AI Referral Analytics

This skill makes answer-engine traffic visible in analytics. Citations only matter if they drive outcomes, but a large share of AI-assistant traffic arrives with no referrer — from mobile apps, copy-paste, and in-assistant browsers — and lands in "Direct," so most brands badly undercount it. This skill captures and attributes that traffic and reads its quality. The examples use GA4 patterns; the approach maps to any analytics platform. It works for any business, B2B or B2C.

When to use this skill

• Setting up tracking for AI-assistant referral traffic.
• Recovering AI traffic that is misattributed as Direct.
• Attributing conversions and revenue to AI sources.
• Reporting AI traffic volume and quality alongside visibility metrics.

Before you start

Read aeo-foundation.md for the site/analytics context and the conversions that matter. If it is missing, point to the aeo-foundation skill.

The core problem

AI-assistant visits often carry no referrer (native apps, copy-pasted links, assistant-embedded browsers), so analytics files them under Direct. Even referred visits may not be grouped as "AI" by default. Without deliberate setup, AI's contribution is invisible and undervalued.

Setup (GA4 patterns)

1. Use the AI channel grouping if available. Newer analytics builds include an AI-assistant channel that classifies referred traffic from major assistants — enable and confirm it.

2. Create a custom channel group / segment for AI sources, matching the referrer/source domains of the assistants, for example:

   chatgpt.com|chat.openai.com|openai.com|perplexity.ai|gemini.google.com|
   copilot.microsoft.com|claude.ai|you.com|grok.com|meta.ai|deepseek.com

   Match on source/referrer to pull these out of generic Direct/Referral buckets.

3. Tag links you control. Where you can influence a link the assistant shows (e.g. in your own content or feeds), add UTM parameters (utm_source, utm_medium=ai_assistant) for explicit attribution.
4. Track conversions by this segment. Attribute sign-ups, leads, or sales to the AI segment to measure outcome, not just sessions.
5. Accept a residual gap. Some referrer-less AI traffic will remain in Direct; watch Direct for unexplained shifts that correlate with AI visibility gains, and report ranges rather than false precision.

Reading the data

• Volume & trend — AI sessions over time, by source.
• Quality — conversion rate and engagement of AI traffic vs. other channels (often high-intent, since the user asked an assistant and clicked through).
• Source mix — which assistants actually send traffic for this brand.
• Tie-back — correlate referral gains with citation/SOV gains from ai-visibility-tracking.

Output format

• The analytics configuration: channel group/segment definition and any UTM scheme.
• A reporting view: AI sessions, conversions, and quality vs. other channels.
• A caveat note on the residual Direct misattribution and how to read it.
• (assumption) tags where platform specifics were unknown.

Common pitfalls

• Assuming Direct is "type-ins" when a chunk is unattributed AI traffic.
• Reporting AI sessions without conversions — volume without value.
• Over-claiming precision when referrer data is inherently incomplete.
• Setting it up once and not revisiting as assistants and referrer behavior change.

Related Skills

• ai-visibility-tracking — the citation/SOV side; pair both for the full picture.
• ai-visibility-audit — connects traffic outcomes back to readiness gaps.
• aeo-foundation — the conversions and goals worth attributing.