Share of voice in AI answers: the metric that replaces rankings

In a search results page you could be #4 and still exist. In an AI answer there are no positions — the assistant names two or three brands and moves on. The scoreboard for that world is share of voice: of all brand mentions across the questions that matter in your category, what percentage are yours?

The math (it’s short)

Track a set of prompts across the engines you care about. Count brand mentions in the answers — yours and your competitors’. Your share of voice = your mentions ÷ all tracked-brand mentions. If ChatGPT-and-friends name someone 100 times this month across your category’s questions and 18 of those are you, you hold 18% share of voice.

Why it beats tracking mentions alone

Raw mention counts flatter you when the category is booming and hide losses when a rival surges. Share of voice is zero-sum by construction — if yours went from 18% to 24%, someone else’s fell, and that’s a sentence a board understands. It’s also the earliest reliable signal of a competitor’s content push working: their share climbs before your traffic dips.

Reading it honestly

• Per engine. Engines disagree; 30% on Perplexity and 5% on Gemini is one story averaged into mush. Look at the breakdown.
• Per segment. Share of voice on “best budget X” and “best enterprise X” are different markets. Tag your prompts and read them separately.
• Alongside the sources. A rival’s share spike almost always traces to specific pages engines started citing. Those pages are your to-do list.

Moving the number

Win the citation gaps (pages that mention competitors but not you), make your own pages quotable enough to be lifted into answers, and fix anything that keeps AI crawlers from reading you. Then watch the weekly series — Asheard computes share of voice per engine, per prompt segment, automatically, alongside your Google rankings, so both halves of visibility live on one chart.