Measuring AI Visibility
GEO without measurement is faith-based marketing. The measurement stack is younger and messier than classic SEO's, but a workable system has three layers: referral traffic, citation presence, and share of voice.
Layer 1: AI referral traffic#
Visitors clicking through from AI answers arrive with identifiable referrers. Segment them in your analytics:
chatgpt.com ChatGPT (links in answers)
perplexity.ai Perplexity citations
copilot.microsoft.com Microsoft Copilot
gemini.google.com Gemini
claude.ai Claude (browsing/citations)In GA4, build a custom channel group or exploration segment matching these sources. Track alongside volume:
- Engagement quality - AI-referred users typically show high engagement and conversion (they arrive pre-recommended); this is your business case
- Landing pages - which content earns AI clicks tells you what's being cited
- Trend - share of total traffic, monthly; this number only grows
Layer 2: Citation presence (prompt tracking)#
The AI equivalent of rank tracking: ask the engines your important questions, record whether you're cited or named.
The manual baseline - build a prompt portfolio:
- Take your top 20–50 commercial and informational queries from your keyword map
- Rephrase each as a user would ask a chatbot ("what's the best X for Y?", "how do I…?")
- Monthly, run each prompt across ChatGPT (with search), Perplexity, Gemini, and Google AI Mode
- Record: were you cited (linked)? mentioned (named, unlinked)? absent? Who was?
Purpose-built tools automate this at scale - the category (Profound, Peec, Otterly, Semrush/Ahrefs AI features, and a fast-changing field of others) runs prompt panels across engines and trends your presence. Evaluate current options when you buy; the vendor list shifts quarterly.
Layer 3: Share of voice and brand signals#
Zoom out from prompts to presence:
- Competitive share of voice: across your prompt portfolio, what % of answers cite you vs. each competitor? This is the GEO equivalent of average position - and the number executives understand.
- Brand search volume (GSC: queries containing your brand) - when AI answers name you, people search for you; this is the lagging indicator of mention-without-click value
- AI crawler activity in server logs -
GPTBot,OAI-SearchBot,PerplexityBot,ClaudeBothits confirm you're being read, and which pages:
grep -iE "gptbot|oai-searchbot|perplexitybot|claudebot" access.log \
| awk '{print $7}' | sort | uniq -c | sort -rn | head -20A monthly GEO report that works#
1. AI referral sessions + conversions (GA4 segment, trend)
2. Citation rate across prompt portfolio (per engine, trend)
3. Share of voice vs. top 3 competitors (prompt panel)
4. Top cited pages (from referrals + spot checks)
5. Brand query volume (GSC, trend)
6. Actions: passages to retrofit, gaps where
competitors are cited and you're notThe last line matters most: every competitor-cited-you-absent prompt is a concrete work item - usually a quotability retrofit or a missing page.
GEO module complete. The final module, Expert Practice, turns everything so far into professional workflows - starting with the full technical audit.
