gaash.ai

Share of Voice

Share of Voice (SoV) is the percentage of all brand mentions on a topic that belong to you, rather than a competitor. In a GEO context, you run a set of real user questions through AI assistants like ChatGPT, Gemini or Claude, tally every brand each one names, and divide your mentions by the total. A high SoV means that when someone asks an AI assistant about your category, you're the name that comes up.

Why Share of Voice matters

Visibility isn't a yes-or-no state, it's a slice of a fixed pie. Knowing your brand shows up in AI answers tells you little on its own — what matters is how often you show up relative to everyone else fighting for the same question. Share of Voice gives you that comparison, which is also what makes it trackable over time: if your share climbs, it climbed at a competitor's expense, because the total mentions in an answer are limited. That scarcity is the real reason SoV matters for GEO. ChatGPT alone counts roughly 900 million weekly users and Gemini has passed a billion monthly users, so a bigger slice of a fixed set of AI-generated recommendations reaches a genuinely large audience.

How it works

Start with a list of the real questions your buyers ask — "which accounting software suits a small business" is a usable prompt, "accounting software" alone is not. Run each prompt many times across the AI assistants you care about, since answers vary from one run to the next, and log every brand named, not just yours. Your Share of Voice is your mentions divided by total mentions across all brands in that set. Treat a first-position recommendation as worth more than a name buried in a list, because position affects whether a user acts on it. Whatever weighting you choose, apply it consistently so runs stay comparable to each other.

Common mistakes

The most common mistake is calculating SoV from a single prompt run. AI outputs aren't deterministic — the same question can surface different brands on different attempts — so one run measures noise, not your position. A second mistake is tracking only your own mentions and skipping the competitive count; that gives you a raw mention rate, not a share. A third is defining the question set too loosely, which dilutes the result with irrelevant traffic. Chasing this through llms.txt or AI-specific markup is also a dead end: Google has said explicitly that AI Overviews and AI Mode need no special files, and an Ahrefs analysis of roughly 137,000 sites that published an llms.txt found about 97% saw no measurable referral traffic from it. Track SoV per assistant, because each pulls from different sources, and only compare across them once you have separate numbers for each.

Relation to AI recommendations

Share of Voice is the bridge between abstract AI visibility and an actual purchase decision. Ask an assistant for a recommendation and you typically get a handful of names, not a page of search results, so missing that shortlist means missing the decision entirely. A rising SoV means you're landing in that shortlist more often than your competitors. It also points to where the lever actually is: Ahrefs' analysis of roughly 75,000 brands found web-mention frequency correlates with AI citation rate at about 0.664, close to three times the correlation for backlinks (about 0.218) — so earning third-party mentions moves SoV more reliably than link-building does.

Example

Take a project management software vendor. Its team runs ten realistic questions, such as "which tool suits an agile team," fifty times each across ChatGPT, Claude and Perplexity, and logs every brand mentioned in the answers — 900 mentions in total. Their own product is named 180 times, the strongest competitor 270 times, and the rest are split among smaller providers. That's a Share of Voice of 20 percent against the market leader's 30. After three months spent earning third-party coverage and tightening comparison content, their share climbs to 28 percent, closing most of the gap.

Common questions

What's the difference between Share of Voice and mention rate?

Mention rate is a raw count — how many times your brand gets named, full stop. Share of Voice divides that count by the total mentions across every brand in the same question set, so it tells you your position relative to competitors, not just your own volume.

How many queries do I need for a reliable Share of Voice number?

More than one. Because AI answers vary between runs, a single query tells you almost nothing reliable. Ask each question several dozen times per assistant so the variation averages out and the share you measure reflects an actual pattern rather than a lucky or unlucky run.

Related terms