TL;DR
Key Definition
AI share of voice is calculated by dividing the number of responses that cite your brand by the total responses generated across your prompt set. According to AthenaHQ State of AI Search 2026 report, a brand appears in only 17.2 percent of AI answers on average, while the best-optimized brands reach up to 56.7 percent.
Why Does It Matter?
- It turns your AI visibility into a metric comparable against your direct competitors
- It reveals the prompts where competitors dominate and where you are absent
- The leader in classic search often holds 25 to 40 percent of AI share of voice in its category
- It acts as a directional indicator for your GEO and content efforts
- Tracked over time, it shows whether your optimization work is paying off
- It helps prioritize the topics where a visibility gain will have the most commercial impact
How Does It Work?
AI share of voice is built from a representative set of prompts, run across several engines, where each response is analyzed to count brand mentions.
The Basic Formula
AI share of voice equals the number of responses mentioning your brand divided by the total responses across your prompt set, times one hundred. If your brand appears in 300 of 1500 responses, your share of voice is 20 percent.
Mentions and Citations
Share of voice can be measured on mentions, which reflect your presence in the conversation, or on citations, which reflect your share of authoritative sources. The two angles tell a different and complementary story.
Position Weighting
A mention cited first weighs more than one buried at the bottom of a list. A simple model assigns 3 points to first position, 2 to second, and 1 to third, producing a weighted share of voice closer to the real experience.
How Do You Measure It?
- Define a set of prompts representative of how your customers question AI
- Select the competitor panel that serves as the denominator of your calculation
- Run the same prompts across several engines such as ChatGPT, Gemini, Perplexity, and Google AI Mode
- Repeat each prompt several times and average, since responses vary from one run to another
- Track share of voice by prompt category and by platform, not just one global figure
- Measure on a regular cadence to tell a real trend from a one-off fluctuation
What Are the Common Mistakes?
- Measuring on too narrow a prompt set, which produces a misleading share of voice
- Tracking only one platform when your customers use several
- Taking a single snapshot without building a regular tracking cadence
- Celebrating a rise in mentions without checking their sentiment or position
- Computing position weighting on a single run, which makes it unstable
- Choosing an unrepresentative competitor panel that skews the denominator
What Should You Do First?
Follow these steps to get started.
- List the prompts that truly matter in your customers buying journey
- Set the competitor panel you want to benchmark against
- Cover several AI engines for a complete view of your category
- Repeat and average each prompt to neutralize response variability
- Segment share of voice by topic to spot your blind spots
- Cross share of voice with sentiment so you do not confuse visibility with reputation
- Schedule periodic tracking and compare each reading to the previous one
Examples
Benchmark Against a Competitor Panel
Across 1000 responses generated for your category prompts, your brand appears 250 times and your main competitor 400 times. Your AI share of voice is 25 percent versus 40 percent for them.
Gap by Topic
You dominate definition prompts with 60 percent share of voice but drop to 10 percent on comparison prompts, revealing a precise topic to work on.