How do you actually track conversions from AI traffic? Attribution is a nightmare

AI attribution requires a multi-method approach. No single method captures everything.

The Attribution Stack:

Layer 1: Direct Referral Tracking What you CAN directly attribute in GA4:

• chatgpt.com referrals
• perplexity.ai referrals
• Other identifiable AI sources

Layer 2: Assisted Conversions Multi-touch attribution in GA4:

• AI touchpoint anywhere in journey
• First-touch vs last-touch analysis
• Weighted attribution models

Layer 3: Correlation Analysis Statistical relationship between:

• AI visibility score and total conversions
• AI traffic and brand search volume
• Visibility improvements and conversion lifts

Layer 4: Qualitative Data Direct customer input:

• “How did you find us?” form field
• Post-purchase surveys
• Sales qualification questions

Combined Approach: Each layer captures different AI influence. Together they tell the complete story.

Correlation analysis bridges the attribution gap.

The Logic: If AI visibility causes conversions, then:

• When visibility goes up → conversions go up
• When visibility goes down → conversions go down
• With some time lag (typically 4-8 weeks)

Analysis Method:

Step 1: Time series data Weekly data for 6+ months:

• AI visibility score
• Total conversions
• Brand search volume

Step 2: Lag correlation Calculate correlation at different lags:

• Visibility week 1 vs conversions week 1
• Visibility week 1 vs conversions week 2
• Visibility week 1 vs conversions week 4
• Etc.

Step 3: Identify strongest correlation Example findings:

• Visibility correlates 0.68 with conversions at 4-week lag
• Visibility correlates 0.72 with brand search at 2-week lag

Step 4: Regression model “10% visibility increase → 8% conversion increase at 4-week lag”

Presentation: Chart showing visibility line and conversion line. Visual correlation builds the case.

Correlation isn’t causation, but it’s compelling evidence.

The “How did you find us?” field is underrated.

Implementation:

Add open-text or dropdown to key forms:

• Lead forms
• Demo requests
• Contact forms
• Purchase flow

Question format: “How did you first hear about us?” (open text)

OR

Dropdown options:

• Google search
• AI assistant (ChatGPT, Perplexity, etc.)
• Social media
• Referral from colleague
• Other

What we see: Before AI focus: 2% said AI After AI focus: 12% said AI Growth matches visibility improvements

Data quality:

• Self-reported has recall bias
• But reveals discovery AI analytics miss
• Captures “I asked ChatGPT, then Googled you”

Our finding: 45% of customers who checked “AI” converted. Only 28% of “Google search” converted. AI-discovered leads are higher quality.

Simple field, powerful insight.

Sales team input reveals AI influence.

Qualification Questions: Train sales to ask: “Before we connected, how did you research solutions like ours?”

Common AI-related responses:

• “I asked ChatGPT for recommendations”
• “Perplexity suggested a few options”
• “AI gave me a starting list”

CRM Tracking: Create field: “Discovery Method” Options include: AI assistant

Weekly Report:

Insights: AI-sourced leads often:

• More educated on problem
• Faster sales cycle
• Higher average deal size

Qualitative data that analytics miss.

Brand search is a proxy for AI influence.

The Logic: When AI mentions your brand:

• Some users ask AI follow-up questions
• Some users Google your brand name
• Brand search increases as awareness increases

Measurement: Track in Google Search Console:

• Brand keyword impressions
• Brand keyword clicks
• Trend over time

Correlation: Chart AI visibility vs brand search volume. If they move together, AI is driving awareness.

Our data:

Brand search grew 95% as visibility grew 71%.

Why this matters: Brand search has high conversion intent. AI drives brand search. Therefore, AI drives high-intent traffic.

Indirect but powerful evidence.

AI leads often have different quality characteristics.

Quality Metrics to Track:

Engagement:

• Time on site (AI vs other)
• Pages per session
• Demo request rate
• Trial sign-up rate

Sales Metrics:

• SQL rate (from MQL)
• Win rate
• Sales cycle length
• Average deal size

Our Findings:

Story this tells: AI visitors are more educated. They’ve already researched via AI. They’re further along the journey. Higher quality, faster conversion.

Quality metrics often more compelling than volume.

What executives actually want to see:

Monthly AI ROI Report:

1. Investment

• GEO team time: X hours
• Tools: $Y
• Content: $Z
• Total: $ABC

2. Direct Attribution

• AI referral traffic: X sessions
• AI direct conversions: Y
• AI attributed revenue: $Z

3. Influenced Attribution

• AI assisted conversions: X
• Correlation-estimated impact: Y%
• Self-reported AI discovery: Z%

4. Quality Indicators

• AI conversion rate vs average
• AI lead quality vs average
• AI customer value vs average

5. ROI Calculation Conservative: Direct attribution only Moderate: Include assisted Optimistic: Include correlated lift

The narrative: “We invested $X in GEO. Directly trackable: $Y revenue. Likely influenced: $Z additional. Lead quality: 40% higher than average. Estimated ROI: ABC%”

Multiple methods, one clear story.

The reality: AI is often one touchpoint of many.

Common paths we see:

Path 1: AI → Google → Convert User asks AI → gets brand name → Googles → converts AI gets no direct attribution, but was catalyst.

Path 2: AI → Website → Retargeting → Convert AI mention → visits site → retargeted → converts Retargeting gets attribution.

Path 3: AI → Social → Website → Convert AI mention → follows on social → later visits → converts Social gets attribution.

How to capture: GA4 path exploration:

• Set up to show all touchpoints
• Look for AI in any position
• Credit AI for influence

Our finding: AI in conversion path: 22% of conversions AI as last touch: only 8% of conversions

AI influence is 3x what direct attribution shows.

Look at paths, not just last touch.

Practical implementation steps:

Week 1: GA4 Setup

• Create AI referral segment
• Set up conversion tracking
• Build basic dashboard

Week 2: Form Field

• Add “How did you find us?” to key forms
• Options include AI/ChatGPT
• Start collecting data

Week 3: Sales Enablement

• Train sales on discovery questions
• Add CRM field for source
• Begin tracking

Week 4: Analysis Framework

• Weekly AI traffic report
• Monthly correlation analysis
• Quarterly ROI calculation

Ongoing:

• Refine attribution model
• Improve data collection
• Build historical baseline

Start simple, iterate: Don’t try to build perfect attribution day one. Start with what you can track. Add layers over time.

Imperfect data > no data.

This gives me a practical framework. Implementation plan:

Attribution Stack:

1. Direct: GA4 AI referral segment
2. Assisted: Path exploration analysis
3. Correlation: Visibility vs conversion trends
4. Qualitative: Form fields + sales input

Dashboard Metrics:

• AI sessions and growth
• AI conversion rate
• AI assisted conversions
• Self-reported AI discovery %
• Lead quality comparison

Monthly Report Structure:

• Investment summary
• Direct attribution
• Influenced attribution
• Quality metrics
• ROI range (conservative to optimistic)

Key insight: Perfect attribution isn’t possible. Multi-method approach tells the full story. Quality metrics often more compelling than volume.

Thank you all - this makes AI ROI provable, not just hoped for.