect the properties AI systems actually access.
Measure machine retrieval, access policy, AI referrals and downstream outcomes from authorized first-party evidence.