wandered

Patrick Kennedy writes about the evidence behind AI traffic.

Patrick authors Wandered’s field notes on AI referrals, search visibility, machine activity, attribution, and the business outcomes that follow. The work is designed to help analytics, growth, SEO, and product teams make claims their evidence can support.

Research focus

The central question is not whether AI matters. It is what a specific observation can prove. A sampled mention, a citation, an identifiable human referral, a crawler request, and a conversion are different events with different limits.

Each guide starts with that boundary, reviews current primary documentation, and turns the result into a practical measurement workflow. Provider behavior and analytics classifications are dated and rechecked rather than presented as permanent rules.

Start with the measurement problem.

AI traffic

AI traffic analytics: connect discovery to business outcomes

A complete measurement model for the traffic and outcomes created by AI discovery.

Read the field note →
AI traffic

How to track AI traffic in GA4—and what GA4 still misses

A step-by-step GA4 workflow for AI Assistant traffic, landing pages, and conversions.

Read the field note →
AI visibility

AI visibility vs. AI traffic: what each metric can prove

A clear boundary between appearing in an AI answer and receiving measurable visits.

Read the field note →
AI agents

AI crawler user-agent directory: purpose, policy, and verification

A purpose-first directory for classifying AI machine traffic without counting it as human activity.

Read the field note →
AI traffic

AI referral source directory: assistants and analytics classification

A citation-ready reference for normalizing observable AI referral traffic.

Read the field note →
AI traffic

AI traffic attribution without false precision

A defensible attribution ladder for AI-influenced customer journeys.

Read the field note →

Sources, dates, and limits remain visible.

Every substantive guide names its author, reviewer, publication date, verification date, methodology, primary sources, and material changes.

Read the editorial policy →