How to Track AI Share of Voice: A Step-by-Step Measurement Guide

How to Track AI Share of Voice

AI Share of Voice is the percentage of relevant AI-generated answers in which your brand is mentioned, recommended, or cited. It is to AI search what keyword ranking position is to traditional SEO: the primary performance metric that tells you whether your AI visibility program is working.

Most brands have no systematic AI Share of Voice measurement in place. They do not know whether AI systems are recommending them, avoiding them, or actively warning buyers away. This is a significant blind spot given how much buyer research now runs through AI.

This guide explains exactly how to set up AI Share of Voice tracking from scratch.

Step 1: Define Your Query Set

The first step is identifying the queries your target buyers are actually asking AI systems. These fall into three categories:

Category queries
Queries buyers use when they are in the consideration phase, researching what type of solution they need. Examples: “what is the best [product category] for [use case]?” or “which companies specialize in [service category]?”

Comparison queries
Queries buyers use when comparing specific options. Examples: “how does [your brand] compare to [competitor]?” or “[your brand] vs [competitor B].”

Validation queries
Queries buyers use when they want to verify a brand before engaging. Examples: “is [your brand] legit?” or “what do people say about [your brand]?” or “reviews of [your brand].”

For initial setup, build a query set of 15 to 25 prompts covering all three categories. This is enough to establish a baseline and detect meaningful trends.

Step 2: Select Your AI Platforms

AI Share of Voice needs to be tracked separately across each major AI platform because the same query can produce different answers on different platforms. The platforms to track:

ChatGPT (OpenAI)
The highest-volume AI search platform for most B2B categories. Relevant for both GPT-4o (with browsing) and the base model responses.

Perplexity AI
A real-time retrieval AI system that actively searches the web before answering. Perplexity is highly relevant for brands targeting buyers who research before purchasing. Its citations make it particularly useful for identifying which sources influence your AI visibility.

Google Gemini / AI Overviews
Google’s AI systems are visible within standard Google search results (AI Overviews) and through the Gemini product. Given Google’s search volume dominance, tracking Google AI Overview appearances for your target queries is high priority.

Claude (Anthropic)
Increasingly used for research and buying decision support, particularly in tech and professional service categories.

For initial setup, prioritize ChatGPT and Perplexity. Add Gemini and Claude once the initial tracking system is running.

Step 3: Run Your Query Set

For each query in your set, run it across each AI platform and record the full response. For initial tracking, do this manually. You are capturing:

Whether your brand is mentioned: Simply yes or no. Was your brand named anywhere in the response?

Where your brand appears: Is your brand the primary recommendation, one of several options, or a passing mention? Being named first is more valuable than being named fourth.

The language used about your brand: What descriptors, qualifiers, and associations does the AI use when mentioning your brand? “A good option for small businesses” is different from “an enterprise-grade solution” is different from “some users report billing issues.”

Whether competitors are mentioned instead: Which competitors appear in the responses where you are not mentioned? This is your competitive displacement data.

Sources cited (for Perplexity): For Perplexity responses, record the sources cited. These reveal which third-party sources are most influencing your AI visibility for each query.

Step 4: Build Your Tracking Sheet

Create a tracking spreadsheet with the following structure:

Rows: One row per query per platform per tracking cycle.

Run this full query set monthly at minimum. Weekly tracking is useful for brands actively running AI visibility programs and wanting to see the impact of new content or citation-building work.

Step 5: Calculate Your AI Share of Voice Metrics

From your tracking data, calculate:

Overall mention rate
(Number of queries where brand was mentioned / Total queries run) x 100

Example: Brand mentioned in 8 of 20 queries = 40% overall mention rate.

Platform-specific mention rates
Run the same calculation per platform to identify where you are strongest and where gaps exist.

Positive mention rate
(Number of queries with positive brand language / Total queries where brand was mentioned) x 100

Share of mentions vs competitors
For queries where any brand in your category is mentioned, what percentage name you vs. each competitor?

Category query vs validation query performance
Track mention rates separately for category queries (discovery phase) and validation queries (verification phase). Different results in these categories point to different strategic fixes.

Step 6: Diagnose the Gaps

Low AI Share of Voice results from identifiable, fixable causes. Common patterns:

Present in category queries, absent in comparison queries
You are getting discovery mentions but not appearing in head-to-head comparisons. Fix: build comparison content pages for each major competitor matchup. Get featured in “vs.” content from third-party sources.

Absent in category queries entirely
AI systems are not associating your brand with the category at all. Fix: entity consistency audit, category-specific Reddit presence building, editorial coverage in category publications.

Mentioned with consistent negative qualifiers
AI systems are accurately reflecting negative sentiment from community sources. Fix: source identification (where is the negative content coming from?), reputation repair in the specific communities generating those signals.

Strong on one platform, weak on others
Different AI platforms draw from different sources. Weakness on Perplexity often means weak real-time web presence (Reddit, review platforms, editorial). Weakness in ChatGPT base responses often means training data gaps.

Tools to Scale AI Share of Voice Tracking

Manual tracking works for initial baseline and monthly monitoring. At larger scale, tools are available to automate the process:

Semrush AI Toolkit: Integrates AI mention tracking with traditional SEO data.

Profound: Purpose-built AI brand mention tracking across multiple AI platforms.

Otterly.ai: AI Share of Voice tracking with automated prompt testing.

For most brands starting AI visibility programs, manual tracking for the first 90 days is sufficient to establish baseline and directional trends. Platform tools become valuable once the brand needs to track 50+ queries across multiple platforms.

Start With a Baseline AI Share of Voice Measurement in Your Free Digital Moat Visibility Audit

The audit includes a baseline AI Share of Voice snapshot covering your top category queries across ChatGPT and Perplexity, so you start with data rather than guesses.

Frequently Asked Questions

How many queries should I track for meaningful AI Share of Voice data?
For initial tracking, 15 to 25 queries is sufficient to establish a baseline. The queries should span category discovery, comparison, and validation types. Adding more queries increases precision but also increases the manual tracking workload. Start with 20 and expand once the tracking system is running smoothly.

How often should I run AI Share of Voice tracking?
Monthly tracking is the minimum for most brands. Brands actively running AI visibility programs should track bi-weekly or weekly to measure the impact of specific initiatives (new content published, new Reddit presence, new editorial citations). Quarterly tracking is too infrequent to detect trends or measure program impact.

Does AI Share of Voice vary by geographic region?
Yes, particularly for brands with strong or weak regional presence. AI systems draw from geographically distributed sources and may produce different recommendations for the same query depending on the user’s location context. For brands with significant regional variation in business, track queries with explicit geographic context (“best [service] in [city]”) in addition to general queries.

What is a good AI Share of Voice benchmark?
Benchmarks vary significantly by category and competitive intensity. For well-established brands in defined categories, 50-70% overall mention rate across category queries is achievable. For newer brands or crowded categories, 20-30% is a reasonable initial target. The more useful benchmark is your share relative to direct competitors: being mentioned more often than your primary competitor matters more than hitting an absolute percentage.

Reviewed by Hank Cai, Founder of Digile Media. AI Share of Voice tracking is the measurement foundation of every Digital Moat System engagement.

How to Track AI Share of Voice: A Step-by-Step Measurement Guide