Share of Voice
Share of Voice (SoV) in generative engine optimization measures the percentage of AI-generated responses that cite your brand for a given topic. Unlike traditional SEO rankings, which are deterministic (you are either position 3 or you are not), AI citation is probabilistic: the same query asked ten times may cite your brand seven times, three times, or zero times. Measuring SoV requires statistically significant sample sizes and confidence intervals, not single-query spot checks.
How to Measure Share of Voice
SoV measurement requires running the same query multiple times across multiple AI platforms and calculating the percentage of responses that include your brand. A single query is statistically meaningless because LLM outputs are non-deterministic. Reliable measurement uses convergence-based sampling: collecting observations until the citation rate stabilizes within a defined confidence interval, typically 95%.
The formula is straightforward: (number of responses citing your brand / total responses sampled) x 100 = SoV percentage. A brand cited in 34 out of 100 sampled responses has a 34% Share of Voice for that query. The confidence interval tells you whether that 34% is stable or noise.
Why SoV Replaces Rankings
Traditional SEO tracks a single position on a ranked list. GEO tracks a probability distribution across AI responses. This shift matters because:
- Multiple brands can be cited in a single response. Unlike search rankings where position 1 gets the majority of clicks, AI responses routinely cite 3 to 7 sources. SoV measures your share of that citation pool.
- Citation rates fluctuate. Model updates, new training data, and competing content cause citation rates to shift over time. SoV tracked weekly reveals trends that single-query checks miss entirely.
- Platform differences matter. Your SoV on ChatGPT may differ significantly from your SoV on Perplexity or Google AI Overviews. Cross-platform measurement reveals where your brand is strong and where competitors are winning.
SoV Benchmarks
There are no universal SoV benchmarks because citation rates vary dramatically by industry, query type, and platform. However, a brand consistently achieving above 30% SoV for its core topic on a given platform is performing well. Below 10% indicates the brand is being outcompeted or not surfaced at all. The most actionable use of SoV is tracking your own trajectory over time and comparing against specific competitors, not chasing an absolute number.
For the complete measurement framework, see the Generative Engine Optimization guide.
Related: Convergence-Based Sampling · Citation Frequency · Confidence Interval · Competitive Dynamics


