How to Measure AI Visibility: The Metrics That Actually Matter

Why Traditional SEO Metrics Don't Capture AI Visibility

The metrics that have defined digital marketing success for the past 20 years (keyword rankings, organic traffic, click-through rate, bounce rate) are optimized for a specific model of how users find information: they type a query, see a list of results, and click a link. That model is changing.

When a user asks ChatGPT a question about your industry, no click-through rate is recorded. When Perplexity synthesizes a response that cites your competitors but not you, no ranking drop is logged. When Google AI Overview appears above organic results and answers the query without requiring a click, your organic traffic dips without any obvious cause in your analytics.

Traditional SEO metrics are blind to this new layer of search. They measure what happens after a user reaches your site, not whether your brand appears in the growing share of queries that AI platforms handle before a user ever visits a search results page. For brands invested in GEO optimization, a new measurement framework is essential.

Share of Model (SoM): Definition, How to Calculate, Benchmarks

Share of Model (SoM) is the foundational AI visibility metric. It measures how often your brand is mentioned or recommended relative to your competitors across a defined set of AI-relevant prompts in a specific AI platform.

Definition

SoM is calculated as:

Share of Model = (Prompts where your brand appears / Total prompts tested) × 100

For competitive analysis, extend this to:

Relative SoM = (Your brand mentions / Total brand mentions across all competitors) × 100

How to calculate SoM

1. Build a prompt bank of 20-50 queries relevant to your brand's category, use cases, and competitive landscape. Include queries your target customers are likely to ask AI assistants.
2. Run all prompts through your target AI platforms (ChatGPT, Perplexity, Google AI Overview) at the same time of day to minimize variability.
3. Record which brands are mentioned in each response. Count your brand's appearances and your competitors' appearances.
4. Calculate your percentage share and track it over time (weekly or monthly baseline).

Benchmarks

SoM benchmarks vary significantly by category and competitive density. As a rough guide for 2026:

• Category leader: 60-80% SoM in high-salience prompts
• Strong competitor: 30-50% SoM in category-relevant prompts
• Market participant: 10-30% SoM in broad category prompts
• Low AI visibility: below 10% SoM, brand is not well-represented in AI systems

These are directional benchmarks, not universal standards. Your actual target SoM depends on your category, competitive set, and business goals. Use our prompt calculator to build your prompt bank and baseline.

Citation Rate: How to Track, What's a Good Rate

Citation Rate is a more granular metric than SoM. Where SoM tracks brand mentions generally, Citation Rate tracks how often your specific content (a specific URL, article, or data point) is cited as a source in AI-generated responses.

How to track Citation Rate

Citation Rate tracking requires prompt-by-prompt analysis:

1. For each prompt in your bank, note not just whether your brand is mentioned but whether a specific page is cited as a source (Perplexity and ChatGPT Search both surface source URLs in many responses)
2. Track which of your URLs are cited most frequently. These are your highest-authority pages for AI citation
3. Monitor changes in citation rates after you make optimization changes to specific pages (schema markup updates, content restructuring, new data added)

What's a good Citation Rate?

Citation Rate benchmarks depend heavily on content type and category. Pages with FAQPage schema, original research data, and strong entity signals consistently achieve higher citation rates. For well-optimized pages in competitive categories:

• High-performing pages: 20-40% citation rate across relevant prompts
• Moderate performance: 5-20% citation rate
• Needs optimization: below 5% for pages targeting high-intent AI queries

AI Referral Traffic: How to Set Up Tracking in GA4

While traditional SEO metrics do not capture AI visibility broadly, there is one overlap: when users follow source links in AI platforms (especially Perplexity and ChatGPT Search), those visits show up in your analytics as referral traffic. Setting up proper tracking for AI referral traffic lets you measure the direct traffic impact of your AI visibility.

Setting up AI referral tracking in GA4

Step 1: Identify AI referral sources

The main AI platforms that generate referral traffic and their referral domains:

• Perplexity: perplexity.ai
• ChatGPT: chat.openai.com
• You.com: you.com
• Google AI Overview: typically shows as direct traffic or google.com referral

Step 2: Create a channel group in GA4

In GA4, go to Admin → Data display → Channel groups → Create new channel group. Add a custom channel called "AI Referral" with the following conditions: Source contains "perplexity.ai" OR Source contains "chat.openai.com" OR Source contains "you.com".

Step 3: Create a custom report

Build a custom Exploration in GA4 that filters by your "AI Referral" channel group. Track: sessions, engaged sessions, conversions, and revenue by AI referral source. Compare monthly to monitor growth as your AI visibility improves.

Step 4: Add UTM parameters to trackable links

For any content you publish on platforms where you control the URL format, add UTM parameters that help distinguish AI-referred traffic more precisely. This is particularly useful for content published on LinkedIn, your newsletter, or press releases that AI systems might surface.

Prompt Coverage: Building a Prompt Bank and Testing Systematically

Prompt Coverage is a qualitative measurement of the breadth of your AI visibility: across how many different question types and user intents does your brand appear in AI responses?

