Prompt
I'm trying to figure out whether AI answer visibility is actually driving leads, not just impressions. How do I attribute that?
Latest observation
To attribute whether AI answer visibility is driving leads rather than just impressions, you need to connect AI exposure → site visit → conversion as cleanly as possible. The hard part is that most AI surfaces don’t give perfect click-level tracking, so you usually need a blended attribution setup.
1) Define what you want to measure
Start with three separate metrics:
- AI visibility: whether your brand/content is being cited, summarized, or mentioned in AI answers
- AI-assisted traffic: visits that appear to come from AI surfaces
- AI-assisted leads: form fills, demo requests, calls, signups, or purchases that happened after AI exposure
If you only track impressions, you’re measuring awareness. To measure leads, you need conversion attribution.
2) Track referrals from AI surfaces
Set up analytics to identify traffic coming from AI tools and AI-powered answer experiences.
In analytics, look for referrers like:
chat.openai.comperplexity.aicopilot.microsoft.comgemini.google.comclaude.aiopenai.comsearch.google.com/ AI Overviews-related traffic, where available- other AI browser assistants or chat surfaces
Also create a segment for:
- Direct traffic after AI exposure
- Branded search after AI exposure
- Referral traffic from AI domains
This won’t catch everything, but it’s the start.
3) Use landing-page and campaign tagging where possible
If you control the links in AI-friendly content, use:
- UTM parameters
- dedicated landing pages
- vanity URLs
- unique offer codes
Example:
utm_source=perplexityutm_medium=ai_answerutm_campaign=product_guide
This helps when AI surfaces actually send users to your site. For organic citations in AI answers, you may not control the link, so this is only partial coverage.
4) Measure assisted conversions, not just last-click
AI often influences users early in the journey. A lead may:
- see your brand in an AI answer,
- later search your brand,
- then convert directly.
If you only use last-click attribution, AI gets undercounted.
Use:
- first-touch attribution
- multi-touch attribution
- conversion path analysis
- assisted conversions
Look at:
- AI referral before conversion
- branded search within a short window after AI exposure
- direct conversions from users who previously landed from AI referrals
5) Build a holdout or lift test
This is the best way to estimate causality.
Possible test designs:
- Geo holdout: optimize AI visibility in certain regions and compare lead volume vs. control regions
- Topic holdout: improve AI visibility for certain content clusters, leave others unchanged
- Time-based test: compare lead volume before/after a major AI optimization effort, adjusting for seasonality
If leads rise in the test group more than the control group, that suggests AI visibility is driving incremental leads.
6) Use lead source surveys
Ask leads how they found you.
Add a field like:
- “What prompted you to contact us?”
- “Where did you first hear about us?”
- “Did you see us in an AI tool or AI search result?”
Provide options such as:
- Google search
- ChatGPT
- Perplexity
- Google AI Overviews
- Referral
- Other
This is noisy, but useful as a supporting signal.
7) Compare branded search lift after AI visibility gains
If AI answer visibility is working, you often see:
- more branded search volume
- more direct traffic
- more conversions on branded queries
Track:
- branded search impressions
- branded search clicks
- branded search conversion rate
- lift after earning AI citations/mentions
This is especially helpful when the AI answer doesn’t click through directly but creates demand.
8) Attribute leads with a practical model
A workable attribution model for AI visibility usually combines:
A. Direct tracking
- AI referral traffic
- UTMs
- landing-page conversions
B. Assisted attribution
- AI referral appears anywhere in the path
- branded search shortly after AI exposure
- direct conversion after prior AI visit
C. Incrementality
- control vs. exposed comparison
- lift in leads after AI optimization
If you want a simple rule:
Count a lead as AI-assisted if the user had an AI referral or AI-cited exposure within a defined lookback window, even if they converted later via another channel.
A common lookback window is:
- 7 days for fast decisions
- 30 days for longer B2B cycles
9) Important caveats
AI attribution is messy because:
- many AI answers don’t send a referrer
- users may copy info without clicking
- AI traffic can be misclassified as direct
- multi-device journeys hide the source
- some AI tools block normal tracking
So don’t expect perfect attribution. Aim for:
- directional accuracy
- incrementality
- multi-signal validation
10) Recommended setup
If you want a practical stack:
- GA4 or similar analytics
- CRM with lead source fields
- server-side event tracking
- custom AI referrer channel grouping
- conversion path analysis
- survey question on lead forms
- periodic holdout testing
Bottom line
To attribute AI answer visibility to leads, you need to go beyond impressions and measure:
- AI referrals
- assisted conversions
- branded search lift
- incremental lift via testing
- lead survey confirmation
If you want, I can help you design:
- a GA4 attribution setup
- a lead source schema for CRM
- or a simple incrementality test plan for AI visibility.