Share of Voice

What Share of Voice measures

Share of Voice counts how many times AI mentions you versus your competitors on category-defining queries. The probe fires 10 user-realistic prompts (per the locked v3 query taxonomy) across ChatGPT, Gemini, and Perplexity, parses each response for brand mentions, and calculates your mention share.

If three competitors get mentioned on every prompt and you only show up on three of ten, your SoV is roughly 8% (3 / (3 × 10 + 3)). The score normalises across vertical and engine so a 50% SoV in SaaS means something different than a 50% SoV in news.

The 7 intent buckets we probe

Each audit fires 10 prompts across these 7 buckets, weighted by your detected vertical. The buckets are based on real query distributions from probe research — not what tools think users ask, but what they actually do.

• compare_arrows Compare — "X vs Y" head-to-head queries (~25% of weight)
• recommend Recommend — "best X for Y use-case" prompts (~10% — downweighted from 50% based on real probe data showing only 2% vendor-cite rate)
• help Explain — "what is X / how does X work" educational queries
• build Troubleshoot — "X not working" / "fix X" practical queries
• extension Integrate — "X with Y" combination queries
• payments Price-research — "how much does X cost" / "X pricing"
• change_circle Alternatives — "alternatives to X" / "X competitors"

How to improve your Share of Voice

SoV is the AURA dim with the most direct ROI on content work. Push these levers:

• article Create category content. AI loves to cite "best X" and "X vs Y" listicles. If you write a fair, well-sourced comparison page that includes your competitors, AI will often cite that page when synthesising answers — and you get mentioned alongside the competitors.
• forum Get into Reddit / HN / forum threads. Discussion-thread sources punch above their weight in AI answers because they read like real-user opinions. Don't astroturf — build genuine community presence.
• menu_book Be in canonical category lists. G2, Capterra, ProductHunt, "awesome-X" GitHub repos. AI uses these as anchor sources for category questions.
• school Get cited in educational content. Tutorial sites, YouTube transcripts, dev.to, freeCodeCamp. AI loves training on educational content — brands mentioned there get free SoV lift.
• refresh Re-audit at 90 days. The drift re-audit (free with every paid audit) shows whether your work moved the needle. AURA shifts faster than AI Brand Recognition because live grounding picks up new content within weeks.

Frequently asked questions

• What queries does Share of Voice probe?expand_more
  Each audit fires 10 user-realistic prompts across the 7 intent buckets above, templated for your detected vertical (SaaS, e-commerce, news, local, blog, generic). Total: 30 LLM calls per audit (10 prompts × 3 engines). The exact prompts are visible in your audit report so you can see what AI was asked.
• How is SoV different from Competitive Rank?expand_more
  SoV counts mentions: how often you appear, period. Competitive Rank measures ordering: when AI lists vendors, where do you rank? You can have high SoV but low Rank (mentioned a lot but always last) or low SoV but high Rank (rarely mentioned but always #1 when you are).
• Why does my SoV fluctuate between audits?expand_more
  AURA probes have a noise floor of ±5–10 points between runs because LLMs have small variation in their outputs. A genuine SoV change is typically 15+ points. The 90-day drift re-audit (free with every paid audit) helps separate signal from noise.
• My competitors are everywhere in AI — how do I catch up?expand_more
  Don't try to win every prompt. Pick your strongest 1–2 intent buckets (usually Compare and Alternatives for B2B, Explain and Troubleshoot for content sites) and focus content there. Owning 70% SoV in 2 buckets beats 15% SoV in all 7.