Measuring the Invisible: From Share of Voice to Share of Model
Traditional marketing measurement is in crisis. Tools like Google Analytics 4 (GA4) cannot track the "Zero-click" interactions occurring within AI search engines (Perplexity, ChatGPT). Even when users receive brand recommendations and make purchase decisions within an AI chat interface, the "Dark Funnel" phenomenon—where no website visit record (Referral) is left behind—is accelerating. Marketers must now shift their KPIs from Share of Voice (SoV) to "Share of Model (SoM)," which measures how frequently and positively AI models mention your brand.
Why Is AI Traffic Invisible in GA4?
The primary reason is that the essence of AI search lies in "Direct Answers," not "Link Clicks." While traditional search engines (Google) acted as "intermediaries" sending users to websites, Generative AI acts as a "terminator," synthesizing information to provide answers directly.
Zero-Click Consumption: Users consume the information summarized by AI and leave without clicking further.
Referral Stripping: Platforms like ChatGPT often strip specific referral data or lump traffic under "Direct" for security and privacy reasons, even when external links are provided.
The Dark Funnel: Brand recommendations occurring in Slack communities, private Discords, and 1:1 conversations with AI are part of the "Dark Funnel"—an area tracking codes cannot reach.
As a result, a "Data Discrepancy" occurs where dashboard traffic figures stagnate or decline, even while a brand's actual influence is growing.
What Is Share of Model (SoM)?
Share of Model (SoM) refers to the frequency with which AI models include your brand in their answers to specific topics or questions. It is a redefinition of traditional Share of Voice for the AI era.
Measurement Target
SERP Rankings
Mentions within AI Answers
Goal
Top 3 on Page 1
Inclusion in "Recommendations" & "Citations"
Competition
Keyword Bidding & Content
Brand Authority within Training Data
Success Metrics
Traffic, CTR
Citations, Sentiment
"AI is not just a search tool; it is an advisor making purchase decisions." Therefore, increasing SoM is not merely about increasing exposure, but a branding effort to make AI recognize your brand as a "trusted solution."
How to Measure SoM: A 3-Step Audit Framework
While perfect automated tools do not yet exist, you can periodically measure and manage SoM through manual audits.
1. Define Personas & Prompts
Create a list of questions your customers are likely to ask. These should be "conversational questions," not simple keywords.
Informational: "What are the pros and cons of marketing automation tools?"
Commercial: "Recommend the top 3 CRM software for SMBs."
Transactional: "Compare the pricing plans of Company A vs. Company B."
2. Test Across Platforms
Input the prompts above directly into major AI engines to collect results.
Target Engines: ChatGPT (GPT-4), Perplexity, Google Gemini, Claude.
Tip: Use "Incognito Window" or a new chat session to exclude personalization effects.
3. Scoring
Quantify the answer results based on the following criteria:
Mentioned: Was the brand name mentioned? (Yes/No)
Cited: Was a source (link) included? (Yes/No)
Recommendation Rank: What was the ranking in the recommendation list?
Sentiment: Was the nuance Positive, Neutral, or Negative?
Proxy Metrics: Proving Performance Beyond the Dashboard
Along with SoM measurement, use "Proxy Metrics" to infer GEO performance from existing data.
Brand Search Lift
Users who discover a brand via AI often search the brand name on Google for verification. If total traffic decreases but "Brand Keyword" search volume increases, it is a signal of rising awareness.
Direct Traffic Spike
Visitors clicking links in AI answers or typing URLs directly are often captured as "Direct." If Direct traffic spikes without a specific marketing campaign, it may be the influence of AI.
Qualitative Attribution
Add a "How did you hear about us?" field to sign-up forms or sales calls, with options like "ChatGPT Recommendation" or "AI Search." This becomes the most definitive proof.
Conclusion
The saying "If you can't measure it, you can't manage it" is only half true. In the AI era, the most critical decisions happen in the unmeasurable "Dark Funnel."
Move away from the "Vanity Metric" of traffic and manage the real influence of Share of Model (SoM). Ensuring your brand is included in AI's "Recommendation List" will be the most powerful marketing asset of the coming future.
Frequently Asked Questions (FAQs)
Q1. Are there tools that automatically measure SoM? A. Tools like Peec AI, Profound, and BrandWell are emerging. They send bulk prompts to various LLMs to track brand mention frequency and sentiment. However, as they are in early stages, it is recommended to combine them with manual audits.
Q2. How do I persuade my boss when traffic is dropping? A. Emphasize that "while traffic has decreased, Conversion Rate and Lead Quality have improved." Users arriving via AI are likely "high-intent users" who have already learned about the product. Also, present the upward trend in "Brand Search Volume" as evidence.
Q3. How can I increase my SoM? A. You must ensure your brand is mentioned in "Authoritative Sources" that AI trusts. Distributing positive reviews and detailed information on industry news sites, Wikipedia, and review platforms (G2, Capterra) is key. (This will be covered in detail in Pillar 3.)
Q4. Do I need to care about all AI models? A. If resources are limited, focus on Perplexity and ChatGPT. Perplexity is a search-specialized engine most active in citations, while ChatGPT has the highest market share and impact.
Q5. Is the Dark Funnel truly impossible to track? A. Technically, perfect tracking is impossible. However, you can estimate it through "Self-reported Attribution (Customer Surveys)" and "Correlation Analysis." Analyzing the correlation between specific AI marketing activities and revenue growth is a realistic alternative.
References
Last updated