Share of Voice, Defined Precisely

Everyone in GEO quotes a Share of Voice number. Here's our exact formula, a worked example, and why per-engine splits matter more than the headline.

5 min readAnovox team

The formula

For one workspace, one engine, one day: Share of Voice = (answers naming your brand ÷ total answers collected) × 100, rounded to one decimal. Run 10 prompts across 4 engines and you have 40 answers; if 17 name you, that's 42.5% for the day. Workspace SOV aggregates prompt-level results upward, and trend lines compare the same prompts week over week — so a rising score means genuinely more citations, never a changing prompt mix.

A worked example

Take an illustrative workspace tracking 10 buyer prompts on ChatGPT and Gemini — 20 answers a day. Monday: ChatGPT names the brand in 6 of 10 (60%), Gemini in 3 of 10 (30%). Overall: 9 of 20, 45%. By Friday, after publishing one FAQ targeting the two weakest prompts, it's 7 and 5 — 12 of 20, 60%. The +15pt move decomposes cleanly: which prompts flipped, on which engine, after which fix. That decomposition is the whole point of measuring daily instead of monthly.

Why per-engine splits beat the headline

A single blended number hides the actionable shape. 45% overall could be 90% on ChatGPT and 0% on Claude — and the fix for 'invisible on Claude' (different retrieval, different sources trusted) has nothing in common with 'slipping on ChatGPT'. Our dashboard reports per-engine SOV first and the aggregate second, and every gap in Aim names the engines where you're missing. Aggregate for the executive summary; split for the work.

What SOV doesn't capture (honestly)

Being named isn't being recommended — that's why every citation also gets a sentiment classification, and why a negatively-framed mention triggers our crisis watch instead of counting as a win. SOV also says nothing about position (first sentence vs buried paragraph six), which is why the citation board records first-mention position separately. One number opens the conversation; it never closes it. The full model, including what we deliberately don't score, is published openly in our scoring methodology.

See this on your own brand

Run the free check — same scoring the dashboard uses, no account needed.

Run free check

Related posts

How-to

Your robots.txt Is Your GEO Front Door

4 min read
Announcement

We Pointed Anovox at Itself: The 30-Day Loop

4 min read