projects / ai-share-of-voice
AI Share-of-Voice Analyzer
A tool that measures how often a brand shows up when buyers ask AI models for recommendations.
2025–present · AI · marketing analytics · CLI · github ↗
## problem
B2B buyers increasingly ask ChatGPT-style tools "what should I use for X?" before they ever touch a search engine. Brands have decades of SEO tooling and essentially no way to measure their presence in AI answers.
If you can't measure it, you can't improve it, and most companies don't even know whether AI models mention them at all.
## approach
The analyzer runs a configurable set of prompts, tagged by buyer persona and funnel intent, against any OpenAI-compatible model endpoint, then parses the responses for brand mentions.
It computes the metrics a marketer actually wants: presence rate, share of voice against competitors, and recommendation rate, plus automated insights.
It deliberately avoids vendor lock-in: it works with free and local options like Ollama, Groq, and Gemini Flash, so measuring costs nothing.
## impact
Turns "do AI models recommend us?" from a shrug into a number you can track week over week.
Open source and extensible, so teams can add their own model providers and prompt sets.
## key learnings
AI answers are becoming a results page, and almost nobody is measuring their ranking on it yet.
Measurement has to come before optimization. Most "AI visibility" advice skips straight to tactics with no baseline.
Building the tool taught me more about how models pick recommendations than any blog post about it had.