Building your prompt bank

A well-structured prompt bank covers five categories of queries:

Category 1: Direct brand queries

"What is [your brand]?", "Tell me about [your brand]", "What does [your brand] do?" These baseline prompts test how AI systems represent your brand when explicitly asked. They reveal your entity clarity and description accuracy.

Category 2: Category queries

"What is the best [your category] tool/agency/product?", "Top [your category] providers in [your target market]", "Which [your category] companies should I consider?" These are the highest-value prompts. Appearing here means AI is recommending your brand as a category solution.

Category 3: Use case queries

"How do I [use case your brand solves]?", "What tools help with [use case]?" These prompts test whether AI systems associate your brand with specific problems and use cases.

Category 4: Comparison queries

"[Your brand] vs [competitor]", "Compare [your brand] and [competitor]", "How does [your brand] differ from [competitor]?" These reveal how AI represents you in competitive contexts.

Category 5: Topic authority queries

Questions about topics your brand has published expert content on. "How do I [topic you've written about]?", "What is [term from your glossary]?" These measure whether your content is recognized as authoritative on specific topics.

Testing systematically

Run your full prompt bank weekly (or bi-weekly for large banks). Keep a structured spreadsheet or database recording: the prompt, the platform, the date, whether your brand appeared, the position in the response, which competitors appeared, and any source URLs cited. Over time, this dataset reveals patterns and trends that inform optimization priorities.

Sentiment Tracking: Not Just Citations But How AI Describes Your Brand

An underappreciated aspect of AI visibility measurement is that appearing in AI responses is not uniformly good. The context and sentiment of how AI describes your brand matters enormously.

What to track in AI brand descriptions

• Category positioning: Is AI describing your brand in the right category? Incorrect categorization (e.g., AI calling a professional services firm a "software company") signals entity confusion that can mislead potential customers.
• Attribute accuracy: Are the specific claims AI makes about your brand accurate? Incorrect pricing, feature descriptions, or target market descriptions are common AI errors that need active correction.
• Tone and framing: Is AI describing your brand neutrally, positively, or with caveats? "Brand X is a leading provider" vs "Brand X, which has faced criticism for..." represent very different citation outcomes.
• Comparison positioning: In competitive comparisons, how does AI position your brand relative to competitors? This reveals AI's implicit ranking of your brand's authority and quality.

Tracking sentiment over time

Assign a qualitative score (1-5) to each AI response about your brand: 1 = inaccurate or negative framing, 3 = neutral and accurate, 5 = positive and authoritative. Average these scores weekly to create a Brand AI Sentiment Score that you track alongside your SoM.

Negative sentiment or inaccurate descriptions in AI responses are the signal that you have LLM perception drift, a problem that requires active content and entity optimization to correct.

Competitor Share of Model Comparison

AI visibility is inherently competitive. Understanding your SoM in absolute terms matters less than understanding it relative to your competition. Competitor SoM analysis reveals:

• Which competitors have stronger AI citation authority than you in your shared category
• Which prompt types your competitors appear in that you do not, revealing content and entity gaps
• How AI positions you relative to competitors in head-to-head comparison queries
• Which competitors are investing in GEO and gaining ground over time

Run your prompt bank against both your brand and your top 3-5 competitors simultaneously. The competitive SoM dashboard you build from this data is your most direct indicator of AI visibility performance.

Free vs Paid Tracking Tools Overview

Free approaches

• Manual prompt testing: Test your prompt bank manually in ChatGPT, Perplexity, and Google AI Overview. Free but time-intensive and not scalable for large prompt banks.
• GA4 AI referral tracking: Set up the custom channel group described above. Free within GA4, gives you direct traffic measurement.
• Akravo GEO Audit: Our free audit tool gives you a baseline assessment of your AI visibility signals including entity recognition and schema completeness.

Paid tools (2026 landscape)

• Dedicated AI visibility platforms: Tools like Profound, Trackr, and Goodie AI automate prompt-based SoM tracking at scale, including competitor tracking and historical trend data.
• Enterprise SEO platforms: Semrush, Ahrefs, and Moz have begun incorporating AI visibility metrics alongside traditional SEO data. Coverage and accuracy varies by platform.
• Brand monitoring tools: Platforms like Mention and Brand24 are beginning to index AI-generated content, though coverage remains partial compared to traditional web monitoring.

Building Your AI Visibility Dashboard

Combine the metrics above into a single weekly dashboard:

• SoM % by platform (ChatGPT, Perplexity, Google AI Overview)
• Relative SoM vs top 3 competitors
• Citation Rate for top 10 pages
• AI referral sessions and conversions (from GA4)
• Prompt Coverage score (# of prompt categories where you appear / total categories)
• Brand AI Sentiment Score

Reviewed weekly, this dashboard gives you a comprehensive picture of your AI visibility that traditional analytics simply cannot provide. It is the measurement layer that makes your GEO strategy accountable and improvable.